+ All Categories
Home > Documents > Using Mathematica to evaluate Laplace Transforms and their...

Using Mathematica to evaluate Laplace Transforms and their...

Date post: 01-Jan-2020
Category:
Upload: others
View: 5 times
Download: 0 times
Share this document with a friend
7
Using Mathematica to evaluate Laplace Transforms and their inverses, plus Cauchy’s PV prescription Laplace Transforms (straightforward in Mathematica) Here are the examples discussed in class LaplaceTransform@1, t, pD 1 p LaplaceTransform@t^n,t,pD p -1-n Gamma@1 + nD LaplaceTransform@Exp@atD,t,pD 1 - a + p LaplaceTransform@Sin@btD,t,pD b b 2 + p 2 The results for L.T. of derivatives also come out simply: LaplaceTransform@y'@tD,t,pD p LaplaceTransform@y@tD,t,pD - y@0D LaplaceTransform@y''@tD,t,pD p 2 LaplaceTransform@y@tD,t,pD - py@0D - y ¢ @0D Checking the solution to Sec. 8.9 #4 (solved in class using L.T.) InverseLaplaceTransform@1 Hp^2 + 1L ^2 + p Hp^2 + 1L,p,tD Cos@tD + 1 2 H- t Cos@tD + Sin@tDL Alternatively, solve directly: TrigExpand@DSolve@8y''@tD + y@tD Sin@tD,y@0D == 1,y'@0D 0<,y@tD,tDD ::y@tD Cos@tD - 1 2 t Cos@tD + Sin@tD 2 >> Convolutions Convolutions are defined for functions with infinite range in Mathematica To use for our functions (which vanish for t<0) need to append a unit step.
Transcript
Page 1: Using Mathematica to evaluate Laplace Transforms and their ...courses.washington.edu/ph227814/228/nb/LT.nb.pdf · Using Mathematica to evaluate Laplace Transforms and their inverses,

Using Mathematica to evaluate Laplace Transforms and

their inverses, plus Cauchy’s PV prescription

Laplace Transforms (straightforward in Mathematica)

Here are the examples discussed in class

LaplaceTransform@1, t, pD1

p

LaplaceTransform@t^n, t, pDp

-1-nGamma@1 + nD

LaplaceTransform@Exp@a tD, t, pD1

-a + p

LaplaceTransform@Sin@b tD, t, pDb

b2 + p2

The results for L.T. of derivatives also come out simply:

LaplaceTransform@y'@tD, t, pDp LaplaceTransform@y@tD, t, pD - y@0DLaplaceTransform@y''@tD, t, pDp2LaplaceTransform@y@tD, t, pD - p y@0D - y

¢@0D

Checking the solution to Sec. 8.9 #4 (solved in class using L.T.)

InverseLaplaceTransform@1 � Hp^2 + 1L^2 + p � Hp^2 + 1L, p, tD

Cos@tD +1

2

H-t Cos@tD + Sin@tDL

Alternatively, solve directly:

TrigExpand@DSolve@8y''@tD + y@tD � Sin@tD, y@0D == 1, y'@0D � 0<, y@tD, tDD

::y@tD ® Cos@tD -1

2

t Cos@tD +Sin@tD

2

>>

Convolutions

Convolutions are defined for functions with infinite range in Mathematica

To use for our functions (which vanish for t<0) need to append a unit step.

Page 2: Using Mathematica to evaluate Laplace Transforms and their ...courses.washington.edu/ph227814/228/nb/LT.nb.pdf · Using Mathematica to evaluate Laplace Transforms and their inverses,

Plot@UnitStep@tD, 8t, -1, 1<, PlotStyle ® 8Red, Thick<, Exclusions ® NoneD

-1.0 -0.5 0.5 1.0

0.2

0.4

0.6

0.8

1.0

Convolve[ ] has the form Convolve[ f[x], g[x], x, y (new variable)]

ConvolveB1

x4 + 1

, x, x, yF

Π y

2

Which we see is equivalent to,

IntegrateB1

Hy - xL4 + 1

* x, 8x, -¥, ¥<, Assumptions ® Im@yD � 0F

Π y

2

The order of the functions in a convolution doesn’t matter:

IntegrateB1

x4 + 1

* Hy - xL, 8x, -¥, ¥<, Assumptions ® Im@yD � 0F

Π y

2

Convolution needed for Sec. 8.10 #6

conv1 = Convolve@Exp@-a tauD UnitStep@tauD, Sinh@b tauD UnitStep@tauD � b, tau, tDHb ã-a t - b Cosh@b tD + a Sinh@b tDL UnitStep@tD

b Ha2 - b2LHere is the result we found in lecture notes:

conv2 = UnitStep@tD HExp@-a tD � H2 bLLHHExp@Ha + bL tD - 1L � Ha + bL + H1 - Exp@Ha - bL tDL � Ha - bLL

ã-a t J 1-ãHa-bL t

a-b+

-1+ãHa+bL t

a+bN UnitStep@tD

2 b

2 RevisedLT.nb

Page 3: Using Mathematica to evaluate Laplace Transforms and their ...courses.washington.edu/ph227814/228/nb/LT.nb.pdf · Using Mathematica to evaluate Laplace Transforms and their inverses,

conv2 �� FullSimplify

Hb ã-a t - b Cosh@b tD + a Sinh@b tDL UnitStep@tDb Ha2 - b2L

They agree!

FullSimplify@conv1 - conv2D0

Inverse Laplace Transforms (with no pain)

Examples discussed in class, done using Bromwich integral:

InverseLaplaceTransform@1 � Hp - aL, p, tDãa t

InverseLaplaceTransform@1 � Hp^2 Hp + 1LL, p, tD-1 + ã

-t+ t

Sometimes you’ll need to FullSimplify a result from an InverseLaplaceTransform[ ] to get back what you

started with:

LaplaceTransform@Sin@tD Exp@tD, t, sD1

1 + H-1 + sL2

InverseLaplaceTransform@%, s, tD

-1

2

ä ãH1-äL t I-1 + ã

2 ä tM

% �� FullSimplify

1

1 + H-1 + sL2

(Cauchy) Principle Value Integral

Typically Mathematica will not let you integrate over a pole as the integral diverges at that point

RevisedLT.nb 3

Page 4: Using Mathematica to evaluate Laplace Transforms and their ...courses.washington.edu/ph227814/228/nb/LT.nb.pdf · Using Mathematica to evaluate Laplace Transforms and their inverses,

PlotB1

x - 2, 8x, 0, 3<F

0.5 1.0 1.5 2.0 2.5 3.0

-5

5

IntegrateB1

x - 2, 8x, 0, 3<F

Integrate::idiv : Integral of

1

-2 + x

does not converge on 80, 3<. �

à0

3 1

-2 + x

âx

If we wanted to make sense of this integral somehow, we can use the Cauchy principle value of the

integral where you integrate excluding the pole and add the parts together.

IntegrateB1

x - 2, 8x, 0, 2 - a<, Assumptions ® a > 0F

LogBa2

F

IntegrateB1

x - 2, 8x, 2 + a, 3<, Assumptions ® 1 > a > 0F

-Log@aD

LimitBLogBa

2

F - Log@aD, a ® 0F

-Log@2DMathematica has this as the built in option “PrincipleValue”

IntegrateB1

x - 2, 8x, 0, 3<, PrincipalValue ® TrueF

-Log@2DThis also works for other integrals which are improper.

Consider a very simple example:

4 RevisedLT.nb

Page 5: Using Mathematica to evaluate Laplace Transforms and their ...courses.washington.edu/ph227814/228/nb/LT.nb.pdf · Using Mathematica to evaluate Laplace Transforms and their inverses,

Integrate@x, 8x, -¥, ¥<D

Integrate::idiv : Integral of x does not converge on 8-¥, ¥<. �

à-¥

¥

x âx

This divergese because improper integrals are defined as:

à-¥

¥

f HxL â x = lima®¥

à-a

0

f HxL â x +limb®¥

à0

b

f HxL â x

and,

Integrate@x, 8x, 0, b<Db2

2

LimitBb2

2

, b ® ¥F

¥

The principle value in this case is defined by:

lima®¥

à-a

a

f HxL â x

Integrate@x, 8x, -¥, ¥<, PrincipalValue ® TrueD0

The result vanishes because the integrand is odd.

Recap of PV derivation done in class:

Often to do a principle value integral by hand you would use a contour integral, for example

IntegrateBExp@I * xD

x

, 8x, -¥, ¥<, PrincipalValue ® TrueF

ä Π

The contour we would use is:

RevisedLT.nb 5

Page 6: Using Mathematica to evaluate Laplace Transforms and their ...courses.washington.edu/ph227814/228/nb/LT.nb.pdf · Using Mathematica to evaluate Laplace Transforms and their inverses,

ShowB:ListPlot@880, 0<<, PlotStyle ® [email protected],PlotRange ® 88-2, 2<, 8-2, 2<<, AspectRatio ® 1D,

GraphicsB:Text@C, 81.5, 1.5<D, Text@S, 8.25, .25<D, Text@r, 8.09, .13<D,

Text@R, 8-.7, .9<D, LineB:80, 0<, :.25

2

,

.25

2

>>F, LineB:80, 0<, :-2

2

,

2

2

>>F,

Line@88-2, 0<, 8-.25, 0<<D, [email protected], 0<, 82, 0<<D,[email protected], 1<, 81.3, 1.3<<D, [email protected], .1<, 8-.75, .1<<D,Circle@80, 0<, .25, 80, Π<D, Circle@80, 0<, 2, 80, Π<D>F>F

C

S

r

R

-2 -1 1 2

-2

-1

1

2

Where we care about the integral of the part over the real line in the limit as as r ® 0 and R® ¥

No poles are contained in the contour so we know:

limR®¥

r® 0

àC

+ àS

+à-R

-r

+àr

R

= 0

The C integral vanishes by Jordan’s lemma so:

limR®¥

r® 0

à-R

-r

+àr

R

= PV à-¥

¥

= -limr® 0

àS

The integral S is simply half what you would get for a full circle, integrated clockwise (so you get a -

sign) around the pole:

- -äΠ * ResidueBExp@I * zD

z

, 8z, 0<F

äΠ

A semicircle being half the value of the full circle is a lemma to the residue theorem which can be

proven by looking at the Laurent series about the pole and bounding non-pole term.

6 RevisedLT.nb

Page 7: Using Mathematica to evaluate Laplace Transforms and their ...courses.washington.edu/ph227814/228/nb/LT.nb.pdf · Using Mathematica to evaluate Laplace Transforms and their inverses,

A semicircle being half the value of the full circle is a lemma to the residue theorem which can be

proven by looking at the Laurent series about the pole and bounding non-pole term.

To show it explicitly in this case:

z=r eiΘ, dz=i z dΘ

àΠ

0

i ei r e

â Θ

As r ® 0:

àΠ

0

i â Θ = -iΠ

RevisedLT.nb 7


Recommended