0% found this document useful (0 votes)
73 views4 pages

Lecture14 Probset1.8

1) The document discusses solutions to the one-dimensional homogeneous and inhomogeneous wave equations using Fourier transforms. 2) For the homogeneous wave equation, taking the Fourier transform reduces it to an ordinary differential equation that can be solved for the general solution. The inverse Fourier transform gives D'Alembert's solution in terms of delta functions. 3) For the inhomogeneous wave equation, the Green's function is defined as the response to a disturbance at the origin. Taking the Fourier transform of the Green's function and evaluating the integrals yields the causal Green's function solution in terms of delta functions.

Uploaded by

JR
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
73 views4 pages

Lecture14 Probset1.8

1) The document discusses solutions to the one-dimensional homogeneous and inhomogeneous wave equations using Fourier transforms. 2) For the homogeneous wave equation, taking the Fourier transform reduces it to an ordinary differential equation that can be solved for the general solution. The inverse Fourier transform gives D'Alembert's solution in terms of delta functions. 3) For the inhomogeneous wave equation, the Green's function is defined as the response to a disturbance at the origin. Taking the Fourier transform of the Green's function and evaluating the integrals yields the causal Green's function solution in terms of delta functions.

Uploaded by

JR
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Section 14: Solution of Partial Differential Equations; the Wave Equation

14. 1. Solution Methods


The classical methods for solving PDEs are
1. Separation of Variablesidea is to reduce a PDE of N variables to N ODEs. You have
used this method extensively in last year and we will not develop it further here.
2. Integral transform and Green functions method
14. 2. Wave equationDAlemberts solution
First as a revision of the method of Fourier transform we consider the one-dimensional (or
1+1 including time) homogeneous wave equation.
1
utt = 0
<x<
u(x, 0) = a(x) ut (x, 0) = b(x) .
c2
We already saw by the method of characteristics that the general solution is of the form
u = f (x ct) + g(x + ct) and here we could fit the boundary conditions to this expression
(see e.g. Riley, Hobson, Bence 16.4). Instead we start from scratch and use the Fourier
transform. (For systems that are translationally invariant i.e. no boundary conditions at
finite x the F.T. is generally the preferred method.)
uxx

Let

dx u(x, t)eikx .

F (k, t) =

Transforming the wave equation gives


k 2 F (k, t) =

1 2F
c2 t2
F (k, t) = A(k) cos(kct) + B(k) sin(kct)

The boundary conditions become, with e


a(k) = F [a(x)] and eb(k) = F [b(x)]
F (k, 0) e
= b(k) = kcB(k)
t
eb(k)

F (k, t) = e
a(k) cos(kct) +
sin(kct)
kc
We invoke the convolution theorem to deduce


1
1 sin(kct)
u(x, t) = a(x) F [cos(kct)] + b(x) F
kc
F (k, 0) = e
a(k) = A(k)

Now we recall integral representation of the dirac delta and Heaviside functions to deduce
Z
 1
1
dk eikx  ikct
1
F [cos(kct)] =
e + eikct = [(x + ct) + (x ct)]
2 2
2


Z

1
dk eikx  ikct
1
1 sin(kct)
F
=
e eikct =
[(x + ct) (x ct)]
kc
4i ck
2c
64

Then we can carry out the convolutions to obtain


Z
1
1
u(x, t) =
[a(x ct) + a(x + ct)] +
dy b(x y) [(y + ct) (y ct)]
2
2c
Z
1 x+ct 0 0
1
=
[a(x ct) + a(x + ct)] +
dx b(x )
2
2c xct
The terms in the square bracket represents the sum of the initial displacement a(x) translated
by ct to the left and ct to the right. The integral represents the cumulative effect of the
initial velocity profile g(x). To see this last point more clearly let us consider the example
a(x) = 0

b(x) = (x)

which we can think of as a unit impulse at x = t = 0 e.g. a plucked guitar string. Then we
obtain
( 1
for ct < x < ct
1
2c
u(x, t) = [(x + ct) (x ct)] =
2c
0
otherwise
Thus the effect of the impulse at x = 0 propagates to the right and left with speed c creating
a characteristic cone or domain of influence delineated by the two characteristics x = ct.
You should sketch this

Figure 1: Domain of influence for an impulse at the origin


Aside: It is interesting to note that there is nothing to stop us considering t < 0 then our
solution gives
( 1
2c for c|t| < x < c|t|
1
u(x, t) = [(x + c|t|) (x c|t|)] =
2c
0
otherwise
Thus somehow we have a negative signal propagating into the past with speed c a wave
equation admits solutions propagating backwards in time!
When playing with the (classical) wave equation and guitar strings one is not usually interested in backwards time travel so t < 0 is not usually considered. Later we shall see how
this solution has stemmed from the way we carried out the inverse Fourier transform and in
particular our integral representation of the step function.

65

14. 3. Inhomogeneous wave equation


We now consider the inhomogeneous wave equation in three (spatial) dimensions
2 u(x, t)

1 2 u(x, t)
= f (x, t) .
c2 t2

We define the Green function G(x, t; x0 , t0 ) as a function which satisfies


1 2 G(x, t; x0 , t0 )
= (x x0 )(t t0 )
(1)
t2
c2
where the Laplacian and time derivative act on the unprimed variables. As usual we construct
a particular solution by convolving the external influence f with the Green function:
Z
u(x, t) = d3 x0 dt0 f (x0 , t)G(x, t; x0 , t0 ) .
2 G(x, t; x0 , t0 )

Now clearly G is a function of x x0 and t t0 therefore we may without loss of generality


take x0 = t0 = 0 in (1) and recover the general case at the end of the calculation.
Thus G(x, t) is the response to a disturbance at x = 0, t = 0.
We begin to solve for G(x, t) by taking the Fourier transform in space and time
Z
Z
3
e
G(k, ) = d x dt G(x, t)ei(k xt)
note the common convention that the Fourier variable , conjugate to t, is opposite sign to
k the variable conjugate to x.
Transforming (1) gives


2
k + 2
c
2


e ) = 1
G(k,

c2
G(k, ) = 2
k 2 c2
We are left to invert the F.T.
Z
G(x, t) =

d3 k
(2)3

d
c2
ei(k xt) .
2 2 k 2 c2

First let us do some work on the integral over k-space


Z
d3 k
eik x
(2)3 2 k 2 c2
We go into spherical polar co-ordinates in k-space. Since x is held constant here it is just
some vector and we have the freedom to choose the e3 -axis for our k-space spherical polars
along the x direction. Then k x = kr cos and we have
Z
dk d d k 2 sin ikr cos
e
where r = |x|
(2)3 2 k 2 c2
 ikr cos 
Z
Z
2
k2
e
1
k
eikr
=
dk 2

=
dk
(2)3 0
k 2 c2
ikr 0
(2)2
ir 2 k 2 c2
66

Thus we have performed the angular integration over , and we are left with
Z
Z
1 c2
eit
ikr
G(x, t) =
.
dk
ke
d
(2)3 ir
( kc)( + kc)

In the integral there are poles at = c|k| and = c|k|. See figure

Figure 2: Poles in complex plane in calculation of the Green function


For t > 0 we close the contour by a big semicircle in the lower half plane (because of eit )
For t < 0 we close the contour by a big semicircle in the upper half plane.
In order to proceed we have to decide how to bypass the poles on the integration contour.
What to do will be specified by the boundary conditions of the problem (recall that Green
functions always depend on the boundary conditions).
First note that including the poles within a closed integration contour will result in residues
from the integral proportional to eik(rct) which lead to contributions (r ct) in G.
Now recall that we want G to be the response to a disturbance at x = 0 t = 0. i.e. we want
causality to hold and G = 0 for t < 0. This implies that the contour must bypass the poles
by going above them so that when t < 0 they are not inside the closed contour.
However the closed contour for t > 0 (which is in a clockwise sense) will include the poles
and we obtain by the residue theorem
 ikct

Z
2i c2
eikct
1
ikr e
t>0
G(x, t) =
dk
ke

=
[(r + ct) (r ct)]
(2)3 ir
2kc
2kc
4cr
Because t > 0 r > 0 the first delta becomes irrelevant and finally, putting back in x0 t0 , we
have
(
0
t < t0
G(x x0 , t t0 ) =
4|xcx0 | (|x x0 | c|t t0 |) t > t0

67

You might also like