+ All Categories
Home > Documents > E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

Date post: 01-Apr-2015
Category:
Upload: leslie-barrow
View: 213 times
Download: 1 times
Share this document with a friend
Popular Tags:
30
E. T. S. I. Caminos, Canales y Puertos 1 Lecture 2 Engineering Computation
Transcript
Page 1: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 1

Lecture 2

EngineeringComputation

Page 2: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 2

Smearing Occurs when individual terms are larger than summation itself. Consider the exponential series with x = -10

Consider formulas such as:

With 7-decimal-digit accuracy:exact answer = 4.54 10-05

computed answer = – 6.26 10-05

(45 terms) wrong sign !

Largest intermediate terms are:9th = –2,755.732 & 10th = 2,755.732

2 3 4 5x x x x x

e 1 x2! 3! 4! 5!

Round-off Error due to Arithmetic Operations

Page 3: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 3

Error caused by the nature of the numerical technique employed to approximate the solution.

Example:

Maclaurin series expansion of ex

2 3 4 5x x x x x

e 1 x2! 3! 4! 5!

2x x

e 1 x2!

3 4 5x x x

3! 4! 5!

If we use a truncated version of the series:

Then the Truncation Error is:

Truncation Error

Page 4: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 4

Approximations and Rounding Errors

• Precautions:– Sums of large and small numbers: due to equaling the

exponent. They are common in sums of infinite series where the individual terms are very small when compared with the accumulated sum. This error can be reduced by summing first the small terms and using double precision.

– Cancellation of the subtraction: The subtraction of very similar numbers.

– Smearing: The individual terms are larger than the total sum.

– Inner products: They are prone to rounding errors. Thus, it is convenient to use double precision in this type of calculations.

n

1inn2211ii yxyxyxyx

Page 5: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 5

Error PropagationErrors which appear because we are basing current calculations on previous calculations which also incurred some form of error

Stability and Condition NumberNumerically Unstable: Computations which are so sensitive to round-off errors that errors grow uncontrollably during calculations.Condition: sensitivity to such uncertainty; "well conditioned" vs. "ill conditioned"Condition Number: measure of the condition; i.e., extent to which uncertainty in x is amplified by ƒ(x)

C.N. 1 ===> "well-conditioned" C.N. >> 1 ===> "ill-conditioned"

Page 6: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 6

Basic Idea:

Predict the value of a function, ƒ, at a point xi+1 based on the value of the function and all of its derivatives, ƒ, ƒ', ƒ",… at a neighboring point xi

Given xi, ƒ(xi), ƒ'(xi), ƒ"(xi), ... ƒn+1(xi),

we can predict or approximate ƒ(xi+1)

Taylor Series Expansion

Page 7: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 7

General Form:

h = "step size" = xi+1 – xi

Rn = remainder to account for all other terms

= O (hn+1) with x not exactly known "on the order of hn+1 "

Note: f(x) must be a function with n+1 continuous derivatives

2 3 nn

i 1 i i i i i nh h h

f (x ) f (x ) hf (x ) f (x ) f (x ) f (x ) R2! 3! n!

n 1n 1h

f ( )(n 1)!

with xi xi+1

Taylor Series Expansion

Page 8: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 8

0th order T.S. approx. (n = 0): f(xi+1) = f(xi) + O (h1)

1st order T.S. approx. (n = 1): f(xi+1) = f(xi) + hf '(xi) + O (h2)

2nd order T.S. approx. (n = 2):

nth order T.S. approximation will be exact for an nth order polynomial

2 nn n 1

i 1 i i i ih h

f (x ) f (x ) h f (x ) f (x ) f (x ) (h )2! n!

O

2n 1

i 1 i i ih

f (x ) f (x ) h f (x ) f (x ) (h )2!

O

Taylor Series Expansion

Page 9: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 9

Zero orderFirst orderSecond order

f(xi )f(xi+1 ) f(xi )

f(xi+1 ) f(xi )+f '(xi )h

f(xi+1 ) f(xi )+f '(xi )h+ )+f "(xi )h2/2!

f(xi+1 )

True

f(x )

xi+1xi

h

x

Taylor Series Expansion

Page 10: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 10

Objective:Evaluate the derivatives of function, ƒ(xi), without doing it analytically.

When would we want to do this?1. function is too complicated to differentiate analytically:

2. function is not defined by an equation,

i.e., given a set of data points (xi, ƒ(xi)), i=1,…,n

i 0 1 2 3 4

xi 1.0 3.0 5.0 7.0 9.0ƒ(xi) 2.3 4.1 5.5 5.7 5.9

0.5x2 cos(1 x )e

1 0.5x

Numerical Differentiation from Taylor Series Expansion

Page 11: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 11

Numerical Differentiation from Taylor Series Expansion

– First derivative with backward difference.

)xx)(x('f)x(f)x(f 1iiii1i

)xx()x(f)x(f

)x('f1ii

1iii

Page 12: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 12

Backward Difference Approx.:

 First Derivative:

first backward difference

2

i 1 i i 1 i ih

f (x ) f (x ) (x x )f '(x ) f "( )2

2

i 1 i ih

f (x ) f (x ) hf '(x ) f "( )2

Letting h = xi - xi-1

2

i i i 1h

hf '(x ) f (x ) f (x ) f "( )2

i i 1f (x ) f (x )f '(x) (h)

h

O

Numerical Differentiation from Taylor Series Expansion

Page 13: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 13

Using data below calculate ƒ'(x1) :

i 0 1 2 3 4xi 1.0 3.0 5.0 7.0 9.0

ƒ(xi) 2.3 4.1 5.5 5.7 5.9

First Backward Finite-Divided-Difference at x1:

1 0f (x ) f (x )f '(x) (h)

h

O

14.1 2.3

f '(x ) (h)2

O

f ' (x1) 0.9 { + O (h) }

Example of 1st Backward FDD

Page 14: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 14

Second Derivative:

2i 2 i

i 2 i i 2 i ix x

f (x ) f (x ) x x f '(x) f "(x )2!

+ O([xi-2– xi]3)

with h = xi– xi-1 and 2h = xi – xi-2

The 2nd order approximation to ƒ(xi-2) becomes:

ƒ(xi-2) = ƒ(xi) – 2hƒ'(xi) + 2h2 ƒ"(xi) +O (h3) [1]

2nd order approximation to ƒ(xi-1):2

3i 1 i i

h(x ) (x ) h '(x) "(x ) (h )

2!f f f f O [2]

Backward Difference Approximation

Page 15: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 15

Subtracting 2*[2] from [1] yields:

f(xi-2) – 2f(xi-1) = –f(xi) + h2f"(xi) + O (h3)

Rearranging:

h2ƒ"(xi) = f(xi) – 2f(xi-1) + f(xi-2) + O (h3)

Second backward difference

3i i 1 i 2

i 2

f (x ) 2f (x ) f (x ) O(h )f "(x )

h

i i 1 i 2i 2

f (x ) 2f (x ) f (x )f "(x ) O(h)

h

Backward Difference Approximation

Page 16: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 16

Using data below calculate ƒ"(x2) :i 0 1 2 3 4

xi 1.0 3.0 5.0 7.0 9.0ƒ(xi) 2.3 4.1 5.5 5.7 5.9

Second Backward Finite-Divided-Difference at x2:

2 1 02 2

f (x ) 2f (x ) f (x )f "(x ) (h)

h

O

2 2

5.5 2*4.1 2.3f "(x ) (h)

2

O

f " (5.0) - 0.1 { + O (h) }

Example of 2nd Backward FDD

Page 17: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 17

What points are used for each form?

Backward:

…, ƒ(xi-2), ƒ(xi-1), ƒ(xi), ƒ(xi+1), ƒ(xi+2), …

Forward:

…, ƒ(xi-2), ƒ(xi-1), ƒ(xi), ƒ(xi+1), ƒ(xi+2), …

Centered:

…, ƒ(xi-2), ƒ(xi-1), ƒ(xi), ƒ(xi+1), ƒ(xi+2), …

Other Forms of Numerical Differentiation

Page 18: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 18

Taylor Series and Truncation errors

- Higher order divided differences.

2iii1i

2iii1i

h!2

)x(''fh)x('f)x(f)x(f

h!2

)x(''fh)x('f)x(f)x(f

- Second finite central divided difference

Page 19: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 19

Forward:

i 1 ii

(x ) (x )f '(x ) (h)

h

O

i 2 i 1 ii 2

(x ) 2 (x ) (x )f "(x ) (h)

h

O

Centered:

2i 1 i 1i

(x ) (x )f '(x ) (h )

2h

O

2i 1 i i 1i 2

(x ) 2 (x ) (x )f "(x ) (h )

h

O

22 1i

- ( ) 4 ( ) -3 ( )f '(x ) = + O(h )

2hi i if x f x f x

Other Forms of Numerical Differentiation

Page 20: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 20

Taylor Series and Truncation errors

• Use of the Taylor series to calculate derivatives.– First derivative with forward difference.

)xx)(x('f)x(f)x(f i1iii1i

)xx()x(f)x(f

)x('fi1i

i1ii

Page 21: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 21

Taylor Series and Truncation errors

– First derivative with central differences.

h2)x(f)x(f

)x('f 1i1ii

2iii1i

2iii1i

h!2

)x(''fh)x('f)x(f)x(f

h!2

)x(''fh)x('f)x(f)x(f

3i

)3(

i1i1i h!3

)x(fh)x('f2)x(f)x(f

Page 22: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 22

Questions:

• Which is a better approximation? Forward, Centered, or Backward?

• Why?

• When would you use which?

Note:

We also can get higher order forward, centered, and backward difference derivative approximations

[C&C Chapter 23, tabulated in Figs. 23.1-3]

Taylor Series and Truncation errors

Page 23: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 23

Determine h to minimize the total error of a forward finite-divided difference approximation for:

i 1 if (x ) f (x )f '(x)

h

xi xi+1

• Round-off Error:

• Truncation Error:

i 1 if (x ) f (x ) hf '(x) f "( )

h 2

ˆx x x f f f with = machine epsilon.

. .i i(x h) (1 ) (x ) (1 ) hˆ "( )

h 2

f ' =

i i i(x h) (x ) 2 (x )RoundoffError h h

As a result:

Example Combining Roundoff and Truncation Error

Page 24: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 24

Total error = truncation error + roundoff error

E = | Total Error | h

f "( )2

NOTE: Truncation error decreases as h decreases Round-off error increases as h decreases

i2 f (x )

h

+

Example Combining Roundoff and Truncation Error

Page 25: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 25

Example Combining Roundoff and Truncation Error

Page 26: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 26

To minimize total error E with respect to h, set the first derivative to zero:

i2

f "( ) 2 f (x )dE0

dh 2 h

Solve for h and approximate f "() as f

"(xi):

i

i

4 f (x )h

f "(x )

Example Combining Roundoff and Truncation Error

Page 27: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 27

Using the first forward-divided-difference approximation with

error O(h) and a 5-decimal-digit machine:

e = b1-t = 101-5 = 10-4 = 0.0001

f '(x) = ; f "(x) = 0

i

i

4 f (x )h infinity

f "(x )

Linear Application:

Determine h that will minimize total error for calculating f’(x) for

f(x) = x at x = 1

Example Combining Roundoff and Truncation Error

Page 28: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 28

f(x+h)= {exact: 3.1415} h (x+h) f(x+h)-f(x) [f(x+h)-f(x)]/h0 3.1415

0.000001 3.1415 0 00.00001 3.1416 0.0001 100.0001 3.1419 0.0004 4.00.001 3.1447 0.0032 3.20.01 3.1730 0.0315 3.150.1 3. 4557 0.3142 3.1421 6.2831 3.1416 3.146

Underlined digits are subject to round-off error. They are likely to be in error by ± one or two units. This does not cause much problem when h = 1, but causes large errors in the final result when h < 10-4.

Example Combining Roundoff and Truncation Error

Page 29: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 29

Nonlinear Application:

Determine h for minimizing the total error for computing f’(x) for

ƒ(x) = ex at x = 3

Using the first forward-divided-difference approximation with

error O(h) and a 5-decimal-digit machine:

e = b1-t = 101-5 = 10-4 = 0.0001

f(x) = f '(x) = f "(x) = ex = 20.0855;

i

i

4 (x )h 0.02

"(x )

or about 0.01

Example Combining Roundoff and Truncation Error

Page 30: E. T. S. I. Caminos, Canales y Puertos1 Lecture 2 Engineering Computation.

E. T. S. I. Caminos, Canales y Puertos 30

Underlined digits subject to roundoff error.Bold digits in error due to truncation.

full precisionh f(x+h)=ex+h f(x+h)-f(x) [f(x+h)-f(x)] [f(x+h)-f(x)]

h h0 20.085 {exact = 20.085}

0.00001 20.085 0.0 0 20.0860.0001 20.087 0.002 20 20.0860.001 20.105 0.020 20 20.0960.01 20.287 0.202 20.2 20.180.1 22.198 2.113 21.13 21.121 54.598 34.513 34.523 34.512

Roundoff Truncation

Example Combining Roundoff and Truncation Error


Recommended