+ All Categories
Home > Documents > ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary....

ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary....

Date post: 05-Jul-2020
Category:
Upload: others
View: 0 times
Download: 0 times
Share this document with a friend
59
NASA Contractor Report 158947 ON SEVERAL ASPECTS AND APPLICATIONS OF THE MULTIGRID METHOD FOR SOLVING PARTIAL DIFFERENTIAL EQUATIONS NATHAN DINAR COLLEGE OF WILLIAM AND MARY W.ILLIAMSBURG, VA 23185 NASA CONTRACT NAS1-14972-3 (NAS'A-CR-158947) ON SEVERAL ASPECTS AND APPLICATIONS OF THE MUTIGRID METHOD FOR 1118-32783 SOVTNG PARTIAL DIFFERENTIAL EQUATIONS (College of William and Mary) 58 p HC A04/NF AO1 CSC SEPTEMBER -1978 12A G3/64 -- Unclas 31570 IASA National Aeronautics and Space Administration Langley Research Center Hampton, Virginia 23665 https://ntrs.nasa.gov/search.jsp?R=19780024840 2020-07-28T01:18:19+00:00Z
Transcript
Page 1: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

NASA Contractor Report 158947

ON SEVERAL ASPECTS AND APPLICATIONS OF THE MULTIGRID METHOD FOR SOLVING PARTIAL DIFFERENTIAL EQUATIONS

NATHAN DINAR

COLLEGE OF WILLIAM AND MARY WILLIAMSBURG VA 23185

NASA CONTRACT NAS1-14972-3 (NASA-CR-158947) ON SEVERAL ASPECTS AND APPLICATIONS OF THE MUTIGRID METHOD FOR

1118-32783

SOVTNG PARTIAL DIFFERENTIAL EQUATIONS (College of William and Mary) 58 p HC A04NF AO1 CSC

SEPTEMBER -1978 12A G364

--

Unclas 31570

IASA National Aeronautics and Space Administration

Langley Research Center Hampton Virginia 23665

httpsntrsnasagovsearchjspR=19780024840 2020-07-28T011819+0000Z

TABLE OF CONTENTS

Page

I Introduction 1

II Multigrid Methods for Systemsof Equations 4

21 Cauchy-Riemann Equations 522 Stokes Equations 923 Navier-Stokes 12

III Control and Prediction Techniquesin Multigrid Methods 16

IV Multigrid Methods for Time DependentParabolic Equations 21

V Improvements and Changes in ExistingMultigrid Algorithms 27

51 The Poisson Equation 2752 Distribution Relaxation 33

VI References 39

VII Appendix 40

ON SEVERAL ASPECTS AND APPLICATIONSOF THE MULTIGRID METHOD FOR SOLVING PARTIAL

DIFFERENTIAL EQUATIONS

By Nathan Dinar

College of William amp MaryWilliamsburg Virginia 23185

Research Sponsored by NASA Langley Research CenterContract NASl-14972-3

SUMMARY

Several aspects of multigrid methods are briefly described in this report The main subjects include the development ofvery efficient multigrid algorithms for systems of ellipticequations (Cauchy-Riemann Stokes Navier-Stokes) as well asthe development of control and predictiontools (based on localmode Fourier analysis) used to analyze check and improve thesealgorithms

Preliminary research on multigrid algorithms for timedependent parabolic equations is also described This reportdeals also with improvements in existing multigrid processesand algorithms for elliptic equations

Some partial and typical results are given More completeand detailed information will be presented in the authorsPhD Thesis to appear at the Weizmann Institute of ScienceRehovot Israel

I INTRODUCTION

This report deals with several aspects concerning multigrid

methods for fast solution of partial differential equations It

covers the research on this subject for the period August 1977 -

August 1978 when the author was spending his sabbatical at the

NASA Langley Research Center This research is part of a PhD

Research Assistant Department of Mathematics

2

Thesis to appear shortly at the Weizmann Institute of Science

Rehovot Israel including more detailed results and conclusions

The research on multigrid methods began in the early 1970s

by the initiative of Professor A Brandt He is today supervising

several projects on these subjects including the present research

The multigrid method can be applied to a wide range of

problems and therefore it interests many people It is known

to be one of the most powerful and advanced methods used today

The multigrid method uses the fact that the numerical

discrete equations we usually want to solve are not independent

They are derived from a continuous problem whose solution we

want to approximate In the process of the solution of the

discrete equations it is convenient to keep in mind the differshy

ential origin of the problem The use of discrete operators

on several levels of meshes interacting strongly in the process

of the solution allows us to solve the problems on the finest

grid very efficiently with a minimum number of arithmetical

operations (0(N)) where N is the number of equations in the

the finest grid

The multigrid method consists generally of several

processes performed in a given order and defining an itershy

ative cycle These processes include generally

Relaxation (Usually of Gauss-Seidel type) Used only to

smooth the errors ie to reduce these high

frequencies in the error that are not described

in coarser grids

3

Transfer of Residuals to a Coarse Grid This allows us to

define a problem on a coarse grid that is

similar to the original one The solution on

the coarse grid is used to improve the approxishy

mate solution given on the finer grid

Interpolation Used to define a new approximation on a

finer grid given an approximation solution on

a coarse grid

Moreover we can intermix the principles of the multishy

grid method with the grid adaptivity principles which mean

adaption of the order of discrete approximations and mesh

size using total smoothness of the solution This keeps

N as small as possible

These basic ideas and others as well as treatment of

theoretical and practical aspects have been extensively

covered in the papers of Brandt I2 3

This report deals with a-wide spectrum of problems

involved in the development of multigrid methods and multishy

grid software and the principal parts are the following

a) Development of multigrid methods and software

for elliptic systems of equations (Including

Cauchy-Riemann equation Stokes equation and

Navier-Stokes equation

b) Development of control and prediction tools for

various steps and processes involved in the

4

multigrid method (Based on local mode Fourier

analysis)

c) Development of multigrid methods for time

dependent parabolic equations (Heat equation

on rectangular domains)

d) Improvement and changes in existing multigrid

processes and algorithms (mainly on Poisson

equation) and treatment of new ideas in relaxation

methods

II MULTIGRID METHODS FOR SYSTEMS OF EQUATIONS

The basic ideas of the multigrid method are not

restricted of course to a unique equation and from a

theoretical point of view no special problem could be

expected in implementing multigrid ideas to a system of

equations However special and detailed algorithms for

this problem did not exist and it was extremely important

to get sharp and practical proofs of the efficiency of

multigrid methods for systems of equations One of the

questions we did not know the answer in advance was for

instance what is the appropriate method of relaxation for

a system

In order to answer this and many other important

questions we developed multigrid algorithms for three

5

typical problems starting with a simple one and each

subsequent problem involving more and new complications

relative to the former Moreover since it is usually

wise and necessary in this kind of research to isolate

different questions which may arise we developed the

algorithm for simple geometrical domains (rectangles)

The past experience showed that more complicated geometries

did not affect the efficiency of multigrid methods (see

for instance Shiftan4) For complicated geometries there

are of course more programing problems

21 Cauchy-Riemann Equations

The equations are

Sx-Vy fl (xy) (i)

Ux+Vy = f3(xY) (2)

for 05xsl Osyll

And the boundary conditions

U(Oy) = U0 (y) V(xO) = V0 (x) (3)

U(ly) = Ul(y) V(xl) = Vl(x)

Figure 1

y V (x)

U (y) iU(y)

v0 )x)

6

The data must fulfill an additional condition due to

the continuity equation (2) in order to assure-a solution

to the system This condition is

0V fIvj- 0 )dx f3 y)dxdy(I c 0

The discretization of the problem was accomplished

on a staggered grid This method has several advantages

in problems of this type For instance it allows us to

define easily second order conservative finite differences

The grid is described in figure 2a

Figure 2a

VVkV

U U2 f3 J3 U

U U V U

V V V

The finite differences approximations for equation-()

are defined at grid intersections (X) for instance

71 P jJ3t 1-(v2 V1 J

And the finite differences for equation (2) are defined in

the middle of the cells () ie

1u 3 -U 2 +V3 vk = f3 (6)

As in the continuous case the data must fill an additional

discrete condition equivalent to (4) in the finest grid ie

7

(Ulj1 -U 0 j)+ L (Vii-Voi) = J f3ij (4a)

Several relaxation schemes appropriate to this system

were considered and all these schemes have a good smoothing

factor around 7=5 The most convenient scheme was chosen

because of its simplicity and its small number of arithmetical

operations This method belongs to a new approach of relaxshy

ations developed by Professor A Brandt and is called

Distributed Relaxations

This approach is characterized by the fact that when

passing through a point (or cell) in the grid for which

the difference equations are defined we change the value

of more than one unknown in order to make zero residual

on this point

In the Cauchy-Riemann equations separate relaxations

are performed to each one of the equations according to

the schemes discribed in figures 3 and 4

Efficiency of relaxation in a multigrid process is measuredby the smoothing factor which is defined as the asymptoticratio between the fast components of the errors after a relaxshyation sweep and the same errors before the relaxation A more

exact definition can be found in Brandt 2 More considerationson this subject can be found in Chapter 3 of this report

8

Figure 3 Relaxation of Ux-V = f1

u2+

C

Figure 4 Relaxation of Ux+Vy

2

4f3 U-1i-+

The values U+j0 and V+ C in figures 3 and 4 represents the

new values after the sweep of this point 0 is always chosen

to make the residual zero at the point The relaxation is a

Gauss-Seidel type relaxation and it passes through all the points

(or cells) of the grid in a usual natural order The important

property of this method relies on the fact that when relaxing

each of the equations the residual of the other equation

remains unchanged

Mode Fourier Analysis shows that the smoothing factor for

this method is i =5 the same as in Poisson equation which

is of course equivalent to the Cauchy-Riemann system The

multigrid algorithm shows convergence factor 7 56 which is about the same value as in the Poisson case The

number of operations in the relaxation of the Cauchy-Riemann

See Chapter 3 in this report

9

system is of course higher than in Poisson equations (about

two times more) On the other hand we can say that the

information we get from the solution of the Cauchy-Riemann

system (functions U and V) is also two times the information

we get from the solution of the Poisson equation

The main reason for our treatment of Cauchy-Riemann

system was for learning purposes It is probably the simplest

elliptic system possible and it allows us to treat complishy

cated problems more easily However this system is intershy

esting in itself as we can see in the works of Lomax5 and

Ghil6 These papers describe fast algorithms for the

solution of the Cauchy-Riemann system We think that the

algorithm just described in this report is preferable

because it does not depend in principle on the simplicity

of the domain and because it is at least as fast as Chil

algorithm and even faster asymptotically (for finer and

finer discretizations) The implementation of the algorithms

for nonlinear equation is also much easier in the multigrid

3method

An example of a computer output is given in the Appendix

Output No 1

22 Stokes Equations

The Stokes equation are given by

AU-px = f (7)

t1 V-py = f2 (8)

U +V = f3 (9)

10

The unknowns are U V and P U and V are given on

boundaries Like in Cauchy-Riemann systems The data must

satisfy the condition (4) The discretization is also in the

same staggered grids already described P and f3 are defined

in the centers of the cells f1 and f2 are defined in the

same points as U and V respectively (see figure 2b)

Figure 2b

V-V -v v

U P U P U P U PII

U P U P U P U P

I -V-u- I --V- 0 -- -V I- VVI_

The difference equations for equation (7) are given by

AucxtxlampiP (x+~y)-P(x- S Y)] f(xy) (10)

for (xy) where U is defined A is the usual five

points discrete Laplace operator Li -

Similarly for equation (8)

I CxY+ )-P(xy- ff2 (xy) (1i)(xxyY)- )

for (xy) where V is defined

For the continuity equation (9) the difference

equations were already defined by (6)

Equations (10) and (11) hold for interior points Near

the boundary the difference equations are different but they

are simple and will not be described here

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 2: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

TABLE OF CONTENTS

Page

I Introduction 1

II Multigrid Methods for Systemsof Equations 4

21 Cauchy-Riemann Equations 522 Stokes Equations 923 Navier-Stokes 12

III Control and Prediction Techniquesin Multigrid Methods 16

IV Multigrid Methods for Time DependentParabolic Equations 21

V Improvements and Changes in ExistingMultigrid Algorithms 27

51 The Poisson Equation 2752 Distribution Relaxation 33

VI References 39

VII Appendix 40

ON SEVERAL ASPECTS AND APPLICATIONSOF THE MULTIGRID METHOD FOR SOLVING PARTIAL

DIFFERENTIAL EQUATIONS

By Nathan Dinar

College of William amp MaryWilliamsburg Virginia 23185

Research Sponsored by NASA Langley Research CenterContract NASl-14972-3

SUMMARY

Several aspects of multigrid methods are briefly described in this report The main subjects include the development ofvery efficient multigrid algorithms for systems of ellipticequations (Cauchy-Riemann Stokes Navier-Stokes) as well asthe development of control and predictiontools (based on localmode Fourier analysis) used to analyze check and improve thesealgorithms

Preliminary research on multigrid algorithms for timedependent parabolic equations is also described This reportdeals also with improvements in existing multigrid processesand algorithms for elliptic equations

Some partial and typical results are given More completeand detailed information will be presented in the authorsPhD Thesis to appear at the Weizmann Institute of ScienceRehovot Israel

I INTRODUCTION

This report deals with several aspects concerning multigrid

methods for fast solution of partial differential equations It

covers the research on this subject for the period August 1977 -

August 1978 when the author was spending his sabbatical at the

NASA Langley Research Center This research is part of a PhD

Research Assistant Department of Mathematics

2

Thesis to appear shortly at the Weizmann Institute of Science

Rehovot Israel including more detailed results and conclusions

The research on multigrid methods began in the early 1970s

by the initiative of Professor A Brandt He is today supervising

several projects on these subjects including the present research

The multigrid method can be applied to a wide range of

problems and therefore it interests many people It is known

to be one of the most powerful and advanced methods used today

The multigrid method uses the fact that the numerical

discrete equations we usually want to solve are not independent

They are derived from a continuous problem whose solution we

want to approximate In the process of the solution of the

discrete equations it is convenient to keep in mind the differshy

ential origin of the problem The use of discrete operators

on several levels of meshes interacting strongly in the process

of the solution allows us to solve the problems on the finest

grid very efficiently with a minimum number of arithmetical

operations (0(N)) where N is the number of equations in the

the finest grid

The multigrid method consists generally of several

processes performed in a given order and defining an itershy

ative cycle These processes include generally

Relaxation (Usually of Gauss-Seidel type) Used only to

smooth the errors ie to reduce these high

frequencies in the error that are not described

in coarser grids

3

Transfer of Residuals to a Coarse Grid This allows us to

define a problem on a coarse grid that is

similar to the original one The solution on

the coarse grid is used to improve the approxishy

mate solution given on the finer grid

Interpolation Used to define a new approximation on a

finer grid given an approximation solution on

a coarse grid

Moreover we can intermix the principles of the multishy

grid method with the grid adaptivity principles which mean

adaption of the order of discrete approximations and mesh

size using total smoothness of the solution This keeps

N as small as possible

These basic ideas and others as well as treatment of

theoretical and practical aspects have been extensively

covered in the papers of Brandt I2 3

This report deals with a-wide spectrum of problems

involved in the development of multigrid methods and multishy

grid software and the principal parts are the following

a) Development of multigrid methods and software

for elliptic systems of equations (Including

Cauchy-Riemann equation Stokes equation and

Navier-Stokes equation

b) Development of control and prediction tools for

various steps and processes involved in the

4

multigrid method (Based on local mode Fourier

analysis)

c) Development of multigrid methods for time

dependent parabolic equations (Heat equation

on rectangular domains)

d) Improvement and changes in existing multigrid

processes and algorithms (mainly on Poisson

equation) and treatment of new ideas in relaxation

methods

II MULTIGRID METHODS FOR SYSTEMS OF EQUATIONS

The basic ideas of the multigrid method are not

restricted of course to a unique equation and from a

theoretical point of view no special problem could be

expected in implementing multigrid ideas to a system of

equations However special and detailed algorithms for

this problem did not exist and it was extremely important

to get sharp and practical proofs of the efficiency of

multigrid methods for systems of equations One of the

questions we did not know the answer in advance was for

instance what is the appropriate method of relaxation for

a system

In order to answer this and many other important

questions we developed multigrid algorithms for three

5

typical problems starting with a simple one and each

subsequent problem involving more and new complications

relative to the former Moreover since it is usually

wise and necessary in this kind of research to isolate

different questions which may arise we developed the

algorithm for simple geometrical domains (rectangles)

The past experience showed that more complicated geometries

did not affect the efficiency of multigrid methods (see

for instance Shiftan4) For complicated geometries there

are of course more programing problems

21 Cauchy-Riemann Equations

The equations are

Sx-Vy fl (xy) (i)

Ux+Vy = f3(xY) (2)

for 05xsl Osyll

And the boundary conditions

U(Oy) = U0 (y) V(xO) = V0 (x) (3)

U(ly) = Ul(y) V(xl) = Vl(x)

Figure 1

y V (x)

U (y) iU(y)

v0 )x)

6

The data must fulfill an additional condition due to

the continuity equation (2) in order to assure-a solution

to the system This condition is

0V fIvj- 0 )dx f3 y)dxdy(I c 0

The discretization of the problem was accomplished

on a staggered grid This method has several advantages

in problems of this type For instance it allows us to

define easily second order conservative finite differences

The grid is described in figure 2a

Figure 2a

VVkV

U U2 f3 J3 U

U U V U

V V V

The finite differences approximations for equation-()

are defined at grid intersections (X) for instance

71 P jJ3t 1-(v2 V1 J

And the finite differences for equation (2) are defined in

the middle of the cells () ie

1u 3 -U 2 +V3 vk = f3 (6)

As in the continuous case the data must fill an additional

discrete condition equivalent to (4) in the finest grid ie

7

(Ulj1 -U 0 j)+ L (Vii-Voi) = J f3ij (4a)

Several relaxation schemes appropriate to this system

were considered and all these schemes have a good smoothing

factor around 7=5 The most convenient scheme was chosen

because of its simplicity and its small number of arithmetical

operations This method belongs to a new approach of relaxshy

ations developed by Professor A Brandt and is called

Distributed Relaxations

This approach is characterized by the fact that when

passing through a point (or cell) in the grid for which

the difference equations are defined we change the value

of more than one unknown in order to make zero residual

on this point

In the Cauchy-Riemann equations separate relaxations

are performed to each one of the equations according to

the schemes discribed in figures 3 and 4

Efficiency of relaxation in a multigrid process is measuredby the smoothing factor which is defined as the asymptoticratio between the fast components of the errors after a relaxshyation sweep and the same errors before the relaxation A more

exact definition can be found in Brandt 2 More considerationson this subject can be found in Chapter 3 of this report

8

Figure 3 Relaxation of Ux-V = f1

u2+

C

Figure 4 Relaxation of Ux+Vy

2

4f3 U-1i-+

The values U+j0 and V+ C in figures 3 and 4 represents the

new values after the sweep of this point 0 is always chosen

to make the residual zero at the point The relaxation is a

Gauss-Seidel type relaxation and it passes through all the points

(or cells) of the grid in a usual natural order The important

property of this method relies on the fact that when relaxing

each of the equations the residual of the other equation

remains unchanged

Mode Fourier Analysis shows that the smoothing factor for

this method is i =5 the same as in Poisson equation which

is of course equivalent to the Cauchy-Riemann system The

multigrid algorithm shows convergence factor 7 56 which is about the same value as in the Poisson case The

number of operations in the relaxation of the Cauchy-Riemann

See Chapter 3 in this report

9

system is of course higher than in Poisson equations (about

two times more) On the other hand we can say that the

information we get from the solution of the Cauchy-Riemann

system (functions U and V) is also two times the information

we get from the solution of the Poisson equation

The main reason for our treatment of Cauchy-Riemann

system was for learning purposes It is probably the simplest

elliptic system possible and it allows us to treat complishy

cated problems more easily However this system is intershy

esting in itself as we can see in the works of Lomax5 and

Ghil6 These papers describe fast algorithms for the

solution of the Cauchy-Riemann system We think that the

algorithm just described in this report is preferable

because it does not depend in principle on the simplicity

of the domain and because it is at least as fast as Chil

algorithm and even faster asymptotically (for finer and

finer discretizations) The implementation of the algorithms

for nonlinear equation is also much easier in the multigrid

3method

An example of a computer output is given in the Appendix

Output No 1

22 Stokes Equations

The Stokes equation are given by

AU-px = f (7)

t1 V-py = f2 (8)

U +V = f3 (9)

10

The unknowns are U V and P U and V are given on

boundaries Like in Cauchy-Riemann systems The data must

satisfy the condition (4) The discretization is also in the

same staggered grids already described P and f3 are defined

in the centers of the cells f1 and f2 are defined in the

same points as U and V respectively (see figure 2b)

Figure 2b

V-V -v v

U P U P U P U PII

U P U P U P U P

I -V-u- I --V- 0 -- -V I- VVI_

The difference equations for equation (7) are given by

AucxtxlampiP (x+~y)-P(x- S Y)] f(xy) (10)

for (xy) where U is defined A is the usual five

points discrete Laplace operator Li -

Similarly for equation (8)

I CxY+ )-P(xy- ff2 (xy) (1i)(xxyY)- )

for (xy) where V is defined

For the continuity equation (9) the difference

equations were already defined by (6)

Equations (10) and (11) hold for interior points Near

the boundary the difference equations are different but they

are simple and will not be described here

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 3: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

ON SEVERAL ASPECTS AND APPLICATIONSOF THE MULTIGRID METHOD FOR SOLVING PARTIAL

DIFFERENTIAL EQUATIONS

By Nathan Dinar

College of William amp MaryWilliamsburg Virginia 23185

Research Sponsored by NASA Langley Research CenterContract NASl-14972-3

SUMMARY

Several aspects of multigrid methods are briefly described in this report The main subjects include the development ofvery efficient multigrid algorithms for systems of ellipticequations (Cauchy-Riemann Stokes Navier-Stokes) as well asthe development of control and predictiontools (based on localmode Fourier analysis) used to analyze check and improve thesealgorithms

Preliminary research on multigrid algorithms for timedependent parabolic equations is also described This reportdeals also with improvements in existing multigrid processesand algorithms for elliptic equations

Some partial and typical results are given More completeand detailed information will be presented in the authorsPhD Thesis to appear at the Weizmann Institute of ScienceRehovot Israel

I INTRODUCTION

This report deals with several aspects concerning multigrid

methods for fast solution of partial differential equations It

covers the research on this subject for the period August 1977 -

August 1978 when the author was spending his sabbatical at the

NASA Langley Research Center This research is part of a PhD

Research Assistant Department of Mathematics

2

Thesis to appear shortly at the Weizmann Institute of Science

Rehovot Israel including more detailed results and conclusions

The research on multigrid methods began in the early 1970s

by the initiative of Professor A Brandt He is today supervising

several projects on these subjects including the present research

The multigrid method can be applied to a wide range of

problems and therefore it interests many people It is known

to be one of the most powerful and advanced methods used today

The multigrid method uses the fact that the numerical

discrete equations we usually want to solve are not independent

They are derived from a continuous problem whose solution we

want to approximate In the process of the solution of the

discrete equations it is convenient to keep in mind the differshy

ential origin of the problem The use of discrete operators

on several levels of meshes interacting strongly in the process

of the solution allows us to solve the problems on the finest

grid very efficiently with a minimum number of arithmetical

operations (0(N)) where N is the number of equations in the

the finest grid

The multigrid method consists generally of several

processes performed in a given order and defining an itershy

ative cycle These processes include generally

Relaxation (Usually of Gauss-Seidel type) Used only to

smooth the errors ie to reduce these high

frequencies in the error that are not described

in coarser grids

3

Transfer of Residuals to a Coarse Grid This allows us to

define a problem on a coarse grid that is

similar to the original one The solution on

the coarse grid is used to improve the approxishy

mate solution given on the finer grid

Interpolation Used to define a new approximation on a

finer grid given an approximation solution on

a coarse grid

Moreover we can intermix the principles of the multishy

grid method with the grid adaptivity principles which mean

adaption of the order of discrete approximations and mesh

size using total smoothness of the solution This keeps

N as small as possible

These basic ideas and others as well as treatment of

theoretical and practical aspects have been extensively

covered in the papers of Brandt I2 3

This report deals with a-wide spectrum of problems

involved in the development of multigrid methods and multishy

grid software and the principal parts are the following

a) Development of multigrid methods and software

for elliptic systems of equations (Including

Cauchy-Riemann equation Stokes equation and

Navier-Stokes equation

b) Development of control and prediction tools for

various steps and processes involved in the

4

multigrid method (Based on local mode Fourier

analysis)

c) Development of multigrid methods for time

dependent parabolic equations (Heat equation

on rectangular domains)

d) Improvement and changes in existing multigrid

processes and algorithms (mainly on Poisson

equation) and treatment of new ideas in relaxation

methods

II MULTIGRID METHODS FOR SYSTEMS OF EQUATIONS

The basic ideas of the multigrid method are not

restricted of course to a unique equation and from a

theoretical point of view no special problem could be

expected in implementing multigrid ideas to a system of

equations However special and detailed algorithms for

this problem did not exist and it was extremely important

to get sharp and practical proofs of the efficiency of

multigrid methods for systems of equations One of the

questions we did not know the answer in advance was for

instance what is the appropriate method of relaxation for

a system

In order to answer this and many other important

questions we developed multigrid algorithms for three

5

typical problems starting with a simple one and each

subsequent problem involving more and new complications

relative to the former Moreover since it is usually

wise and necessary in this kind of research to isolate

different questions which may arise we developed the

algorithm for simple geometrical domains (rectangles)

The past experience showed that more complicated geometries

did not affect the efficiency of multigrid methods (see

for instance Shiftan4) For complicated geometries there

are of course more programing problems

21 Cauchy-Riemann Equations

The equations are

Sx-Vy fl (xy) (i)

Ux+Vy = f3(xY) (2)

for 05xsl Osyll

And the boundary conditions

U(Oy) = U0 (y) V(xO) = V0 (x) (3)

U(ly) = Ul(y) V(xl) = Vl(x)

Figure 1

y V (x)

U (y) iU(y)

v0 )x)

6

The data must fulfill an additional condition due to

the continuity equation (2) in order to assure-a solution

to the system This condition is

0V fIvj- 0 )dx f3 y)dxdy(I c 0

The discretization of the problem was accomplished

on a staggered grid This method has several advantages

in problems of this type For instance it allows us to

define easily second order conservative finite differences

The grid is described in figure 2a

Figure 2a

VVkV

U U2 f3 J3 U

U U V U

V V V

The finite differences approximations for equation-()

are defined at grid intersections (X) for instance

71 P jJ3t 1-(v2 V1 J

And the finite differences for equation (2) are defined in

the middle of the cells () ie

1u 3 -U 2 +V3 vk = f3 (6)

As in the continuous case the data must fill an additional

discrete condition equivalent to (4) in the finest grid ie

7

(Ulj1 -U 0 j)+ L (Vii-Voi) = J f3ij (4a)

Several relaxation schemes appropriate to this system

were considered and all these schemes have a good smoothing

factor around 7=5 The most convenient scheme was chosen

because of its simplicity and its small number of arithmetical

operations This method belongs to a new approach of relaxshy

ations developed by Professor A Brandt and is called

Distributed Relaxations

This approach is characterized by the fact that when

passing through a point (or cell) in the grid for which

the difference equations are defined we change the value

of more than one unknown in order to make zero residual

on this point

In the Cauchy-Riemann equations separate relaxations

are performed to each one of the equations according to

the schemes discribed in figures 3 and 4

Efficiency of relaxation in a multigrid process is measuredby the smoothing factor which is defined as the asymptoticratio between the fast components of the errors after a relaxshyation sweep and the same errors before the relaxation A more

exact definition can be found in Brandt 2 More considerationson this subject can be found in Chapter 3 of this report

8

Figure 3 Relaxation of Ux-V = f1

u2+

C

Figure 4 Relaxation of Ux+Vy

2

4f3 U-1i-+

The values U+j0 and V+ C in figures 3 and 4 represents the

new values after the sweep of this point 0 is always chosen

to make the residual zero at the point The relaxation is a

Gauss-Seidel type relaxation and it passes through all the points

(or cells) of the grid in a usual natural order The important

property of this method relies on the fact that when relaxing

each of the equations the residual of the other equation

remains unchanged

Mode Fourier Analysis shows that the smoothing factor for

this method is i =5 the same as in Poisson equation which

is of course equivalent to the Cauchy-Riemann system The

multigrid algorithm shows convergence factor 7 56 which is about the same value as in the Poisson case The

number of operations in the relaxation of the Cauchy-Riemann

See Chapter 3 in this report

9

system is of course higher than in Poisson equations (about

two times more) On the other hand we can say that the

information we get from the solution of the Cauchy-Riemann

system (functions U and V) is also two times the information

we get from the solution of the Poisson equation

The main reason for our treatment of Cauchy-Riemann

system was for learning purposes It is probably the simplest

elliptic system possible and it allows us to treat complishy

cated problems more easily However this system is intershy

esting in itself as we can see in the works of Lomax5 and

Ghil6 These papers describe fast algorithms for the

solution of the Cauchy-Riemann system We think that the

algorithm just described in this report is preferable

because it does not depend in principle on the simplicity

of the domain and because it is at least as fast as Chil

algorithm and even faster asymptotically (for finer and

finer discretizations) The implementation of the algorithms

for nonlinear equation is also much easier in the multigrid

3method

An example of a computer output is given in the Appendix

Output No 1

22 Stokes Equations

The Stokes equation are given by

AU-px = f (7)

t1 V-py = f2 (8)

U +V = f3 (9)

10

The unknowns are U V and P U and V are given on

boundaries Like in Cauchy-Riemann systems The data must

satisfy the condition (4) The discretization is also in the

same staggered grids already described P and f3 are defined

in the centers of the cells f1 and f2 are defined in the

same points as U and V respectively (see figure 2b)

Figure 2b

V-V -v v

U P U P U P U PII

U P U P U P U P

I -V-u- I --V- 0 -- -V I- VVI_

The difference equations for equation (7) are given by

AucxtxlampiP (x+~y)-P(x- S Y)] f(xy) (10)

for (xy) where U is defined A is the usual five

points discrete Laplace operator Li -

Similarly for equation (8)

I CxY+ )-P(xy- ff2 (xy) (1i)(xxyY)- )

for (xy) where V is defined

For the continuity equation (9) the difference

equations were already defined by (6)

Equations (10) and (11) hold for interior points Near

the boundary the difference equations are different but they

are simple and will not be described here

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 4: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

2

Thesis to appear shortly at the Weizmann Institute of Science

Rehovot Israel including more detailed results and conclusions

The research on multigrid methods began in the early 1970s

by the initiative of Professor A Brandt He is today supervising

several projects on these subjects including the present research

The multigrid method can be applied to a wide range of

problems and therefore it interests many people It is known

to be one of the most powerful and advanced methods used today

The multigrid method uses the fact that the numerical

discrete equations we usually want to solve are not independent

They are derived from a continuous problem whose solution we

want to approximate In the process of the solution of the

discrete equations it is convenient to keep in mind the differshy

ential origin of the problem The use of discrete operators

on several levels of meshes interacting strongly in the process

of the solution allows us to solve the problems on the finest

grid very efficiently with a minimum number of arithmetical

operations (0(N)) where N is the number of equations in the

the finest grid

The multigrid method consists generally of several

processes performed in a given order and defining an itershy

ative cycle These processes include generally

Relaxation (Usually of Gauss-Seidel type) Used only to

smooth the errors ie to reduce these high

frequencies in the error that are not described

in coarser grids

3

Transfer of Residuals to a Coarse Grid This allows us to

define a problem on a coarse grid that is

similar to the original one The solution on

the coarse grid is used to improve the approxishy

mate solution given on the finer grid

Interpolation Used to define a new approximation on a

finer grid given an approximation solution on

a coarse grid

Moreover we can intermix the principles of the multishy

grid method with the grid adaptivity principles which mean

adaption of the order of discrete approximations and mesh

size using total smoothness of the solution This keeps

N as small as possible

These basic ideas and others as well as treatment of

theoretical and practical aspects have been extensively

covered in the papers of Brandt I2 3

This report deals with a-wide spectrum of problems

involved in the development of multigrid methods and multishy

grid software and the principal parts are the following

a) Development of multigrid methods and software

for elliptic systems of equations (Including

Cauchy-Riemann equation Stokes equation and

Navier-Stokes equation

b) Development of control and prediction tools for

various steps and processes involved in the

4

multigrid method (Based on local mode Fourier

analysis)

c) Development of multigrid methods for time

dependent parabolic equations (Heat equation

on rectangular domains)

d) Improvement and changes in existing multigrid

processes and algorithms (mainly on Poisson

equation) and treatment of new ideas in relaxation

methods

II MULTIGRID METHODS FOR SYSTEMS OF EQUATIONS

The basic ideas of the multigrid method are not

restricted of course to a unique equation and from a

theoretical point of view no special problem could be

expected in implementing multigrid ideas to a system of

equations However special and detailed algorithms for

this problem did not exist and it was extremely important

to get sharp and practical proofs of the efficiency of

multigrid methods for systems of equations One of the

questions we did not know the answer in advance was for

instance what is the appropriate method of relaxation for

a system

In order to answer this and many other important

questions we developed multigrid algorithms for three

5

typical problems starting with a simple one and each

subsequent problem involving more and new complications

relative to the former Moreover since it is usually

wise and necessary in this kind of research to isolate

different questions which may arise we developed the

algorithm for simple geometrical domains (rectangles)

The past experience showed that more complicated geometries

did not affect the efficiency of multigrid methods (see

for instance Shiftan4) For complicated geometries there

are of course more programing problems

21 Cauchy-Riemann Equations

The equations are

Sx-Vy fl (xy) (i)

Ux+Vy = f3(xY) (2)

for 05xsl Osyll

And the boundary conditions

U(Oy) = U0 (y) V(xO) = V0 (x) (3)

U(ly) = Ul(y) V(xl) = Vl(x)

Figure 1

y V (x)

U (y) iU(y)

v0 )x)

6

The data must fulfill an additional condition due to

the continuity equation (2) in order to assure-a solution

to the system This condition is

0V fIvj- 0 )dx f3 y)dxdy(I c 0

The discretization of the problem was accomplished

on a staggered grid This method has several advantages

in problems of this type For instance it allows us to

define easily second order conservative finite differences

The grid is described in figure 2a

Figure 2a

VVkV

U U2 f3 J3 U

U U V U

V V V

The finite differences approximations for equation-()

are defined at grid intersections (X) for instance

71 P jJ3t 1-(v2 V1 J

And the finite differences for equation (2) are defined in

the middle of the cells () ie

1u 3 -U 2 +V3 vk = f3 (6)

As in the continuous case the data must fill an additional

discrete condition equivalent to (4) in the finest grid ie

7

(Ulj1 -U 0 j)+ L (Vii-Voi) = J f3ij (4a)

Several relaxation schemes appropriate to this system

were considered and all these schemes have a good smoothing

factor around 7=5 The most convenient scheme was chosen

because of its simplicity and its small number of arithmetical

operations This method belongs to a new approach of relaxshy

ations developed by Professor A Brandt and is called

Distributed Relaxations

This approach is characterized by the fact that when

passing through a point (or cell) in the grid for which

the difference equations are defined we change the value

of more than one unknown in order to make zero residual

on this point

In the Cauchy-Riemann equations separate relaxations

are performed to each one of the equations according to

the schemes discribed in figures 3 and 4

Efficiency of relaxation in a multigrid process is measuredby the smoothing factor which is defined as the asymptoticratio between the fast components of the errors after a relaxshyation sweep and the same errors before the relaxation A more

exact definition can be found in Brandt 2 More considerationson this subject can be found in Chapter 3 of this report

8

Figure 3 Relaxation of Ux-V = f1

u2+

C

Figure 4 Relaxation of Ux+Vy

2

4f3 U-1i-+

The values U+j0 and V+ C in figures 3 and 4 represents the

new values after the sweep of this point 0 is always chosen

to make the residual zero at the point The relaxation is a

Gauss-Seidel type relaxation and it passes through all the points

(or cells) of the grid in a usual natural order The important

property of this method relies on the fact that when relaxing

each of the equations the residual of the other equation

remains unchanged

Mode Fourier Analysis shows that the smoothing factor for

this method is i =5 the same as in Poisson equation which

is of course equivalent to the Cauchy-Riemann system The

multigrid algorithm shows convergence factor 7 56 which is about the same value as in the Poisson case The

number of operations in the relaxation of the Cauchy-Riemann

See Chapter 3 in this report

9

system is of course higher than in Poisson equations (about

two times more) On the other hand we can say that the

information we get from the solution of the Cauchy-Riemann

system (functions U and V) is also two times the information

we get from the solution of the Poisson equation

The main reason for our treatment of Cauchy-Riemann

system was for learning purposes It is probably the simplest

elliptic system possible and it allows us to treat complishy

cated problems more easily However this system is intershy

esting in itself as we can see in the works of Lomax5 and

Ghil6 These papers describe fast algorithms for the

solution of the Cauchy-Riemann system We think that the

algorithm just described in this report is preferable

because it does not depend in principle on the simplicity

of the domain and because it is at least as fast as Chil

algorithm and even faster asymptotically (for finer and

finer discretizations) The implementation of the algorithms

for nonlinear equation is also much easier in the multigrid

3method

An example of a computer output is given in the Appendix

Output No 1

22 Stokes Equations

The Stokes equation are given by

AU-px = f (7)

t1 V-py = f2 (8)

U +V = f3 (9)

10

The unknowns are U V and P U and V are given on

boundaries Like in Cauchy-Riemann systems The data must

satisfy the condition (4) The discretization is also in the

same staggered grids already described P and f3 are defined

in the centers of the cells f1 and f2 are defined in the

same points as U and V respectively (see figure 2b)

Figure 2b

V-V -v v

U P U P U P U PII

U P U P U P U P

I -V-u- I --V- 0 -- -V I- VVI_

The difference equations for equation (7) are given by

AucxtxlampiP (x+~y)-P(x- S Y)] f(xy) (10)

for (xy) where U is defined A is the usual five

points discrete Laplace operator Li -

Similarly for equation (8)

I CxY+ )-P(xy- ff2 (xy) (1i)(xxyY)- )

for (xy) where V is defined

For the continuity equation (9) the difference

equations were already defined by (6)

Equations (10) and (11) hold for interior points Near

the boundary the difference equations are different but they

are simple and will not be described here

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 5: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

3

Transfer of Residuals to a Coarse Grid This allows us to

define a problem on a coarse grid that is

similar to the original one The solution on

the coarse grid is used to improve the approxishy

mate solution given on the finer grid

Interpolation Used to define a new approximation on a

finer grid given an approximation solution on

a coarse grid

Moreover we can intermix the principles of the multishy

grid method with the grid adaptivity principles which mean

adaption of the order of discrete approximations and mesh

size using total smoothness of the solution This keeps

N as small as possible

These basic ideas and others as well as treatment of

theoretical and practical aspects have been extensively

covered in the papers of Brandt I2 3

This report deals with a-wide spectrum of problems

involved in the development of multigrid methods and multishy

grid software and the principal parts are the following

a) Development of multigrid methods and software

for elliptic systems of equations (Including

Cauchy-Riemann equation Stokes equation and

Navier-Stokes equation

b) Development of control and prediction tools for

various steps and processes involved in the

4

multigrid method (Based on local mode Fourier

analysis)

c) Development of multigrid methods for time

dependent parabolic equations (Heat equation

on rectangular domains)

d) Improvement and changes in existing multigrid

processes and algorithms (mainly on Poisson

equation) and treatment of new ideas in relaxation

methods

II MULTIGRID METHODS FOR SYSTEMS OF EQUATIONS

The basic ideas of the multigrid method are not

restricted of course to a unique equation and from a

theoretical point of view no special problem could be

expected in implementing multigrid ideas to a system of

equations However special and detailed algorithms for

this problem did not exist and it was extremely important

to get sharp and practical proofs of the efficiency of

multigrid methods for systems of equations One of the

questions we did not know the answer in advance was for

instance what is the appropriate method of relaxation for

a system

In order to answer this and many other important

questions we developed multigrid algorithms for three

5

typical problems starting with a simple one and each

subsequent problem involving more and new complications

relative to the former Moreover since it is usually

wise and necessary in this kind of research to isolate

different questions which may arise we developed the

algorithm for simple geometrical domains (rectangles)

The past experience showed that more complicated geometries

did not affect the efficiency of multigrid methods (see

for instance Shiftan4) For complicated geometries there

are of course more programing problems

21 Cauchy-Riemann Equations

The equations are

Sx-Vy fl (xy) (i)

Ux+Vy = f3(xY) (2)

for 05xsl Osyll

And the boundary conditions

U(Oy) = U0 (y) V(xO) = V0 (x) (3)

U(ly) = Ul(y) V(xl) = Vl(x)

Figure 1

y V (x)

U (y) iU(y)

v0 )x)

6

The data must fulfill an additional condition due to

the continuity equation (2) in order to assure-a solution

to the system This condition is

0V fIvj- 0 )dx f3 y)dxdy(I c 0

The discretization of the problem was accomplished

on a staggered grid This method has several advantages

in problems of this type For instance it allows us to

define easily second order conservative finite differences

The grid is described in figure 2a

Figure 2a

VVkV

U U2 f3 J3 U

U U V U

V V V

The finite differences approximations for equation-()

are defined at grid intersections (X) for instance

71 P jJ3t 1-(v2 V1 J

And the finite differences for equation (2) are defined in

the middle of the cells () ie

1u 3 -U 2 +V3 vk = f3 (6)

As in the continuous case the data must fill an additional

discrete condition equivalent to (4) in the finest grid ie

7

(Ulj1 -U 0 j)+ L (Vii-Voi) = J f3ij (4a)

Several relaxation schemes appropriate to this system

were considered and all these schemes have a good smoothing

factor around 7=5 The most convenient scheme was chosen

because of its simplicity and its small number of arithmetical

operations This method belongs to a new approach of relaxshy

ations developed by Professor A Brandt and is called

Distributed Relaxations

This approach is characterized by the fact that when

passing through a point (or cell) in the grid for which

the difference equations are defined we change the value

of more than one unknown in order to make zero residual

on this point

In the Cauchy-Riemann equations separate relaxations

are performed to each one of the equations according to

the schemes discribed in figures 3 and 4

Efficiency of relaxation in a multigrid process is measuredby the smoothing factor which is defined as the asymptoticratio between the fast components of the errors after a relaxshyation sweep and the same errors before the relaxation A more

exact definition can be found in Brandt 2 More considerationson this subject can be found in Chapter 3 of this report

8

Figure 3 Relaxation of Ux-V = f1

u2+

C

Figure 4 Relaxation of Ux+Vy

2

4f3 U-1i-+

The values U+j0 and V+ C in figures 3 and 4 represents the

new values after the sweep of this point 0 is always chosen

to make the residual zero at the point The relaxation is a

Gauss-Seidel type relaxation and it passes through all the points

(or cells) of the grid in a usual natural order The important

property of this method relies on the fact that when relaxing

each of the equations the residual of the other equation

remains unchanged

Mode Fourier Analysis shows that the smoothing factor for

this method is i =5 the same as in Poisson equation which

is of course equivalent to the Cauchy-Riemann system The

multigrid algorithm shows convergence factor 7 56 which is about the same value as in the Poisson case The

number of operations in the relaxation of the Cauchy-Riemann

See Chapter 3 in this report

9

system is of course higher than in Poisson equations (about

two times more) On the other hand we can say that the

information we get from the solution of the Cauchy-Riemann

system (functions U and V) is also two times the information

we get from the solution of the Poisson equation

The main reason for our treatment of Cauchy-Riemann

system was for learning purposes It is probably the simplest

elliptic system possible and it allows us to treat complishy

cated problems more easily However this system is intershy

esting in itself as we can see in the works of Lomax5 and

Ghil6 These papers describe fast algorithms for the

solution of the Cauchy-Riemann system We think that the

algorithm just described in this report is preferable

because it does not depend in principle on the simplicity

of the domain and because it is at least as fast as Chil

algorithm and even faster asymptotically (for finer and

finer discretizations) The implementation of the algorithms

for nonlinear equation is also much easier in the multigrid

3method

An example of a computer output is given in the Appendix

Output No 1

22 Stokes Equations

The Stokes equation are given by

AU-px = f (7)

t1 V-py = f2 (8)

U +V = f3 (9)

10

The unknowns are U V and P U and V are given on

boundaries Like in Cauchy-Riemann systems The data must

satisfy the condition (4) The discretization is also in the

same staggered grids already described P and f3 are defined

in the centers of the cells f1 and f2 are defined in the

same points as U and V respectively (see figure 2b)

Figure 2b

V-V -v v

U P U P U P U PII

U P U P U P U P

I -V-u- I --V- 0 -- -V I- VVI_

The difference equations for equation (7) are given by

AucxtxlampiP (x+~y)-P(x- S Y)] f(xy) (10)

for (xy) where U is defined A is the usual five

points discrete Laplace operator Li -

Similarly for equation (8)

I CxY+ )-P(xy- ff2 (xy) (1i)(xxyY)- )

for (xy) where V is defined

For the continuity equation (9) the difference

equations were already defined by (6)

Equations (10) and (11) hold for interior points Near

the boundary the difference equations are different but they

are simple and will not be described here

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 6: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

4

multigrid method (Based on local mode Fourier

analysis)

c) Development of multigrid methods for time

dependent parabolic equations (Heat equation

on rectangular domains)

d) Improvement and changes in existing multigrid

processes and algorithms (mainly on Poisson

equation) and treatment of new ideas in relaxation

methods

II MULTIGRID METHODS FOR SYSTEMS OF EQUATIONS

The basic ideas of the multigrid method are not

restricted of course to a unique equation and from a

theoretical point of view no special problem could be

expected in implementing multigrid ideas to a system of

equations However special and detailed algorithms for

this problem did not exist and it was extremely important

to get sharp and practical proofs of the efficiency of

multigrid methods for systems of equations One of the

questions we did not know the answer in advance was for

instance what is the appropriate method of relaxation for

a system

In order to answer this and many other important

questions we developed multigrid algorithms for three

5

typical problems starting with a simple one and each

subsequent problem involving more and new complications

relative to the former Moreover since it is usually

wise and necessary in this kind of research to isolate

different questions which may arise we developed the

algorithm for simple geometrical domains (rectangles)

The past experience showed that more complicated geometries

did not affect the efficiency of multigrid methods (see

for instance Shiftan4) For complicated geometries there

are of course more programing problems

21 Cauchy-Riemann Equations

The equations are

Sx-Vy fl (xy) (i)

Ux+Vy = f3(xY) (2)

for 05xsl Osyll

And the boundary conditions

U(Oy) = U0 (y) V(xO) = V0 (x) (3)

U(ly) = Ul(y) V(xl) = Vl(x)

Figure 1

y V (x)

U (y) iU(y)

v0 )x)

6

The data must fulfill an additional condition due to

the continuity equation (2) in order to assure-a solution

to the system This condition is

0V fIvj- 0 )dx f3 y)dxdy(I c 0

The discretization of the problem was accomplished

on a staggered grid This method has several advantages

in problems of this type For instance it allows us to

define easily second order conservative finite differences

The grid is described in figure 2a

Figure 2a

VVkV

U U2 f3 J3 U

U U V U

V V V

The finite differences approximations for equation-()

are defined at grid intersections (X) for instance

71 P jJ3t 1-(v2 V1 J

And the finite differences for equation (2) are defined in

the middle of the cells () ie

1u 3 -U 2 +V3 vk = f3 (6)

As in the continuous case the data must fill an additional

discrete condition equivalent to (4) in the finest grid ie

7

(Ulj1 -U 0 j)+ L (Vii-Voi) = J f3ij (4a)

Several relaxation schemes appropriate to this system

were considered and all these schemes have a good smoothing

factor around 7=5 The most convenient scheme was chosen

because of its simplicity and its small number of arithmetical

operations This method belongs to a new approach of relaxshy

ations developed by Professor A Brandt and is called

Distributed Relaxations

This approach is characterized by the fact that when

passing through a point (or cell) in the grid for which

the difference equations are defined we change the value

of more than one unknown in order to make zero residual

on this point

In the Cauchy-Riemann equations separate relaxations

are performed to each one of the equations according to

the schemes discribed in figures 3 and 4

Efficiency of relaxation in a multigrid process is measuredby the smoothing factor which is defined as the asymptoticratio between the fast components of the errors after a relaxshyation sweep and the same errors before the relaxation A more

exact definition can be found in Brandt 2 More considerationson this subject can be found in Chapter 3 of this report

8

Figure 3 Relaxation of Ux-V = f1

u2+

C

Figure 4 Relaxation of Ux+Vy

2

4f3 U-1i-+

The values U+j0 and V+ C in figures 3 and 4 represents the

new values after the sweep of this point 0 is always chosen

to make the residual zero at the point The relaxation is a

Gauss-Seidel type relaxation and it passes through all the points

(or cells) of the grid in a usual natural order The important

property of this method relies on the fact that when relaxing

each of the equations the residual of the other equation

remains unchanged

Mode Fourier Analysis shows that the smoothing factor for

this method is i =5 the same as in Poisson equation which

is of course equivalent to the Cauchy-Riemann system The

multigrid algorithm shows convergence factor 7 56 which is about the same value as in the Poisson case The

number of operations in the relaxation of the Cauchy-Riemann

See Chapter 3 in this report

9

system is of course higher than in Poisson equations (about

two times more) On the other hand we can say that the

information we get from the solution of the Cauchy-Riemann

system (functions U and V) is also two times the information

we get from the solution of the Poisson equation

The main reason for our treatment of Cauchy-Riemann

system was for learning purposes It is probably the simplest

elliptic system possible and it allows us to treat complishy

cated problems more easily However this system is intershy

esting in itself as we can see in the works of Lomax5 and

Ghil6 These papers describe fast algorithms for the

solution of the Cauchy-Riemann system We think that the

algorithm just described in this report is preferable

because it does not depend in principle on the simplicity

of the domain and because it is at least as fast as Chil

algorithm and even faster asymptotically (for finer and

finer discretizations) The implementation of the algorithms

for nonlinear equation is also much easier in the multigrid

3method

An example of a computer output is given in the Appendix

Output No 1

22 Stokes Equations

The Stokes equation are given by

AU-px = f (7)

t1 V-py = f2 (8)

U +V = f3 (9)

10

The unknowns are U V and P U and V are given on

boundaries Like in Cauchy-Riemann systems The data must

satisfy the condition (4) The discretization is also in the

same staggered grids already described P and f3 are defined

in the centers of the cells f1 and f2 are defined in the

same points as U and V respectively (see figure 2b)

Figure 2b

V-V -v v

U P U P U P U PII

U P U P U P U P

I -V-u- I --V- 0 -- -V I- VVI_

The difference equations for equation (7) are given by

AucxtxlampiP (x+~y)-P(x- S Y)] f(xy) (10)

for (xy) where U is defined A is the usual five

points discrete Laplace operator Li -

Similarly for equation (8)

I CxY+ )-P(xy- ff2 (xy) (1i)(xxyY)- )

for (xy) where V is defined

For the continuity equation (9) the difference

equations were already defined by (6)

Equations (10) and (11) hold for interior points Near

the boundary the difference equations are different but they

are simple and will not be described here

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 7: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

5

typical problems starting with a simple one and each

subsequent problem involving more and new complications

relative to the former Moreover since it is usually

wise and necessary in this kind of research to isolate

different questions which may arise we developed the

algorithm for simple geometrical domains (rectangles)

The past experience showed that more complicated geometries

did not affect the efficiency of multigrid methods (see

for instance Shiftan4) For complicated geometries there

are of course more programing problems

21 Cauchy-Riemann Equations

The equations are

Sx-Vy fl (xy) (i)

Ux+Vy = f3(xY) (2)

for 05xsl Osyll

And the boundary conditions

U(Oy) = U0 (y) V(xO) = V0 (x) (3)

U(ly) = Ul(y) V(xl) = Vl(x)

Figure 1

y V (x)

U (y) iU(y)

v0 )x)

6

The data must fulfill an additional condition due to

the continuity equation (2) in order to assure-a solution

to the system This condition is

0V fIvj- 0 )dx f3 y)dxdy(I c 0

The discretization of the problem was accomplished

on a staggered grid This method has several advantages

in problems of this type For instance it allows us to

define easily second order conservative finite differences

The grid is described in figure 2a

Figure 2a

VVkV

U U2 f3 J3 U

U U V U

V V V

The finite differences approximations for equation-()

are defined at grid intersections (X) for instance

71 P jJ3t 1-(v2 V1 J

And the finite differences for equation (2) are defined in

the middle of the cells () ie

1u 3 -U 2 +V3 vk = f3 (6)

As in the continuous case the data must fill an additional

discrete condition equivalent to (4) in the finest grid ie

7

(Ulj1 -U 0 j)+ L (Vii-Voi) = J f3ij (4a)

Several relaxation schemes appropriate to this system

were considered and all these schemes have a good smoothing

factor around 7=5 The most convenient scheme was chosen

because of its simplicity and its small number of arithmetical

operations This method belongs to a new approach of relaxshy

ations developed by Professor A Brandt and is called

Distributed Relaxations

This approach is characterized by the fact that when

passing through a point (or cell) in the grid for which

the difference equations are defined we change the value

of more than one unknown in order to make zero residual

on this point

In the Cauchy-Riemann equations separate relaxations

are performed to each one of the equations according to

the schemes discribed in figures 3 and 4

Efficiency of relaxation in a multigrid process is measuredby the smoothing factor which is defined as the asymptoticratio between the fast components of the errors after a relaxshyation sweep and the same errors before the relaxation A more

exact definition can be found in Brandt 2 More considerationson this subject can be found in Chapter 3 of this report

8

Figure 3 Relaxation of Ux-V = f1

u2+

C

Figure 4 Relaxation of Ux+Vy

2

4f3 U-1i-+

The values U+j0 and V+ C in figures 3 and 4 represents the

new values after the sweep of this point 0 is always chosen

to make the residual zero at the point The relaxation is a

Gauss-Seidel type relaxation and it passes through all the points

(or cells) of the grid in a usual natural order The important

property of this method relies on the fact that when relaxing

each of the equations the residual of the other equation

remains unchanged

Mode Fourier Analysis shows that the smoothing factor for

this method is i =5 the same as in Poisson equation which

is of course equivalent to the Cauchy-Riemann system The

multigrid algorithm shows convergence factor 7 56 which is about the same value as in the Poisson case The

number of operations in the relaxation of the Cauchy-Riemann

See Chapter 3 in this report

9

system is of course higher than in Poisson equations (about

two times more) On the other hand we can say that the

information we get from the solution of the Cauchy-Riemann

system (functions U and V) is also two times the information

we get from the solution of the Poisson equation

The main reason for our treatment of Cauchy-Riemann

system was for learning purposes It is probably the simplest

elliptic system possible and it allows us to treat complishy

cated problems more easily However this system is intershy

esting in itself as we can see in the works of Lomax5 and

Ghil6 These papers describe fast algorithms for the

solution of the Cauchy-Riemann system We think that the

algorithm just described in this report is preferable

because it does not depend in principle on the simplicity

of the domain and because it is at least as fast as Chil

algorithm and even faster asymptotically (for finer and

finer discretizations) The implementation of the algorithms

for nonlinear equation is also much easier in the multigrid

3method

An example of a computer output is given in the Appendix

Output No 1

22 Stokes Equations

The Stokes equation are given by

AU-px = f (7)

t1 V-py = f2 (8)

U +V = f3 (9)

10

The unknowns are U V and P U and V are given on

boundaries Like in Cauchy-Riemann systems The data must

satisfy the condition (4) The discretization is also in the

same staggered grids already described P and f3 are defined

in the centers of the cells f1 and f2 are defined in the

same points as U and V respectively (see figure 2b)

Figure 2b

V-V -v v

U P U P U P U PII

U P U P U P U P

I -V-u- I --V- 0 -- -V I- VVI_

The difference equations for equation (7) are given by

AucxtxlampiP (x+~y)-P(x- S Y)] f(xy) (10)

for (xy) where U is defined A is the usual five

points discrete Laplace operator Li -

Similarly for equation (8)

I CxY+ )-P(xy- ff2 (xy) (1i)(xxyY)- )

for (xy) where V is defined

For the continuity equation (9) the difference

equations were already defined by (6)

Equations (10) and (11) hold for interior points Near

the boundary the difference equations are different but they

are simple and will not be described here

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 8: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

6

The data must fulfill an additional condition due to

the continuity equation (2) in order to assure-a solution

to the system This condition is

0V fIvj- 0 )dx f3 y)dxdy(I c 0

The discretization of the problem was accomplished

on a staggered grid This method has several advantages

in problems of this type For instance it allows us to

define easily second order conservative finite differences

The grid is described in figure 2a

Figure 2a

VVkV

U U2 f3 J3 U

U U V U

V V V

The finite differences approximations for equation-()

are defined at grid intersections (X) for instance

71 P jJ3t 1-(v2 V1 J

And the finite differences for equation (2) are defined in

the middle of the cells () ie

1u 3 -U 2 +V3 vk = f3 (6)

As in the continuous case the data must fill an additional

discrete condition equivalent to (4) in the finest grid ie

7

(Ulj1 -U 0 j)+ L (Vii-Voi) = J f3ij (4a)

Several relaxation schemes appropriate to this system

were considered and all these schemes have a good smoothing

factor around 7=5 The most convenient scheme was chosen

because of its simplicity and its small number of arithmetical

operations This method belongs to a new approach of relaxshy

ations developed by Professor A Brandt and is called

Distributed Relaxations

This approach is characterized by the fact that when

passing through a point (or cell) in the grid for which

the difference equations are defined we change the value

of more than one unknown in order to make zero residual

on this point

In the Cauchy-Riemann equations separate relaxations

are performed to each one of the equations according to

the schemes discribed in figures 3 and 4

Efficiency of relaxation in a multigrid process is measuredby the smoothing factor which is defined as the asymptoticratio between the fast components of the errors after a relaxshyation sweep and the same errors before the relaxation A more

exact definition can be found in Brandt 2 More considerationson this subject can be found in Chapter 3 of this report

8

Figure 3 Relaxation of Ux-V = f1

u2+

C

Figure 4 Relaxation of Ux+Vy

2

4f3 U-1i-+

The values U+j0 and V+ C in figures 3 and 4 represents the

new values after the sweep of this point 0 is always chosen

to make the residual zero at the point The relaxation is a

Gauss-Seidel type relaxation and it passes through all the points

(or cells) of the grid in a usual natural order The important

property of this method relies on the fact that when relaxing

each of the equations the residual of the other equation

remains unchanged

Mode Fourier Analysis shows that the smoothing factor for

this method is i =5 the same as in Poisson equation which

is of course equivalent to the Cauchy-Riemann system The

multigrid algorithm shows convergence factor 7 56 which is about the same value as in the Poisson case The

number of operations in the relaxation of the Cauchy-Riemann

See Chapter 3 in this report

9

system is of course higher than in Poisson equations (about

two times more) On the other hand we can say that the

information we get from the solution of the Cauchy-Riemann

system (functions U and V) is also two times the information

we get from the solution of the Poisson equation

The main reason for our treatment of Cauchy-Riemann

system was for learning purposes It is probably the simplest

elliptic system possible and it allows us to treat complishy

cated problems more easily However this system is intershy

esting in itself as we can see in the works of Lomax5 and

Ghil6 These papers describe fast algorithms for the

solution of the Cauchy-Riemann system We think that the

algorithm just described in this report is preferable

because it does not depend in principle on the simplicity

of the domain and because it is at least as fast as Chil

algorithm and even faster asymptotically (for finer and

finer discretizations) The implementation of the algorithms

for nonlinear equation is also much easier in the multigrid

3method

An example of a computer output is given in the Appendix

Output No 1

22 Stokes Equations

The Stokes equation are given by

AU-px = f (7)

t1 V-py = f2 (8)

U +V = f3 (9)

10

The unknowns are U V and P U and V are given on

boundaries Like in Cauchy-Riemann systems The data must

satisfy the condition (4) The discretization is also in the

same staggered grids already described P and f3 are defined

in the centers of the cells f1 and f2 are defined in the

same points as U and V respectively (see figure 2b)

Figure 2b

V-V -v v

U P U P U P U PII

U P U P U P U P

I -V-u- I --V- 0 -- -V I- VVI_

The difference equations for equation (7) are given by

AucxtxlampiP (x+~y)-P(x- S Y)] f(xy) (10)

for (xy) where U is defined A is the usual five

points discrete Laplace operator Li -

Similarly for equation (8)

I CxY+ )-P(xy- ff2 (xy) (1i)(xxyY)- )

for (xy) where V is defined

For the continuity equation (9) the difference

equations were already defined by (6)

Equations (10) and (11) hold for interior points Near

the boundary the difference equations are different but they

are simple and will not be described here

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 9: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

7

(Ulj1 -U 0 j)+ L (Vii-Voi) = J f3ij (4a)

Several relaxation schemes appropriate to this system

were considered and all these schemes have a good smoothing

factor around 7=5 The most convenient scheme was chosen

because of its simplicity and its small number of arithmetical

operations This method belongs to a new approach of relaxshy

ations developed by Professor A Brandt and is called

Distributed Relaxations

This approach is characterized by the fact that when

passing through a point (or cell) in the grid for which

the difference equations are defined we change the value

of more than one unknown in order to make zero residual

on this point

In the Cauchy-Riemann equations separate relaxations

are performed to each one of the equations according to

the schemes discribed in figures 3 and 4

Efficiency of relaxation in a multigrid process is measuredby the smoothing factor which is defined as the asymptoticratio between the fast components of the errors after a relaxshyation sweep and the same errors before the relaxation A more

exact definition can be found in Brandt 2 More considerationson this subject can be found in Chapter 3 of this report

8

Figure 3 Relaxation of Ux-V = f1

u2+

C

Figure 4 Relaxation of Ux+Vy

2

4f3 U-1i-+

The values U+j0 and V+ C in figures 3 and 4 represents the

new values after the sweep of this point 0 is always chosen

to make the residual zero at the point The relaxation is a

Gauss-Seidel type relaxation and it passes through all the points

(or cells) of the grid in a usual natural order The important

property of this method relies on the fact that when relaxing

each of the equations the residual of the other equation

remains unchanged

Mode Fourier Analysis shows that the smoothing factor for

this method is i =5 the same as in Poisson equation which

is of course equivalent to the Cauchy-Riemann system The

multigrid algorithm shows convergence factor 7 56 which is about the same value as in the Poisson case The

number of operations in the relaxation of the Cauchy-Riemann

See Chapter 3 in this report

9

system is of course higher than in Poisson equations (about

two times more) On the other hand we can say that the

information we get from the solution of the Cauchy-Riemann

system (functions U and V) is also two times the information

we get from the solution of the Poisson equation

The main reason for our treatment of Cauchy-Riemann

system was for learning purposes It is probably the simplest

elliptic system possible and it allows us to treat complishy

cated problems more easily However this system is intershy

esting in itself as we can see in the works of Lomax5 and

Ghil6 These papers describe fast algorithms for the

solution of the Cauchy-Riemann system We think that the

algorithm just described in this report is preferable

because it does not depend in principle on the simplicity

of the domain and because it is at least as fast as Chil

algorithm and even faster asymptotically (for finer and

finer discretizations) The implementation of the algorithms

for nonlinear equation is also much easier in the multigrid

3method

An example of a computer output is given in the Appendix

Output No 1

22 Stokes Equations

The Stokes equation are given by

AU-px = f (7)

t1 V-py = f2 (8)

U +V = f3 (9)

10

The unknowns are U V and P U and V are given on

boundaries Like in Cauchy-Riemann systems The data must

satisfy the condition (4) The discretization is also in the

same staggered grids already described P and f3 are defined

in the centers of the cells f1 and f2 are defined in the

same points as U and V respectively (see figure 2b)

Figure 2b

V-V -v v

U P U P U P U PII

U P U P U P U P

I -V-u- I --V- 0 -- -V I- VVI_

The difference equations for equation (7) are given by

AucxtxlampiP (x+~y)-P(x- S Y)] f(xy) (10)

for (xy) where U is defined A is the usual five

points discrete Laplace operator Li -

Similarly for equation (8)

I CxY+ )-P(xy- ff2 (xy) (1i)(xxyY)- )

for (xy) where V is defined

For the continuity equation (9) the difference

equations were already defined by (6)

Equations (10) and (11) hold for interior points Near

the boundary the difference equations are different but they

are simple and will not be described here

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 10: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

8

Figure 3 Relaxation of Ux-V = f1

u2+

C

Figure 4 Relaxation of Ux+Vy

2

4f3 U-1i-+

The values U+j0 and V+ C in figures 3 and 4 represents the

new values after the sweep of this point 0 is always chosen

to make the residual zero at the point The relaxation is a

Gauss-Seidel type relaxation and it passes through all the points

(or cells) of the grid in a usual natural order The important

property of this method relies on the fact that when relaxing

each of the equations the residual of the other equation

remains unchanged

Mode Fourier Analysis shows that the smoothing factor for

this method is i =5 the same as in Poisson equation which

is of course equivalent to the Cauchy-Riemann system The

multigrid algorithm shows convergence factor 7 56 which is about the same value as in the Poisson case The

number of operations in the relaxation of the Cauchy-Riemann

See Chapter 3 in this report

9

system is of course higher than in Poisson equations (about

two times more) On the other hand we can say that the

information we get from the solution of the Cauchy-Riemann

system (functions U and V) is also two times the information

we get from the solution of the Poisson equation

The main reason for our treatment of Cauchy-Riemann

system was for learning purposes It is probably the simplest

elliptic system possible and it allows us to treat complishy

cated problems more easily However this system is intershy

esting in itself as we can see in the works of Lomax5 and

Ghil6 These papers describe fast algorithms for the

solution of the Cauchy-Riemann system We think that the

algorithm just described in this report is preferable

because it does not depend in principle on the simplicity

of the domain and because it is at least as fast as Chil

algorithm and even faster asymptotically (for finer and

finer discretizations) The implementation of the algorithms

for nonlinear equation is also much easier in the multigrid

3method

An example of a computer output is given in the Appendix

Output No 1

22 Stokes Equations

The Stokes equation are given by

AU-px = f (7)

t1 V-py = f2 (8)

U +V = f3 (9)

10

The unknowns are U V and P U and V are given on

boundaries Like in Cauchy-Riemann systems The data must

satisfy the condition (4) The discretization is also in the

same staggered grids already described P and f3 are defined

in the centers of the cells f1 and f2 are defined in the

same points as U and V respectively (see figure 2b)

Figure 2b

V-V -v v

U P U P U P U PII

U P U P U P U P

I -V-u- I --V- 0 -- -V I- VVI_

The difference equations for equation (7) are given by

AucxtxlampiP (x+~y)-P(x- S Y)] f(xy) (10)

for (xy) where U is defined A is the usual five

points discrete Laplace operator Li -

Similarly for equation (8)

I CxY+ )-P(xy- ff2 (xy) (1i)(xxyY)- )

for (xy) where V is defined

For the continuity equation (9) the difference

equations were already defined by (6)

Equations (10) and (11) hold for interior points Near

the boundary the difference equations are different but they

are simple and will not be described here

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 11: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

9

system is of course higher than in Poisson equations (about

two times more) On the other hand we can say that the

information we get from the solution of the Cauchy-Riemann

system (functions U and V) is also two times the information

we get from the solution of the Poisson equation

The main reason for our treatment of Cauchy-Riemann

system was for learning purposes It is probably the simplest

elliptic system possible and it allows us to treat complishy

cated problems more easily However this system is intershy

esting in itself as we can see in the works of Lomax5 and

Ghil6 These papers describe fast algorithms for the

solution of the Cauchy-Riemann system We think that the

algorithm just described in this report is preferable

because it does not depend in principle on the simplicity

of the domain and because it is at least as fast as Chil

algorithm and even faster asymptotically (for finer and

finer discretizations) The implementation of the algorithms

for nonlinear equation is also much easier in the multigrid

3method

An example of a computer output is given in the Appendix

Output No 1

22 Stokes Equations

The Stokes equation are given by

AU-px = f (7)

t1 V-py = f2 (8)

U +V = f3 (9)

10

The unknowns are U V and P U and V are given on

boundaries Like in Cauchy-Riemann systems The data must

satisfy the condition (4) The discretization is also in the

same staggered grids already described P and f3 are defined

in the centers of the cells f1 and f2 are defined in the

same points as U and V respectively (see figure 2b)

Figure 2b

V-V -v v

U P U P U P U PII

U P U P U P U P

I -V-u- I --V- 0 -- -V I- VVI_

The difference equations for equation (7) are given by

AucxtxlampiP (x+~y)-P(x- S Y)] f(xy) (10)

for (xy) where U is defined A is the usual five

points discrete Laplace operator Li -

Similarly for equation (8)

I CxY+ )-P(xy- ff2 (xy) (1i)(xxyY)- )

for (xy) where V is defined

For the continuity equation (9) the difference

equations were already defined by (6)

Equations (10) and (11) hold for interior points Near

the boundary the difference equations are different but they

are simple and will not be described here

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 12: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

10

The unknowns are U V and P U and V are given on

boundaries Like in Cauchy-Riemann systems The data must

satisfy the condition (4) The discretization is also in the

same staggered grids already described P and f3 are defined

in the centers of the cells f1 and f2 are defined in the

same points as U and V respectively (see figure 2b)

Figure 2b

V-V -v v

U P U P U P U PII

U P U P U P U P

I -V-u- I --V- 0 -- -V I- VVI_

The difference equations for equation (7) are given by

AucxtxlampiP (x+~y)-P(x- S Y)] f(xy) (10)

for (xy) where U is defined A is the usual five

points discrete Laplace operator Li -

Similarly for equation (8)

I CxY+ )-P(xy- ff2 (xy) (1i)(xxyY)- )

for (xy) where V is defined

For the continuity equation (9) the difference

equations were already defined by (6)

Equations (10) and (11) hold for interior points Near

the boundary the difference equations are different but they

are simple and will not be described here

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 13: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

11

The relaxations are based on the same principles already

described The relaxation of equations (10) and (11) the

Momentum equations are performed separately by the usual

Gauss-Seidel method corresponding to the Poisson operator

The relaxation scheme for the continuity equation (6) is

described in figure 4

Figure 4II

ampis chosen so that equation (6) is fully satisfied on the

current cell and the changes of the values of the P are

chosen in a way that preserves the residuals of equations

(10) and (11) when relaxing equation (6)

The theoretical smoothing factor in the relaxation

method just described can be shown to be I =5 like in

Poisson and Cauchy-Riemann equations

In the numerical experiment we get a multigrid convershy

gence factor - 65 This is a very good value and assures

This is exact only on unbounded domains On bounded domainsthis property cannot be strictly maintained near the boundary

This is the reduction factor of the residuals per unit workwhich is equivalent to one relaxation sweep on the finest grid

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 14: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

12

a very fast method for solving the Stokes equations However

it was a little worse than our expectations based on the

previous experience of Poisson and Cauchy-Riemann equations

In order to check this point and to reduce the possishy

bility of errors in the computer program we performed a

complete multigrid Mode Fourier Analysis (with the aid of the

techniques to be described later) The results of this analysis

showed complete agreement between numerical and theoretical

results leading to the conclusion that the value of 65

is intrinsic to this system which is characterized by strong

interaction between the equations The Mode Fourier Analysis

helped to rule out possible programing errors or bad influences

of the boundary on interior regions

An example of a computer output is given in the Appendix

Output No 2

23 Navier-Stokes Equations

Two-dimensional Navier-Stokes equations are given by

[3x(12)

LxR~~x1 1 Vtj~- 21)(13)

)X C (14)

and the same boundary condition as in Stokes equation The

parameter R is called the Reynolds number (R=O corresponds

to the Stokes equation)

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 15: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

13

These equations are more interesting from the practical

and physical point of view than the former equations The

numerical treatment is more difficult and complicated due

to several reasons like the nonlinearity and the fact that

boundary-layers may appear for large Reynolds numbers

These facts demand a careful choice of the difference

equations in order to keep the ellipticity of the difference

operator This can be accomplished in several ways based

usually on one-sided first differences instead of central

first differences that may cause instability unless we make

a very drastic and unpractical restriction on the mesh size h

1(h must be of the order h=O(K)) In this work we define

the following difference approximations to the first derivatives

of U and V for example for we define

x- 0-a0 definULe (15)

for 0ltaxpl (ax = 5 corresponds to central differences)

a is defined byx

_i )(16)

I-- 1

21 = LRU(xy)and

U gtVan - aedfndi The approximation for and - are defined in

a similar way It can be shown that with this approximations

we get finite differences with the desired properties The

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 16: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

14

equations defined on the same staggered grid already

described are given by

MK~ue1) - P F(tI (8

(19)

(20)

(21

-k~U lRk (22)

4 4shyc-+(23)

(These are the equations for interior points The operator

is a little different near boundaries)

The relaxation method for (18) and (19) are done as in the

Stokes equation by freezing U and V in the calculation of

(9(1) and N)

The relaxation scheme for equation (20) is described in

figure 5

ORM1INAL PAGE ISQ ooR QUALM

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 17: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

ORIGINAL PAGE IS OF POOR QUALITY

115

Figure 5

Vt C

is chosen so that the residual of the continuity equation

in the current cell is zero The changes in P wouid keep

the residuals on (18) and (19) unchanged if we froze the

coefficients

Mode Fourier Analysis shows that the smoothing

factor (calculated for equations with frozen coefficients)

will generally depend on the direction of the relaxation of the

Momentum equations For instance if the solutions U and V

satisfy U)O VG and the direction of the relaxation is as

usual (say by columns and increasing y and increasing x)

then 5 If U or V are negative in some part of the

domain and R is big enough the smoothing factor will be

higher and therefore the multigrid method less efficient

These difficulties may be easily overcome by making relaxations

sweeps in different directions improving in this way the

smoothing factor or better by the technique of distributed

relaxations already mentioned

The first preliminary version of the algorithm does not

include the features just pointed out and therefore it works

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 18: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

16

with the full multigrid efficiency only for cases where U and

V do not change sign in the domain or for general cases with

R not too big (up to R=100)

These first experiments are very important in showing

how the algorithms work in principle independent of secondary

problems like the direction of the relaxations mainly

technical

The numerical results shows a good multigrid convergence

factor bounded for all R by 7-75 This provides

a very fast algorithm for solving Navier-Stokes and

almost the same efficiency as in the linear Stokes case

The full multigrid cycle (explained later) was also

implemented Preliminary experiments shows that only

8-9 works units are needed to solve the equations

to the level of truncations errors (Execution time

on CDC 6400 for a grid of 64x64 is about 15 seconds

and the program is not optimized and includes all kind

of calculations for debugging purposes) An example

of a computer output is given in the Appendix Output

No 3

III CONTROL AND PREDICTION TECHNIQUESIN MULTIGRID METHODS

One of the remarkable advantages of the multigrid

method is the possibility of predicting in advance its

efficiency given the relaxation method interpolation

Equivalent to relaxations sweeps on the finest grid

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 19: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

17

etc This can be done by means of local Fourier Mode

Analysis in an infinite space ie neglecting boundaries

influence This can be justified because of the fact that

Fourier Analysis is very accurate in descibing the fast

components behavior while is less accurate for slower

components However most of the calculating work in the

multigrid processes is invested in the reduction of the

fast components of the errors in the finest grid and this

is achieved by means of relaxation More detailed justishy

fications can be found in Brandt 3 who shows a very good

agreement between theoretical results based on Fourier Mode

Analysis and between numerical data obtained by multigrid

algorithms

We can distinguish two levels of analysis that are

appropriate for multigrid methods In the simplest one

we calculate the smoothing factor 4 which depends only

on the relaxation process The more complete and complicated

analysis takes into account all the multigrid processes and

estimates theoretically the value of t (definitions of 7

and 4 are given in the footnotes in Chapter 2) In many

cases it is enough to perform the simpler analysis In - i i0

these cases the formula I = I wheref is the

ratio of the number of points between coarse approximate

and fine grid (usually S-) and d is the dimension of

the problem provides a fair approximation to

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 20: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

18

The calculatibn of the smoothing rate involves the

search for extremum of functions in bounded domains and the

ca-cu-lation of etgenvalues (defined -shysrally by a gheralshy

ized eigenvalue problem Ax=ABx) of complex matrices of

size q9 where 4 is the number of equations in the

system The estimation of the convergence factor

involves the same kinds of calculations but the size

of the matrices are much bigger ( 2 x 2 where d

is the dimension of the problem)

It is clear that the kind of calculations just

described cannot be performed in a closed form

with the exception of extremely simple cases like the

smoothing rate for Poisson equations and one dimension

simple problems

Because of this and other important reasons an

algorithm that performs these calculations was constructed

A general FORTRAN subroutine was written for this purpose

and the main inputs are the following

d - The dimension of the problem

q - The number of differential equations = the

number of unknowns

r - The number of relaxations in one multigrid

iterative cycle

In the relaxation process the Fourier components areindependent But when we take into account the transferfrom grid h to grid 2h we get interdependence between

groups of 2d components

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 21: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

19

AB - Complex matrices that are derived from the

particular difference equations and methods

of relaxations The size of these matrices

is A and theydepend on the Fourier

component defined by e=(SJampL )6A)

A A function of the Fourier component 0

describing the initial distribution of

amplitudes in the errors (For a random

initial distribution of errors A0 =i

for all amp)

L - Number of points defining a mesh in the ampdomain

S- The ratio between the number of points on grid

and the number of points on grid

The output contains mainly

a) The smoothing factor 4 defined by

4ncz-~(25)

where

d eE CA V- e0(26)

and JAIX1 b) The weighted smoothing factor j given

the number of relaxations r and the weight

function A amp

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 22: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

20

c) The location (mode) where the smoothing

factor 7 is found

d) A distribution map of 10) (only for the

two-dimensional cases d=2)

The same subroutine with some changes can be used

for the complete Fourier Multigrid Analysis In this case

the user has considerably more work in defining the complex

matrices A and B that include in this case the transfer

of residuals from grid h to grid 2h and the interpolation

from grid 2h to grid h

These programs are very important in the research of

multigrid methods it can be used for several purposes

for example

1) Checking new relaxation methods or new multigrid

processes before the algorithm is translated into

the computer

2) Comparison of several algorithms for the purpose

of optimization ie looking for the faster and

stable ones

3) Debugging of multigrid computer programs

We used the subroutine in several cases including

Cauchy-Riemann and Stokes equation We also used it for the

Poisson equation In all these cases we did the complete

Fourier multigrid analysis checking and comparing various

methods of relaxation residual weighting and interpolation

We got full agreement between theoretical and numerical

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 23: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

ORIGINAL PAGE IS OF POOR QUALITY

21

results (in all cases considered) and this fact increases

of course the reliability and prediction power of this

theory

An example of these facts are summarized on table 1 which

include theoretical and numerical results for Stokes equation

Table 1

Comparison Between Theoretical and Numerical Resultsfor Stokes Equation

(The number of Parameter related ultigrid Convergence Factor relaxations on to the degree of the finest grid accuracy in solving Theoretical multi- Numerical in one iterative coarse grid correctshy grid complete results multigrid cycle) ion equations Fourier analysis

1 4 661 6381 5 648 6342 3 680 6952 4 710 125 3 1 714 722

The subroutine was also applied for checking a special approach

to distributed relaxations that may perhaps be applied to

Navier-Stokes equations This will be shortly described in

Chapter 5 An example of a computer output is given in the

Appendix Output No 4

IV MULTIGRID METHODS FOR TIME DEPENDENT

PARABOLIC EQUATIONS

A possible quite obvious way to use multigrid procedures

for initial value problems is considered in the work of

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 24: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

22

Brandt3 By this procedure we use a multigrid algorithm

for solving the implicit equations usually defined at each

time step If for instance we want to solve the Heat

equation in two space dimensions then for each time step

an elliptic problem similar to the discrete Poisson equation

is defined (and in this case the Gauss-Seidel relaxation

has even better smoothing properties than in the Poisson

case) The typical amount of work needed in advancing each

time step by multigrid procedures will be accordingly

equivalent to 5-6 relaxations (see 3) if a solution for

M time steps is required then the total amount of work

will be about 6M relaxations

The question which naturally arise is whether we can

use multigrid principles anI ideas to get more efficient

methods for these Problems

The answer appears to be positive and Professor A

Brandt pointed on some apnroaches which eventually can develop

in efficient multigrid algorithms for these problems

A basic idea is that marching in time can be done for

most of the time steps on coarser grids

The appropriate equations on a coarse grid are carefully

corrected in a way that assures the correct representation

of the information from the fine grid so that even when

marching on coarser grid the accuracy of the finer grid is

kept To update the information from the finer grid we need

some sporadic and infrequent time steps on the finest grid

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 25: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

ORIGINAL PAGE IS OF POOR QUALITY

23

that constitute of course most of the computational work

The base of this approach relies on the fact that fast

Fourier components of the solution (represented on finer

grids) converge to steady-state after a very short time

Changes in the solution after this are due mainly to slower

components that also change slower in time This allows

us to march on coarser grids with large time steps after

the influence of fast components have disappeared

In order to check these and other ideas we developed

two different algorithms for the Heat equation in a

rectangular domain with Dirichlet Boundary Conditions The

first algorithm uses the Crank-Nicholson implicit scheme

and the second one uses the simplest explicit scheme It

must be pointed out that the stability restriction in the

explicit scheme that makes this method very unpractical

does not appear in this case because most of the time we

march on very coarse grids where large time steps are allowed

a The Implicit Scheme

The equations to be solved on the finest grid

(with mesh size h0 and time step k0 ) are

J tL+Pe X (27)

and the unknowns are u0(xyt+k0) for each (xy) definedId

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 26: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

24

the finest grid Similarly umhmkm will represent marching

on a grid which is m levels coarser The equations on coarser

grids are given by

Ut ~ ~ A ~ ~ shy S1L~~r

L(XA 7) (23)

-where usually and T represents the appropriate

correction -C has a very important significance and it

represents the spacial truncation error of grid m relative

to grid 0 TU is given by

A s (29)

where n s) satisfying S C

The transfer from a given coarse grid m to a finer grid

m-i is done by rn-i rn-i in-m rn-i

uULast + m u -u Last) (29a)

where lmm I means interpolation (cubic) from coarse m

to fine grid and Last renresent the last marching in

time on the m-i grid

b The Explicit Scheme

The equations on the finer grid are

jO Lt t )-t1l t - ixc y )= f~ xy) (30)

(

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 27: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

ORIGINAL PAGE IS 25OF POOR QUALITY

On coarser grids the following equations are defined

t~ (3 1)

In this algorithm we keep a constant ratio A over

all grids given by A 1 The value of A 2

was chosen as very appropriate from the point of

view of the fast Fourier components

represents

here the complete relative truncation error (in

space and time) ishyis defined as follows

For m=l

T S) L4txL sxt xy sV fx ) (32)

and (t4 Then for general m

j 1 k- (32a)

and s is defined as in (29) The transfer from coarse

to finer grids is performed like in (29a)

It must be pointed out that this research which deals

with new approaches in implementing multigrid ideas has

a very preliminary character and the principal aim was

to check potentially the various promising possibilities

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 28: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

26

In this context some numerical experiments were performed

in which we tried some criteria in order to define the

times where we switch from coarse to finer grids and viceshy

versa In addition we compared the accuracy we got in these

algorithms with the same schemes when we marched in time on

the finest grid only The numerical results we got from

these experiments in both algorithms are good and encouraging

However the explicit scheme appears to be preferable in some

aspects The numerical results in these experiments and some

theoretical considerations arising from the interpretation

of these results lead to the conclusion that applying these

multigrid ideas with full efficiency (ie solving the problem

in an amount of work comparable with the work invested in

solving elliptic problems) means using adaptive techniques

where we can change and adapt the order of the discrete

approximation

But even without adaptive techniques we can use the

present algorithms to solve the Heat equation very efficiently

The amount of work needed will depend on the smoothness of the

solution For instance for a problem with smooth initial

conditions we can save 98 of the work in relation to the

same numerical scheme defined on the finest grid only (of

size 64x64)

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 29: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

27

V IMPROVEMENTS AND CHANGES IN EXISTINGMULTIGRID ALGORITHMS

In this chapter we deal with existing multigrid proceshy

dures and we check various approaches in order to improve

their efficiency The first part deals with the practical

solution of Poissons equation in a small number of numershy

ical operations Although in this part we concentrate on

Poissons equation for reason of convenience when checking

and comparing numerical results it is quite clear that the

techniques developed here can be applied to other elliptic

problems as well

The second part of this chapter deals with a special

approach to distributed relaxations for general five points

difference operators of the form la -s c s = a+b+c+db

51 The Poisson Equation

The first numerical multigird algorithm and program

was written for the Poisson equation For this equation

there exists today more information related to multigrid

methods than for any other problem

In Brandt I the algorithm called there Cycle C

for solving the Poisson equation is described with detail

In Cycle C the multigrid iterative cycle starts at the

finest grid where an initial approximation to the solution

is defined This approximation is then improved by solving

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 30: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

--

28

correction equations on coarser grids The switching from

fine to coarse grid and vice versa is controlled by some

internal criterplusmna

Cycle C is very useful in learning and understanding

multigrid performance and several theoretical aspects

like the asymptotic convergence factor etc It is

not generally the most efficient algorithm for solving

real problems for which we do not know in advance a

good approximation on the finest grid In addition it

is usually difficult to know in advance how many iterative

cycles are needed to get the desired accuracy One can

perform several iterative cycles and then get very close

to the solution of the difference equations but this will

generally be wasteful because in a real problem we do

not need more accuracy than the accuracy defined by the

differential truncation error

Because of these and other considerations a fixed

algorithm for Poisson equation is described in Brandt3

In this algorithm we start with an approximation on the

coarsest grid and after we perform one iterative cycle

on each level the solution on each grid is used as a first

approximation on the next finer grid by means of a cubic

interpolation This algorithm is based on the knowledge

and past experience in solving Poisson equation by multishy

grid techniques and provides a solution to the problem

up to an accuracy comparable with the truncation errors

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 31: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

ORIGINAL PAGE IS OF POOR QUALITY 29

accuracy and this solution is found in a minimal number

of arithmetic operations (we do not even need to calculate

the residual norms used generally-forinternal criteria)

In the present work a similar but more general algorithm

is constructed This algorithm can serve for learning

purposes as well as for solving practical problems It can

also be easily applied to other elliptic operators The

algorithm was implemented in the FAS (Full Approximation

Storage) mode (For a detailed description of FAS see

Brandt2) In this mode we can look at the coarse grid

correction equations in a somehow differentpbut equivalent

way as follows2k z U

Tj~(~ Ljuk(33)

U2h is a new and better approximation than Uh to theI

exact solution Mh of the difference equation

L (34)

Zk T given by (33) describes the truncation error of the

grid 2h relative to grid h It is well known that the exact

solution u of the differential equation Lu-f satisfy the

following

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 32: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

30

L2 hU = f+ 2h (35)

and T2h is the differentia- truncation error of the 2h

grid The comparison between (35) and (33) that arises

naturally during the FAS Multigrid Cycle without additonal

investment of computational work shows the remarkable

differential aspect of the multigrid method This is in

sharp contrast with other methods where the algebraic

aspect is usually the most important and central one

These properities open new and interesting possibilities

in implementing the FAS algorihm z2h in (33) satisfiesh

the following

L (36)

and the truncation errors can be represented iby Tkz kkUdL

(provided additional smoothing conditions) Then we have

cVk +()(7

Then we can change the FAS algorithm by defining the

following correction equation on a coarse grid (instead

of 33)

2k 3 (38)

ORIGINAL PAGE IS

OF poop QUALITY

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 33: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

31

A series of numerical experiments were performed to

check these interesting points including printouts of

and I U -ii(we chose for this purpose problems where

the exact differential solution u is known) As a result

of these experiments we can point out the following

conclusions

a) The assumption that after only one multigrid cycle

(5-6 work units) we get the needed accuracy in the

solution (truncation error accuracy) was confirmed

In addition truncation error behaves clearly like

O(h 2 )

b) We can get the same desired accuracy without even

performing a complete multigrid cycle ie we can

stop the process before we interpolate back to finer

grids The solution is then defined on coarse grid

points only but the accuracy is the finest grid

accuracy This feature can be very helpful in

the development of multigrid methods that use small

amount of storage much smaller than the number of

unknowns in the finest grid

c) Using thd T-extrapolation (38 instead of 33) we can

get a much better approximation to the solution of

the differential equation on the finest grid much

better in fact than the approximation achieved by

ot 06

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 34: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

32

the exact solution of the difference equations

in this grid where the accuracy is bounded by

truncation errors In order to get this better

approximation we do not have to invest any addishy

tional work Our numerical results do not show

however that the approximation is 0(h4) in

spite of (37) This is because of the fact that

at least 5-th order interpolation is needed for

this purpose

Nevertheless the improvement we can get by this

small change in the FAS algorithm is impressive

The improvement will generally depend on the

smoothness of the solution The more smoothness

the better approximations

In table 2 we cancompare the results we get in the

different variations we tried The problem is Au = f

the solution is known given by u = sin (3x+3y) (04x43 Osy$2)

Table 2 LII i- tL

Size of the Experiment Experiment Experiment ExperimentFinest Grid 1 2 3 4

9 ) 16 x 24 14x10 2 (23) l8xl0 2 (59) 53x10 - 3(5 23x10 3 (91)

32 x 48 34xi0 (23) 42x0 -3 (56) 47xi0 -4(55) 23x0 -4 (89)

64 x 96 86x10 4 (23) l0xl0 3 (54) 52x105 (54) 18x0-5 (89)

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 35: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

ORIGINAL PAGE IS OF POOR QUALITY 33

Key to table 2

Experiment 1 shy is the exact solution of the difference

equations (up to randoff errors)

Experiment 2 - U is the approximation after only one cycle

in all grids

Experiment 3 - L is the approximation after only one cycle

in all grids but using (38) instead of (33)

Experiment 4 - Like experiment 3 but the number of relaxation

in each level was doubled

The numbers in parenthesis represent the number of work

units which are equivalent to the total number of relaxations

on the finest grid (printed in each line of the table)

52 Distributed Relaxation

One of the important application of the method of distributed

relaxations is in finding relaxation schemes with good smoothing

properties This is generally hard for very asymmetric operators

In these cases the usual Gauss-Seidel relaxation may be very good

or very bad depending on the direction in which the relaxation

is performed The problem is even more complicated in case of

non-linear difference operators because in this case we may need

one direction for some parts of the domain and an opposite

direction for other parts In these cases we can perform

as we explained in a previous chapter a series of altershy

nating direction relaxation sweeps but this cannot generally

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 36: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

34

be an ideal solution If we think for example of a threeshy

3dimensional problem we need there at least 2 =8 relaxations

O a multigrid iterative cycle indrder to doOn eampch level for

relaxations in all possible directions This may be too much

because we know that in most multigrid procedures we use

only 2-3 relaxations for achieving optimal multigrid efficiency

So alternating rirection relaxations may in some cases reduce

considerably the efficiency and in addition it will cause

more programming difficulties

Because of this and other reasons it is convenient to

find relaxations with good smoothing properties that are

to some extent independent on the direction of the relaxation

so we can relax always in the same direction no matter how

the operator behaves

As a first example we consider the one-dimensional

operator represented by Ei I - 7- Iti C C

which can be derived for example from the differential

equation 5-R1 (by taking one-sided first differences)

The smoothing factor can be easily calculated in this case

and for extreme values of it behaves in the following way

1 Relaxation from left to right

(39)

Relaxation from right to left

for

In the present work we check a special principle for

the construction of distributed relaxations The relaxation

4

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 37: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

35

is always-done from left to right according to the scheme

represented in figure 6

Figure 6

Relaxation

LX

and in figure 6 represent the values of the

approximation before relaxing on point i and the new values

after the relaxation on this point are represented by k

J is chosen so that the residual at point i is

zero The changes performed at points i and i-i make the

residuals at the points i-i and i-2 nonzero although they

were previously zero when we relaxed these points So we

can choose the parameter lt so that the deterioration of

these residuals is minimal say in the L2 norm

If we chose A in this way it can be shown that the

relaxation scheme is always stable and satisfies tV

for all 0 For P7 0 The optimum parameter

t depends of course on but satisfies 4l iamp 5

and in fact we can chose a constant suboptimal m say

= 45 without significant lose of efficiency

It must be pointed out that this scheme is not

necessarily the best one for all values of M In fact

there exists better stable schemes (Brandt7)for large

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 38: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

36

Using the same principles just stated we calculated

distributed relaxation schemes for general five points

operators in the form

d

a -s c

b

where abcd O and a+b+c+d=s

Operators of this type are typical for instance

in Navier-Stokes problem The relaxation scheme is

described in figure 7

Figure 7

Direction of the relaxation

oL and fare the solutions of an appropriate optmization

problem As in the one-dimensional case A+ A+-to JA+fA7

represent the new values after the relaxation of the central

point of figure 7

The results (calculated by the algorithm described in

Chapter 3) show that the smoothing factor is bounded far

below 1 even in the cases of great asymmetry of the operator

except in the cases where the problem degenerate practically

to a one-dimensional problem (like the case a- cz 0)

In the case of Poissons equation (a=b=c=d=l)4 =315 which

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 39: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

ORIGINAL PAGE IS OF POOR QUALITY

37

is significantly better than in the normal Gauss-Seidel

relaxation (1=5) For this Poisson operation we get ri

For other nonsymmetrical operators the comparison between

this distributed relaxation and the usual Gauss-Seidel is

even more remarkable as we can see on table 3

Table 3

Operator Smoothing Factor 4 a b c d Gauss-Seidel Distributed

Relaxations Relaxations

5 5 631 J 313

105 5 105 5 913 449

5 1005 1003 5 990 481

5 10005 10005 5 999 484

5 5 10005 10005 1000 752

Like in the one-dimensional case it may be possible to find

more efficient scheme for the extremely asymmetric cases But

the present methods assures a good smoothing factor in all the

cases bounded far below 1 and consequently an efficient multishy

grid algorithm Moreover if the problem is nonlinear we do

not have to know in advance the behavior of the solution in

order to define an appropriate relaxation scheme In addition

This distributed relaxation scheme for Poissons equationwas implemented in a multigrid computer program and we getperfect agreement with the theoretical smoothing factor

= 315

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 40: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

38

as in the one-dimensional case it is not necessary to invest

extra work in the exact determination of the parameters and

They can be estimated in a suboptimal way without significant

loss of efficiency

It must be pointed out that in the case of the Poisson

equation the Gauss-Seidel relaxation can be still considered

a little more efficient if we take into account the number of

arithmetic operations needed for each relaxation method On

the other hand for all other operators even if they are

small perturbations of the Laplace operator the proposed

distributed relaxations are much more efficient than the

Gauss-Seidel re1axation

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 41: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

39

REFERENCES

1 A Brandt Multi-Level Adaptive Techniques (MLAT) forFast Numerical Solution to Boundary-Value ProblemsProc 3rd International Conference on NumericalMethods in Fluid Mechanics (Paris 1972) LectureNotes in Physics Vol 18 Springer-Verlag Berlinand New York 1973 pp 82-89

2 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Research Report RC-6159 IBM Thomas JWatson Research Center NY 1976

3 A Brandt Multi-Level Adaptive Solutions to Boundary-Value Problems Mathematics of Computation 13 1977

4 Y Shiftan Multigrid Method for Solving EllipticDifference Equations M Sc Thesis Weizmann Instituteof Science Rehovot Israel 1972 (Hebrew)

5 H Lomax E Martin Fast Direct Numerical Solution ofthe Nonhomogeneous Cauchy-Riemann Equations Journal ofComputational Physics 15 1974

6 M Ghil and R Balgovind A Fast Cauchy-Riemann Solverwith Nonlinear Applications Research Report CourantInstitute of Mathematical Sciences New York University1978

ORIGINAL PAGE IQ OF POOR QtALIT

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 42: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

APPENDIX

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 43: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

OUTPUT NO 1

CAUCHY R EHAN P2) UX6VYCSo(I) UY-VXFINJECTION SwO F=XD24Y2 StrMAtG gELx ATtOzf COARSEST GRID I5 - 2 X-INTERVALStANO 2 Y-INTERVALS HO 500 NUNRER OF GRIDS 6 LtvEf 6

METHOD PARAMETERS ETAu 600 DELTAS 200 ERWi 5000 - - TOLm 1O OE-04

LEVEL RESNORN(I) - RESNORM(I ) WelGIlre tsWdo WORK (c)0 --shy -shy

6 4T3E400 311E-i2 oTSE-01 1000 - _ - - - -__-shy6 449Eo0 A1TE-12 T48E-Ol 2000 - - - - - - - - shy5 413E00 460E-12 688E-011 22505 377E00 S17E-12 628E-01 2500 4 3E E600 493E-12 - 534E-01 2563 4 9266E00 496E-12 443E-01 26253 193E600 9432E-12 322E-01 26413 134Eoo 409E-12 224E-01 2656 shy

z 9806E-01 331IE-12 l34E-0 2660 2 464E-O1 29TE-12 7T3E-02 2664 -2 4264E-01 270E-12 441E-02 2668 -

3 280E-01 6952E-02 126E-01 2684 shy0 554E-02 2699

4 36TE-01 -15sE-01 1192E-01 2762 3 183E-01 299E-02

S206E-01 592E-02 836E-02 2R24-3074 5 456E-01 1OE-Ol 2QlE-0 1

5 6264E-01 - 727E02-I1OSE-01 3324 shy6 488E-01 124E01 184E-01 4324

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 44: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

6 316E-1 -shy 784E--shy 118E-01 - 5324 5 242E-01 747E02 103E-01 5s74shy 5 187E-01 729E-02 919E-02 - 5824 4 148E-01_ 661E-02 -shy - 797E-02 587 4 116E-01 - 608E-02 - 99E-02 - - 5949 3 852E-02 459E-02 525E-02 5965 -

3 9608E-02 343E-02 387E-02 5980 2 - 343E-02 IS9E-02 190E-02 - 59 - -shy2 3

ol40E-02 1OOE02

556E-03 - 109E-02

696E-03 1OTE-02

5988 6004--shy - -

shy

4 195E-02 145E02 153E-02 6066 5 6 6 -

395E-02 775E-02 52TE-02 -

__

160E-02 -187E-02 i OE-02

- -199E-02 285E-02 204E-02

6316 7316 8316

5 391E-02 125E02 170E-02 8566 - 5 -----shy 276 E-02 shy - - 119E02 _ 145E-02 - e16 4 20AE-02 - O1E-02 118E-02 8879 4 - 144E-02 9932E-03 - m102E-02 8941 -

3 -

3-764E-03 107E-02

-622E03

715E-03 773E-03 646E-03 -

8957shy 973---------

-shy--shy - - ----shy2 2

- 502E-03 9276E-03

415E-03 307E03

j429E-03 302E-03

8977 _8 PO

- ---Oshyl

1 - 949E-04 o-10-5-E-03 - d103E-03 8981 ---

I 142E-13 p116E-12 990E-13 8982 -shy - - - - - _ 2 - 774E-04 o113E-03 - 107E-03 8986 -

3 -12CE-03 o197E-03 il84E-03 9002 4 - 297E-03 247E-03 255-03 990-64 - 5 686E03 - 289E-03 355E-03 -shy9311 __-shy

6 9139E-02 - 320E-03 498E-03 10 314 --shy -6 - - - 920E-03 207E-03 325E-03 - - 11314--- 5 696E-03 174E-03 261E-03 11 5 502E-03 158E-03 216E-03 11814 4 4

-

377E-03 267E-03

128E-03 E108E03135E-03

170E-03 - - 1187T 11939shy

3 190E-03 -745E-04 shy 937E-04 - 1 955 shy3 q124E-03 510E-04 633E-04 1971 2 76SE-04 - 279E-04 -361E-04 11975 2 423E-04 156E-04 200E-04 1979 2 257E-04 - 105E-04 9130E-04 11982 1 -751E-05 1 - Al42E-13 _

-358E-05

- 108E-12 424E-05 927E-13

1983-_ 11984 -shy - - - --

2 651E-05 -444E-05shy - o479E-05 11988 3 20E200-04 181E-04 a191E-04 12004 4 613E-04 331E-04 378E-04 12066-------- 5 -129E-03 ---- 443E-04 - 585E-04 - 12316shy 6 6 5

1249E-03

T170E-03 130E-03

_ o509E-04

- 320E-04 0270E-04

1839E-04 550E-04 442E-04

13316 14316 14566

5 940E-04 249E004 9364E-04 34816

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 45: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

4 4 3 3 2 2 1 I142E-13 2 3 414E-05

4 5 6 6

698E-04 479E-04 329E-04 203E-04 s 126E-04

0676E-05 296E05

183E-05

sI12E-04 243E-04 463E-04 315E-04

o2l4E-04 198E-04 156E-04 i136E-04 o876E05 598E-05 o173E-05

1181OE-12 230E-05 424E-05 601E-05 9742E 05 832E-05 556E-05

295E-04 244E-04 9185E-04 147E-04 9940E-05 611E-05 o194E-05

_ 100E-12 222E-05

-422E-05 o687E-05 102E-04 147E-04 988E-05

CYCLE NO 2 CYCLE EFF - - CYCLE NO 3 - CYCLE EFF

CYCLE NO 4 CYCLE EFF CYCLE NO 5 CYCLE EFF

rog AcLt

14e79 14941 14957 1497314977 14980 1498114962 14986 1500215064 15314 16314 17314 574 56 42 q93

CV(FcentrActOI

-

-

- -shy

- - -

ACCoEFF 574- ACCEFFamp565

ACCEFF 558 ACCEFF 557

2 l- - --- - shy

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 46: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

__

OUTPUT NO 2 Yd

jE irjnJs - _ srn s 1t 3 (2 DEL(V)ZPYG (t)--Y lit1rI~II V~ F- 43Y) G $

IjIM111 VAU S- y +y(1y23) PC)

- 400- FqFutAT5 DELAY[ 1IEAU IqTIEr SChEE qES 4EIGTIrm EVENYiiHEPE -G [Oj)IS K- u4ERVLSACOASLST III 2 Y-1 -TEDVAL5 J 5tj

E_ TA IFLTamp i OLc _ I l Elh-

2LT9P1SANA _ET 1011 211)

EPFL3 5 _ 55 3 Q 0 - - L e VEL -J 6 tu( 61 - - - - - - - - - -

cc _ -- - ioiM amp4 AEFS$ M shyf XoR

LFVFI W(I-lt (WE IC$TEDI 41 VL 1t E 3 -shy l e l~l 1 51F 0 1 - I Ij flj p l1 il+ 1 2 E+ 11(

F~J shy-- 1F4 J

_ S - no - IIFI - l7Ftn I3amphl LI CYCLE l

- I3F ---ln -I -shy - C E-l------ -IE-I S 251 - 0 W7AJ7lI IP3FfaIl OQE+III A14pF4II

- a_ - b __13A f I I1 I I - - S ECI SIU l11 __ __ _ _ _ _ ___ __ __ __ _ __ _ __ __ _ 51 -1 I Fl +111 +1pl E i ~ 14I Fflll~t

__ - h ) flq 4-un 1VE1P- EIIEtii

I llI I 2 dF4 tI 90E4 011 - qE I shy

-shy hA -- E -+-IliuI J - L2hS$ l3 il II7E+II - _ - 766 sr-ol -- lidB -lI 1)El5pq -- - +r shy

- -E -I A 1 F - - A I f 1 - - - - - -n - shy

- to I Ef hQ Q $OE (I -- - shy

--shy t( Pl3( FQ FIII 110 02F

15_ 1 IVnE4+110 I EIf I -n I E4 qli5 3 IE4 l IIEI R_E~f l FSi Ell 110A ktE4Ii

l --------- S3 A ---- - - - - - - - - - - - - -- ---- - - --- - - - - - - -E- - C Y C L E - -- N ~ E --- -

U-

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 47: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

-~ ~~ ~A~7 F +Icw7c A - 11n-F+ PiI)hf~ Izzz-----------nshy -shy - -shy -shy -

a 5 9tr 9

- ___

A0 Eshy 1o) 7E-nl 4J IE O1

5F-1 t 7 E-)IAIFP- Il

77SF-l _

-shy_4f)

It7 + -II -

-)-11

-shy---

--

Q shy- - tql_ |rfl-oI

L| QL tE-I 2 -

9qt

P___hOhf001 (IMEt 3 I) 11P -

2

- - - )0 F-112 3I F-O I ) 3r0- - olIJ~f)I

- thE -- IE-O 331)E-13

11rE-O -n 7 0 -03 _shy -fl -shy SR-q

A711E IP I $RE-III

- 17F--lj - 1171F1(p

- 7pE-0 I I1shy-f-Qi7AF-iS

-shy F lP 77P - I

-

-

-

--shy

-

-

--shy

-shy

-

-

_

S- 5 ___43 4 3Aq 9 P97df

-

- seeuroF- r MEnI 67E-nII|

-I-I - IIEn) 9SE-flI j3- En I7J -11I- -

454(t -shy -

SE-dnl-shy Fl-F I

- -shy ------

-FNn CYCLE MII

_ 5 R _ __--shy SnwFI __ 4shy A 1n11 PIE-1 t -

Al71E l P( __ AP -I tm I

P 0 9 4 -P-_Ppr-lj__ _ | 149Q _ 1 1 -3

___3 qK12ff1 YSE- _shy

PE-Ol 1s7E-nRhPE --) 9QE-O 61 f - TtIi - )

p

6-1 E-02 7PEshy

p I - Q7r -I

tE- I3 l F-iVI it I PE I7P0Ia -shy E3

9rn 1-)2 IAF-F-)

luV o-n IphEI I

tbcoF-112

- r

1 1 -I _ M -

-- - --shy

-

-

-shy

--shy

-shy

-

- -

I _17 I 1 x - 110Fshyf _SEO 6LF II FI uI --shy

____ _ _I (shyMi3(l

I97(I-OAA -I

- tEi3 77fr~hRV

- -shy lI7PE-l t2

o -7It-I

PcF-n

S7~II_f3 u-11

Q

1117E-02a

__- 1

_ 1

A15

- - 411I _ 1 I

CtE

AP

77 E-01 tl I Zll

74I-O1

- 1 iFflt 5q 3E-113

310-o

-

QAF (3p HF b

1~

71 7 - P - 3E n 7-

-

A

J01I

123il

I-J

14 -13-

-shy

-IO -shy

~Q3SI)11

-

7711E-c3 1qfl-n VO-03p

-

-

j4-EM -

l-_h 13 31 __

1033 V-IO-flP

aou1F-oe - 1 F-OAj 3MlF-ii

II~Q14PEeB2

0Et1

-

__

5 __

-1

it_ --shy

1-

~ shy4 3S

1I 11~75 431

3 uS- 14

-

-shy

9hlE-13 AhfF-v3 1AfE-n3

(E O G5E-03

b11 q)F6113

-

-

-

t 77E-03 3POE-11

P1-PE-flIS 1 71I1E-05 01 6EO04

-A1

-

-

4AE-i3

5U7Ed-3

3$AEE-a

P65E-113frII

1114E-03 At~IIE-fI4

-

-

- IMq -1a

1 7P)F-1)

Ij

R-1 I bh-f-3

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 48: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

--II~~~~~j _ J$0 7t

IAQ 32~lnstzr--__1t--- -9

11197- 2$AE- Afl__50i ~oq __e113too - 33r-0i-----j1___ 35tE-5 $7A -115 1Q4JE-115 531fl-n5l

_ 2 pound_JoI _ 17VE-QI 4tlE- g - IP5E-1O _ 3 ___ 1 140 -7 523E-114 -- - 4 l rSE-I 4711p1E-04 h6F-nf

_ 15160 1E3 -I3E-3 _ --- 151 E-11 3 ld2E-03 5 3 I 322E-A3 21 GE-03 3AE-113 3oE-n3

I__ I) _ 7SE-n - 7ubE-I3 shy7111E-113 111E- _ _ -- I731 1177E-n3 - 21 E-13 3h 1 025F-13

----- -_---- ENO CYCLE O6 S__ 7 67Q I EAE-0l3 10 1F-(03 - th6E-13 l 1 -IIi3 -II IA_ __ PIQ(E- 1 _ l l h~i -ll I )AE-13 t q)E l - shy

- _ -_ ] 777-a - 51E- t sIF-Ilu 17 IF-0S 11 17 9$75 Q E tE - 37 0E 0E- -- --O 1

3 17QS 33PE- - INIF-14 $ PQAE-(0 50E-l0J --- 1 17 1)A IlE-0I - 11IE-IlII t lIF I)J 39 F-1IIshy

_ I 7u7fl 5eQE-15 372E-05 7hE-S qQ7E-I -shy

74 _ 3- 7 E- 15 I U15E- SI AI F-n r - N5IF-IIb3 I77 q q I i I UEIlU I PAFIll - I59 -EIII

W39 I - _ I s 3SF - d Jp-nj

E-14 l3-d i - IM3A II - 19 fqJ-Cu6 - lbI-

1 IJ bbl E-U t U E -I -

A I93nAP IP 3E-1 94)IF-011 I72E-03 -_P jF-n - _n3 fl 12IF-(3 QF -i I I ftAE6 - I P AF-03 ----

------- -- - - - -- ---------- -- - -- ---------------------------- CYCLE Lf _ - 20( 2i II-_ -- 5SE-( i - 4 9I-01 1 IjIIL-il -

S__ 21 n0 - - 150 EfJ -) - nf -im - -shy- 31o~~l - gp F-( shy

(l - - A 11 - PIIF tIQ I 7P-11 a IqUE-Ild 11P E- Itl - --I 9) 7 __I_ r- - Q thE-0s5 IF-)I 71F- i -

--- I _ )fn91 A 3hF 0c 44 E-nS 7 173F-ilr 11F-I)

__ 1 E5 - l-501amp-05 q F- C - - 6k-)1 -- E shy

-- __ PiIUAS I| tE( bull 7F-nS _ 07E-(9 -IE-()

__ P)- 7 P aF-I hb lIE-b7 - 301 F-qh - - -

-- 2flAP _ 1t -b0h 133E-n6 IIE-ob t JE-07 - P - I q7) M hEIN I P3F-IS - 6Q -0ASF VF-l 116

P 21 012 - j7)E) - 1()5 39E-115 77 E-Pt11

-06 shy

S- F- 5 1E-I5 - RQ1U-s 4AI

- q7 oR-o p lE-p 1W 1 - --- shy-hq2E-n5 JlE-IoS

9 3Ii(l V(IU II I6IF- lJ - - -1PPF-nq -oa P2 14APF1114 - - - F 114 oc(j5F (l

__~ ~A P3 30) iflOE-rll I nIIE-All - AEf5ARE41S -- FN CY NJ) shy ~~ ~ I ___--- I)5IpE-nfj I 01-tS|ltF114 IIhF-tIII

____ P1-55 __IEil tlE-05 Ir-al -nl ---- -shy

_ S 23$1) IIOF-o0 l fE-i$ 77E- lS - P 1F-O0I _ ___ 2 bl 54e-li -f 302F-Ob 0 JtnFIl5 7TSF-ilc -

- L -45_ 1i17F-05 P3E-nS 3 I F-9 kP7F-n shy

_ 21odI 76E-n5 1 lIE-O5 ISPE-oS PM I F-fl5 shy

3 PPQsh II4 -centA 7PE-I~b I021-15 I 7A- - -_ ___ Pt hn 37 lE-ilh puF-Ot 377E-)h - WAUE-06

-- P1o 60 - 17F-nh17PFE-h degI 3F-A IhAF-Oh

1I 2 Q9E-QS _hIEQ5 2- plusmn) P7E-iPt

Ul

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 49: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

26uP p RIE-(I -F-15 I II -oi 11 7F-1141 )_- ------- - --------- -- - - - ------- ------------- CYCLE M- - shyENDshy

5 a 2 I F-Ob - S--I)q - - YtCLE P9FOF _ _P - A 17__ c55 -L I I Fq-i-S 7E-05 5 shy7shy 1

-- -I

-

_ I -bS4 -

2ho17 933

s~~lq l|dF-iiS u1E53

- 71E-IA - 4t-E-16

E-(Ih -

III1 IF-415 7 4 (F-)6_

33E-16

PPF-1 I- r-n5

h 87Fshy 3 t9R _P 7rE- I)h -uaIP- Io- --2F--hb F - - - - - - - - - --- 2 __ PbR I 4E-117 bt3E-A 7 3QJP -t07 1 I 0F-06h

_3 I

~2h ~b ~lQ~rI$7 ____

FAJVtUQP-1A -shy -7 u _ __ pound 6F-flb __

53-f30E~-tp7 EI h-S 6 - b --S _ n -

+

I rE-4 ER -~

rSIF-117 IYF-l fliE-Ie

-

-

- shy

- 784d _ qa iE-ii - -shy i6I4E-I - nbhFnA I PhF-I9

___

___ _ 6 Pi PAORJ

3)-I7E-05 211 E-rIS_ -

-1j-1307V-ol

Ok~tF-I qshyE-hIIF-tI4

~ ~ -

-

E -

___ S - 7a - __ r -- shy---------shy - -shy shy

-- c - 24A1J7 3 17E-16____ 171 E-oIb pbRE-l6 5IUF-0t - - __ __I _ _ 16E-16 QF-06 iA ___PSOE-p - 1 1 F-0A6 ---- 3 P S __ lF-b _ 63E-nl 0 F-0l7 1 7tE-1

_ 29 0111 71 t E-17 - IItE-O7 h Eh-07 I17-F-00shy 2 __QUI Fh-17 I7-3F-07 I I1UE-to7 - 1 1E-

_ 1J A IPE-17 I r-7 QbbF-119 I E-(7 - -

3 9 1 1 IE- 1lE-l) l I 4E-7 71IF-t1l 0 t07_102 I- h -- 5E -17 IP+- -- It7F-OAI

- 1 2 - 3 1FIS_+( I S5r 9F-0 3IF-lA ---- P - 0 h

_ 6_ 1 al7 Q-I 3OE-Oh I 5 I 5 Il rfF- )7Fshy---- p I rlmS~F0Ib 1 orr-n-Th 05

9~ -it-0O5F rta

- I Ad -ih lME-6 75-nh--------------- 13E-dt 1shy- _ _ _ 3 776 1S97E 0 A E-n7 - I IIE- 6 3l tl--

- 7 3 -) - ( - 9 -- jI~ 7~4o~il - - 11P I r-117 ( 1371j-ilt--------shyM7E1)7 I

1__ __I3 R 27I 1E7Ia I k I - oS_ F-7 _-3I_

2 )Il 23 31

3 3 A+tIr-f77E-7

_h bEI) A y - 37 E-n7~tI7gE-1l7

471E-118 aA(E717rE-+77cE-07 50O -ON0shy

1F- sr-n9E

--

-

3 __37_E 37E-2 3)6E-AA--------IO7 E - - 3 _ 845-_ 7U E-I)1t_ 1- 9E-0Q 7 I3 F-)7

____ ____

3301 l$4 -Ee7 31 Q 5 7E 7 -

15deg90 2 1

jqcF-f7 7 3E- 117

IC_ 6 l b1

32E-07 1 t -7-shy306E-OA

F-17 f711 E- 7 2F 7F-0ks

-

- -

- - - - 35V19 -shy tAA-11H OcSE-O7 lb IF-Oh 2- - - - - - - - - - - - -shy

~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~t -~4EQCYCLE NOQ IZ -

~~~_c9 E-O7 2-71E-nt - - I9fl -07 - 471E-117 - -shy

1 0 -E-07 17rE117 - 2A2E-117 IRU3-01 -shy -shy -

____ _shy 3$131 __ IqqF117 1 j ufF-f7 I6JE-07 3JRF-1i7 - - - -

359 thfE-u7 TbIE-PR 11 71)7l SF -0j 3 qq 17- R E- cB 2IEDampshy - - 4rIshyh 1 1dA I 1F9

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 50: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

V 5 - -52s E r A 1-JE _ M E i I6E 2 3 I 6I E-15217E E-I11 IIIE-AH E -fl5

A 31 _ 01E1t7bEQ - -P - _35_ I - F -- I _ 354L 5EtJE-07 b aA - 7 - E7-Oamp |E-I)L

L 67 E E- - I8FE- E-I 7 ----- 76 7 -- -shy5 - - - 1 - 37E- 2E - 07 2 113 L17 -- -- -- 4

37a4btI- 7S J 04dE-117 MIh E-97 9AAE-117 -shy

- h - 7E-Q7 2I7E-n7 1SdE-h7 -- 5Fn

SCYCLE NO 2 CYCLE EFF 64i1 CCEFF )41 -

CYCLE 0 3 CYCLE EFF 524 ACCFFF 3 CYCLE NO I CYCLE ERF 5A5 ACCFFF b8M3-------

cyIE Nn 9 CYr E FFF aSA ArCCFF ShJs- CYCLF 14ih CYCLE FFr h 6 ampCC FF 591

CYCL E 1 7 CYCL E FFF 6 ) AC C F F S97 C YCIE N A CYCLE FF A36 ACCEFF 2 -shy

CYCLE Nrl q CYCI E FFF f35 ACCEFF 106 CYCLE hOlI rYCL7 E EFP 612 ACCFFF hII -

E wcaiEI CYCLE FF A51 ACCEFF ble q - --- CaLIE 121P -- Lu I ACC FFF Iq - shy --- - shy

--- -- 0 13 CYCLE EFF 6s ACC FFP h2 1-- CYCLE

Ott G i CL F CT o

o shy - - -

00

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 51: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

co

OUTPUT No 3

PtWMHWMwM 4 N iMWNW WM iHM d HH eMHV841 WHA MWe14 AM bebeb Wbe4)WM WAW 4W MAW 4vi MIr4$ 444W18MWIM AMSwJ W 1W0tiM f4 MwH 441g kMtM 1tH 41 gMWH H4hI i M

TOKEq (1)(DEL-RUODX-RVOOY) U-PXFIEQUATIONS - NAVIEshy- () (EL-IJIO-PVA(OV)V A PYF2 -- (3) tLX+VYFI - F FUF2F30 -

VxSQRT(R)FXY PR22(X2+Y2) (FXYmI+X(I-w)Y(CI-Y)FUNCTION VALUES -- JYSQIRT()FXY8Utr)ARY CUNITIOHi AND rhtrAL

FASLIKF INJECTION3FPAPhTE RampLAXATIONSLINEAR INTMETOD

COAPSEST GRIm) 15 2 XINTFRVALSAN() 2 Y-INTERVALS Hil 51) -(IF IbS shy --(WUZER L EVEL 4 44 XG4 ampRL

METHOD PARAMETFWS ETA io OELTAu 300 T ElL s jnnEnu

-FI1 350 3I(Q - - - -- RFYNOLt) 11FR 41OE+ -shy

6w O 1J W tIVJ41 8 aE 3 i-LT-2LEVEL 0Rt - shyEo -shy

4 ~ 0nE09 kPARF41)36 1OnIoa 37utr+nm

END CYCLE NO tshy

2m 653F~lh I 41114a1 UlItE~fb u7E0 --+ -- -

3P50 7 +06nE6 -wE m-n)so n28E+U o 1147E+405 3QEioS JME+bit-E+04

9E+n3 h+E+A3 - -E+03 6 9PSO 8iAMA dE+(

--oN) CYCLE NO 1130r4 qnats t YLE N shy 6 b25O I6AE403 5bE (13 lft+A3 5 m mm mm sgmmm m m m~~m (

-- S 511lr03 QIAn2 136F$(ILI I E+1 - -

1hFsh dOAE+t3 ptiRE+o3 9RSE40167SA - IA3P IIOE+3 TjI3 2AIF+0shy

b -r_ 113s 32En03 - 273E+03 157E+01 7251 +112 168k 4(3 3qQE+(O

u

3 $153F+nl IRA E3 iAE+O0I58E+13 267E+03I 60O0b

-- I7E+n3 - 00IF-I

2 q1 71QF+np = lh6E+ 2 6910 - 57ni411 -- 357E+02 shy- 9F+n3 shy

PQE(12qlq -Q1E+n2 - 17DE42127E+I 7qE+(3 ---- 70hF403 I46E-012 h919 - ShFO

90$ 156E403 IiAEo3 71QFmO42 P q17Fn2 - 2QE-n2

2 6927 771E+n2 It P6E+113 E+02 -- 37E+QI F+ -A02 40 aE-03

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 52: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

2 0~ o3E+024e~ -- 166F+02 Il7E-2S2

1 wEo I A E P3 I115E+03 403E-013 6)67

3Fn F + OO11 7 1 17 g fl hF (13 6 7 E + n 3 I20 F)3

2 7 4 +0 I

s )P l h P7F + )2 r t+n3 - 6 9F centO I s qE + fI5 I 7oI oh lIOE+ 03 I5 iE~n (aISE + 3

5 iti 7n $1 +w p sqfkk +12 352E +02 10 1E+o

6 Q i 1 +O +111| 3+ 6 573 E+VS5 5 sE+O qtTu5

6 1 1170 6 17 e (p l7 R +113 497F n 77PF+ ni

6 t217 0 30 r 02 4 PE+ (1 297E+02 nek+0 1 = -END CYCLE NO 3

-0

5 1 2 42 0 M1 F n 2 I b F+ 0)2 I 7 E + 3 18 8F+n t

5 1h7r 33QE+ti 6R4E Q P7PF+D 12 QE +6

12 13 2 31 IE +A 22U E+to2 IQqE 0P 372 F+0 0

2 12 A30 InE A2 I S E 2 I3 E +02 IlQE- na

3 1 2 46 2 07 E u 3 4 0 1E + n3 3 5 9E 03 lIb5F - 01

I P 6 1 S 6 F R + 0 2 F + (3 UP SE +1)2 I (19E - 61I

2 1 P F 34UF401 11 PE+nl A~tE+nI P(F02

131F-402 23 E n 613AE+02 110TF-n23 1289h -

3 1 412 Sh tF~n I bOHE+n q93p+Ai 308E-n

11 1 75 A 2 F + 3 2 E+ o I P+ b ampO 9 7 2 6 E - 0 1 eshy

4 13037 Ll4QF~n ttII 276F+112 (E-At0 17 4t (02 p32E-nj -

S 13 tOOn1+pI71 2 E 02

3 1$ 1it anaE+Ol 6t7F+01 53I)E+01 5 7F-02 C

3 13 13 1 s3 F o| 7u E+nl 7 9E +01 37 E- 02

2 lt its5 t5 P4 t 3t E~n I 1 3sF+O1 I OPP-02-

3 13 15 0 2 5 3F+ n 2 2b E 4op $45 4E+ n2 P IP E n 2 w (

3 131Ih6 OQQ F4nI Il7E+02 87qF+ I I oF -Op

133 17 U qbE nn 141 E+4 1 OE4 0E1 64 7E-63

3 13)IA1 IQOE+A 2511 +P2 2F (12O 0971EiO

3 1 P 513F+01 663E+nl 781E+01 09~F-63

3 13 p3b 2cqF+( I57qE~nI 2b EOt 615E-03

13 qq_ 8 ujF 07 IibE+t3 125E+03 - I(+hE-n

Z13 36 1 18 E+ n 2 E+ nl S PE - 0 2 1 bO + O 0 5 9 8S

494E+041 Lq$E+O4 467E+O005 11611 S3u E+og 2H7E~n2 2q5E+00s 13Ahl i 7 F+n3 33F+03

i+)It11 7FT +n IP3E+02 P2iIE+001 5+nlS 151tt jheE~n q Isq k 0 s 1 3E+05 35

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 53: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

A 171II t5En I7F+P 130E+nI AIE+0 I 6 18111 8A4n1 - 18 fLA2 - 62ampb1 tIOF+OI

- - END CYCLE NO--Q 5 18361 1qEn2 7E 1 - 496E+02 SQqF+n0G5 l86it 1n0 +112 P35E 624E+0t - L26E+1O a 1d 67L 82OE+Ia hlSE+01 - 17PE+A2 t12E+no- - a tt 7 30 lq97F+n1 U+o2 902E+Ol 51IE-01 3- I75p - SuE~nl 320L~nl 9OQE-Oa -3QAE+01 U ip8iIu I~nEn3 2311E413 - 3311-3 3 P64F-01 - shy

- I - EiA7 - abt6367F+01 b5E - -- - - 1dE3O7 11M439 Ptl)F+Al I A3jE~o1 PPf3F+0I 933FA ------ shy

141t19 a qbAP+fi I33Efl - - lLt+01114 I4E+oO --- -- shy

IUa39 -234K+02 tS2EoA2 - 7QIEernl10 J+I- 8+ 5 19 nd -- Ih E+1l - 3d+6 - I20e+0I 5113E-0I0

6 PA~6RQ - 3OqE+n)4 - 42PE~ou -- Oq~bFtOL04 746E+00 ---- - - -- - - ---shy

6 2loAC 1IQE+)3 313k+03 - - 272E+02 SE A6 2bt 27tF4 I 7E+0l - -143E+O1 SF+00 4 23689 pIiQ 401 4L69r+ot 896Q~F+00 15E+0(l 3 a3 END 0+0 CYCLE NO- shy5 a3q- 17F+5l 7E+eFl1 IiE42 6lE+005 24 -1-927 F+III 4 1V4I -n 4E+O - IOqE+0)- shyu aj52 I7E 0I 137F+I0 3h1E +0j q 3ni --- I 243i3 231E nl f+ni - 1L4IE+01 I I -3E-0I

Fn46E 64 5E+(10E+0 21E-11 3 P133 4 0 shyd a39Q3 3qEn WE-02IiSE -616E+02

LI - 2ub -- 317F+Al 533E+o0 3 7 3 F+ I -- 3A - aSI0 a---- 1A3---l-2+ +3r031 P t --- 1 liELo - - -3

3 2a 5I+n0-Uq t +f 0 3 n SLI5Q6 it7+ I r35Eol 9 9 +1t - 145E-02 -

S 24658 2IAF+10 1 171E+ -l 249E+01 F--- -- shy

3 p - - 67L p AE o SIE+nO---- 30t1so+00396E-03 UI P4736 - 3 qSF4 nl II +nI - 7tF01 - 701F03 --- shy

176FO1 - 11 nP701 11- E+01 -- -- - 700F03 - I - 2d$IU +n k+ 0O 75F-03 ---- --- shy2) Fff aI) P3 3E0 -o - shy

a a t7 7 - 3PIlF +I11 1ta71+ UhL40+01 536f -03 - - - - - - - - - shy

dI J3q lalF+n1i 2624 nj4I 41L5E+01 - qpu~En3 ---- - -- -

I LAss 17Qr1P4 73t)E+on t F+O6 CS5E 3 lI 5AnIA P2 F 1I 347F+ni Phigt4Ot 40I3E-n3 ---- -- shy

4I -5OA I(OF4oII 9+(- O I - I IIFo I - 3ASE-03 -shy ______shy

3 240O 6IF+n 2bi+E00 nqmE+GoCIQ1 +O0 a 2r158 147TE+01 ZhE+nI 1$5F+ll I - 313E-39 52bp RLF4A IU7E+nt bS+ 3jqFfl3

5 5J7I P31 -+o3 SCt+I3 -- 111IIE+03 139QE-0l5 25721 uc1ir+nI q3blE+fl 2AdE+oI lIJF-0I

5 P597L 6qQO~F 6 -13SE+o I 640E4+00() 121E -0q I -shy

5 6f)21 2AUC+Ae 5~33EWn 71E+00 SpqEA- --- -- ------

IS 2721 7h IPttc13l IO1E+n4 -- I( +0U IBIE+00 b 29 12It 19 Ut+ A2 513E+np LnF IIEO -- -shyn 8F + - - shy

6 2qppt 811F00 - I 606uii - 683E+00 - 117E+O00----- -- - --- - - - -shy

6 3)n22 I QRqL400 7bE+on 1135E+00 8MQ7-01

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 54: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

s 31) 721Si--S ~~ ~ ~

I7E i t-- pF(iEI -shy

-shy I E On p6tlE~ont-shy

-I IE I h47 E+oo

183E-01 aonE-02

I 3

3nw 3n677

I48F460 3AIF(10

I h +00 6037E+00

59F+00

65nF+O0 332E-03 173F-tl

-

I SOAAh t 17F+00 -- IIAAE+nO I 7E+00 W( E-

2 30890 5 (AF + nit 7 - t S3+I F l tAOE-O

2 3n J49 1 F4 0PIE-n nAEo IsinE noI hl7E05

3 AQst 1734 cent0o i1E+0i( 1 SE400 176 =O

3 3)907 I ft)I

35 F11-n jV3F+nl

|75FfEshyl 2u7F+Aj

3lhF01

70IE+ol PNOE-o

7AI -OU

3

2

lo 9)4

303ia

+10

6137E-llO

IP~E-nlshy

thE+1)O

659L400

qalE+O01

SakE-00

I5 E-04

3

3 3shy

31105

R10 9a 3 1 L

17J+00

jF4(MO SAIE-nI

13A E+Wl

ahF41 shy- I(S -ol+0

oISUE+O(

1023 065F01O

17CF-04

a5F+AQ11 641E-05 -

a a

31 (1660 31 12 AhlF+011

I67E+o ISIE+n

762E+01

95UE+00 37nE-0I 4O O

4 3

31 0141-O -1 317070 5750 -ni jq

E 0

6IE II

pj]lon161)F4041

7021 -01 SM -flil

5 a

3Z$092 39 1)s

274k+AOl 57 E-ntl

qT -Fo1

2E-nl 234h4-00 i 1 E+10

VE-02 SPSF-03

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 55: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

01

2 2

z rn~1t 1 3732o shy 37 I4

37328

-5 u0--q 257E-C0-2tqFAl

shy I62EA I

76Eo 1ZIEo -hQTE-02 777E-01 - 31RE-01 -shy

SI007- E-0 t 3AOE-03 -IPIE+0 1 lE-O1

777

-shy-

shy

0F-a06 lQOt-051-0E -2E0

-

-

- -

3 571Uda PLF+ 351L+Ofl 1Iliutho tl71 F04 3 113jq I d2~oO 6hE+II0 lqE+00 IJF-o 04 2 3 3 2 -

37361 3717 373Q5 37398 -

1IQ -I hUE-oi 2ubEriI 260E-0

-

2B31)E-IIOEkfo S7E-(

367F-(2

- - 6AU2EO IhSFOo

- - PA7E-O 3 7E-0

pqPE-0)5 63PE- 3QOE--

02f-f5

-shy

- --- ---

3 3

37114 7U3)

3ql -n5Eshy$nFtp IIA-tl

- 119E-0I9-SEP01i - 6aIaAE1 i97-E-05

-

-

Ul u

37taA 37 555

71sE 0 Isk+0A

-) 11 7E401 2R15flO

H iMlt+oo IhAE+on

I IIE0113-- 705F0Ls--shy

- --shy

--

3 17570 -F) - 3 7l-nlIPPP -shy 2 E-01 - 15Enashy 14 17633 15lF00 P41A0of tIAbF+00 339F-na -

U 37A 59 I E-1 7 E-A-(-l 237k-ot PI F-04 a I7758 S II -07 137E-fl 11AE-O1 156E-0 C r 5 -shya -

1 nno 34 t 3R50 3-657

u2sr+o 3kn0 SQQLI

1 2E-nlI -shy

hUqE~noI uViE +10 I)L+0Oshy173F-11

-

-

biiE~fl 177E4nP

0b77k-0l 28 E-0I

-shy

A726i13 2t)E-03 htlt-03 QiAEtu

- -

--shy

shy

r 5

31APO 39070

IMSE+o l14I7p

-shy 1s68F+00 3A4E- 1

-shy 53Fo+00 I6hf-0l

- 1IAQE-03 SE-03

6 6

CU7CI41n4

1AqitIP03E+no

- 24AE4np I3E+ol -

21gt~E+02 237E+MO -

50260 93E-02 --shy --shy -shy

h 6

-4u7 - 6 E-ot 03070 2op-ol

-a

-shy I12E+0(0 57E-oI

--shy 72iE-01 293E 01

flOE-n2

18(F-() ns powENO CYCLE NO

5 b

433 O5 357n I3n33

- 21060-shytI E-1I

IflQEtll

- QOEshy2SEshy 1 49J(E-02

-shy-

sp5E-oll 7QE-02

262P-ll 1

h0A-3l nqbs-03

1111603 --- - - -

UI 41h4i5 - qA7E-OPshy - I SUE- I 1I- 7E-101 - 011 -fOillI - - --shy

3 d37 - 0 7UE-02 7-IIFn - - Q FIE-vt IME0shy3 2

437V

417 1A 121I Fnj

84hpnl

7 -t2

E-i -

-- 25 E-01

423-E-02 -

71 LF0 I IE-nS

--

3 3

P

a3i7t 1

413 7 60 - -

bVIE-0lI 37h2131l 1 PF1

A tIoEI PA 7F-o

Pshy1Et(I2 -

9AhE01 031F02 3276fl

lqE-05- 1 06E

-997E -olh-shy - -

-shy

_ ___

I 9178t 2hE-nl 41HE-01 - 320E-n 33E- - -------shy3 a5747 -SQUPOp IVE1-01 - 6172 nI8SF05 2

3 a3 at)1I 13tl

I5aF2 6b5E-fA

)S-l7PLIE)-O shy

PIAE-02

I I -01 AAOEnh 12116-Os

-shy

3 UL3832 L4 fiI 11i ASJEWn UatEshy -A 11 3E-09 2 aI33 I(I006-02 ISOE-02 i IF02 - 37Q6-flb - --

I 3

33 A 2 1387

U0hi-Ip 21qEnp

622Ff-)

048Eshy5U2EitP

- 236E-02 I41(T-A6723b-n6

------shy

----

-shy

41 4 3- IsqFE+11lA 90E+nl 2 iE+o0 2PIEwOU - - -

I S0I2 PAAE-l I q6PE-11 P12E1O I uF -04

3 U4U60711 2a sJ 27E0 -IAA~ Lashy -

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 56: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

a UL2133 276E-1li a27E-n - - 3626-01 460E-05 - d atlaIQS -aa 5 7E-ai - 177E-2 - 2861-os

5-73 +o- IEl---- Q969E-noo - 26E-33 - - - shy

aUa hQs W E -P l iSOFt a - 3 7 9 E- 0 1 - QhPE - o - - shy

5 44a945 I1UOF -01 233E-ot 16713-01 taqE-on--- --- -- --------shy

4 LEAnA IAF a-E-92 -- 0 1uE-O 2E-O0 -

5 u3SA (IIRE fn( 137F+60 I22E+0n 377E-na 5 5i0lSOO 57F-(ll I A1E-nl - - 144E-2 agqEou shy6 IO0A - ZUIEA4i 341E402- -- 35k+1 I h3E-n3 - - shy

6 U7 5h8 I 6 + vIn Pq2E+oia 673E-fll - iP6E-03- ----shy

6 L850$8 16AE-11l dMA3501 1Q3E-01 l3SE-l3- - shy

6 o C418l 711 FoA) I PAE-nl - 7)PE-A 377E-n3 EN CYCLE No

S U9Q758 -514E- IbF-aP- -pIE-0I 136E-03shyr - SO 006 P23F-oP MI~E-12 - 1S-02f) - QUF-0Li - - - -- - - - --- shy

I 5n070 25PE-I1P I23E-112 5SqQk-2 Pb6IE-flu l 50 133 2 0E-u 2UE-o12 221E-02 46E-04 - - - - shy

3 5otl 1PE-On E03E-nl173E-Da - 3014E-AS- - - shy1 51 Ih 23F-o2 3EOA2 1 SE-02 III E---shy2 6 E --- -- ------ ------------ - ------------- N ---- -shy2 - 28EE 3 - 1) AS - t31 111 01E -0i1 h 17F -I S5o IQ PStF- 53E-o -2 16E-0l2 32E-oh - 2 50(3 172E-03 IM8E-03 - 312E-03 hh6iE-n7 - 3 5ad9 AA-2 -319 -- 3 E0 ME-Oh --

50 1 PS3E -( I t

3 50 J34 7AgFO1 153E-o2 -- 71 IF-03 I136E-Oh -L - J - ~ -- -- - shy

2 S03~~ - hE - 63 1l70t-03 II~i ~~-7-shy3 - 0 -5 lIIPE -AP 141E-nA2 7 9 L0 P - M62E-07 --shy

3 rkl0 l2E-)3 33hF03 - 214LF-03 612E-074 5033 hcSE-11 bQ7E-aI 63[E-01 443E-05-shya S39 8- ABPn2 - Wl6E-02 2k52-05--shy14 1 U57 6 I hE12 161 F-as shy09F-03 104I4E-13

3 511 7 3 16E-0 21 - 7E-n flhE-06o -- - L3 0 - l ti AOEn b1b-J shy5n 35 L63E-na F n 2 - -63hD - - shy

a 5n5qm L UPFaI)I 7q()ujE-on3 U7Uh-03 513E-06shy55 1951000 291F400 5AL13 267F+001)Eo

5 - OIqM I 11-AF ~ u~~ - UQ -h4-o 23RE-04S I 311A 135 A 21t-4 laF-Da lpl-oa I -shy

6 5 3L18 57E+A0 MIpftO 719E+00 147E-03I6 533418 3SlFflI HIPE-fll t~fF- Il - 1352-03 - --- ---- - shy

6 qu3U4 1qfl -v 371f(12 29E-n2 l021)36 55l3I 891-03 ShF-n2 7QUF()3 Inro

mawwn END CYCLE NO tOs 55sq SOF II2 phqE-p3 113jF 2 640E0

s s5sU$ SQ5E-o3 143E-n2 2S7E-53 51 E-0451E0np - - --- -shy ss91a 6ghr-03 ot222A3 1 AA II - - -- ---shy

14 55973 AEL-03 IUIiE-02 21)F -13 30E-0

3 5S988 205E-03 - 29E~nl ItbL03 An7E-ob - -- - - - - --shy

3 S6004 13701-03 5U3E03 bPE-03 lunE-062 5600A 59520(I 1 inIn btIIUe-n 5oE-073 lSL23 311h02 tHEn2 - E a2- n h_____________I7P~l7 - 2 A n F

Cr wt

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 57: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

-- - - - - -

- --------

- - - - -- ----- ---

-- - - - -

---

-- --- ---

--- -

- - -- -

----

---

-nS4

3 sh055 I1QE-(3 -- I50E=O3- - 39T d3 76SE07

Sh =3 xEn

4 56111 1I36-n1 - 2IE-l 17 E-I(j OE- 5 5lbo IfE-13 - - -I IQF-i -- 51E-03 726k-O6

(I 5hlJa) t6jF-3 IlI3R-l3 06-IAE-fl3 ttBEO6shy3 e49a 70OC-Ot IO6w4no - 9JIF-O1 O75E-O5 5 5 7UZ 66F-(12 67qE-1 815E-03 ha6 E-I15 s S699 67E -1t3 uL71En3 2IsE-(u3 - LAJE -05 57U -s 8e ldS $O3 halI-CII 367E-O

6 2 2 ItI I -- -PEnl --- I3RF= 2 - Q9 -un --saaaa = 2-O--4DA PS+O-- --shy =6 doz Pn - - hlh-03 - l E-n3 - E-ob

6 612U2 9A (13 l- n IEE4E-03 shy 351E-u u-- 1- w ww

5 6IUJ a 39mP-(3 1T3E-P13 QQUF-03 -31E-07

9 61 2T2 2PSE -03 - - sloE-O 11-0 - 2 03 -j913 n a 6l AiU5 l57F-CI I9E-(3 3 175E-03 l n5-ou 11 6I1Ph7 ISAFP-01 3FQE03 67 4E-04 IU0 3 61h83 470Of1 53Et 04 81 E-0l 1E-ob

1- 1 Q T7FL5fp - 5QQEI- l4P-A2 rsok-nshyu 62Qod IOPV-nl IhiE-n3 1I32E-03 1EO ua - A07l - POE-04 IWt -03 -- l(SE-03 - 2P2E-11 3 62l146 216fF-1U 2E0 34I9E-04 537amp--1

-3 -- b2 Ina 7$UE(104 430F riO 18 H e03 364tF -07-shy2 12111 131E0LJ 243E-04- 13O0EQ 93HE-nil 3 $12 - aPE03 L8 7-C3 9AI)F413 22IE-07 S 0P137 -- 17 0 E-03 - 374P-3-- 134E-0O3 - 107F07 P o - q O 6 1 I7Fr4 - I- 3E (I-- l- - -162EA1 h7t~

3 0) Io IJhnF01 33F- n3 P25SE-03 9pqFOA 3 6) 72 aULok-0flU-0 M33j U43nlF-OU 2 i2I7b - s7oEt-l h77E-6S 9blE-05 17E-08shy1 63 IQ I 5t I Ful h11PE 00l fi5qFOpaq -0I h 2A7 lbuP-0 2boE-Ad - lriJF-114 Uffl

70 - Ah 27 0 RE-a3 ~I iSE12 - qh7E-03 shy 84IL3E-07-M 62 332 - QiI 04- 19 1E- 03 - 790 -0~b uQQE-07 a 62Q1 5qF-J iSEnq 3pQ0uQ 330-07

S 6)haU j~nFlP 293L-01 250E-111 PSRF-Ds-shy5 h2945 ui3lII3 I iA-02 87IF-04 4A-0

q 63 145 IQIAF-(4 7204-n4 I1Ak-lh omoEflnik

h 612145 hUAr-oI Q19nEhFOI1 - 3EOa ushy6 65 ls 1 Ii[12 694E-02 qghE-03 2IQFa-041 h 6I1U5 A73FOIS 1b-(03 975E-04U 11010IJ b71U5 649 4F-04s OIE-113 2PhE0U 41E

S 0195 811E04n 3113E114 I9ME-03 331E-0S r 67hL3 Sb~rF0U Iqob-nI - 79- 21S3t-05

LI 077 240R~0U IIE-nU S34JEOU0 U1$F-0A u b 77071 3J3u R(I4AfEAuJ 1~76EmOI4 20l3F06

3 bJ785 I17E-013 7VIEC5 - 198F04 1F-1

La 67910 _plusmniL3Eft4- 1170-EflQu2 It3 IEI04 - -lO E-n7

Ln

-

- - -

-

- - ---- --- -- -- -- oN --- Y--E -O --- shy

- -

e wEND CYCLE NO

-

-

-

----- -- - ------shy

- -

-

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 58: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

- - -- --- - - -

- - ------------

-- - - - - - - - - - - ----------------

OUTPUT NO 4IO LAO

oSOOTHING rACTaP = 5000 THEII-571 03WEIGHTED IHOOTHINO Aol -28

OPEATO 119ooooXAr

USUAL CAVLCSIamp o

AC 1- NUMbEA J7 10 7~H0t P REfaetwTs ~ o 4 t10c EL rM

3 0 100 0-00 1 0

p-Io0 L

fozzcz o 0-

1 ~-- ------------ 0

I-COZZZ 0000

f 5 N w O - C f ~ I C h ~ b 8 b O b ~ b ~t

~~~ -

-------------------------

V0

---- -C -0 c0

Ph t -- -- - - --shy

------- -~~- -0- - -- -5- - N

lz 11 1

IN1 - --

0 b

1

I3

-- - - - -- - - - Z az tI-Z3zz

]t-- --- ----0-shy0 0 0

-2fshy

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515

Page 59: ASPECTS AND APPLICATIONS OF THEdifferential equations. nathan dinar. college of william and mary. w.illiamsburg, va . 23185. nasa contract nas1-14972-3 (nas'a-cr-158947) on several

- r r-1R7-tlJrr - -r - InFfl --

--- 4~E~ q1 3 iPOn a 0 I a E u 2 - 6 2 Eshy- 0 E ~b - amp S Ft--7- E - O 6 shy

----U~ -- -- ---

C 7U3EO2b-R -67t723 6- - - shy

6d73 Abqna l732-na2E-Ou 339Ej04I7Q9-P4 A40-RM 2-5E-06E-05--

AQ73 15A P 34Etl P F -11 365Enr -

II 5E-fl2 - oqiAE-OM 279E05

C bRA7P3 QO-0

6 7fl73 41qE-01 - 1iEO04 - 7ppEoO5

shy

6 It723 iL17E-114 28UEpu e 7 1A~nU7f 31EflL 112E-04 190E-05

CYCLE NO 1i

a CYCLt Fo ACCtEFF 2Qb77CYCLF NO 7q2 ACCampFF o42CYCLE NO a CLF EFF a77 ACCE FF F 9 CYCLE kO a 5YCLE EFFP

5 CYC E EF 77 ACCEFF 5b6 ---shy- hOCYCE -- CICLE NO 7 CYCLE FFF A 2 CEPFF b2UCYCLE No 6 cYrLF EF 7A ACCEFF 5b

- CYCLE N0-O-CYCLE EFF i CCFF 66A

n - - ----- - - - CYCLE NO 12 YL E~ 7 30 CCF q shy _ 7 2 CCEFF 82

- CYCLE NO 13 CYCLE EFF

--0

tWOn

1515


Recommended