Professional Documents
Culture Documents
MAP 5345
Partial Differential Equations
PROJECT I
RAMIN SHAMSHIRI
Spring 2009
1.
= ,
, = ,
, = ,
(1.1)
+ +
(1.2)
Solution:
The PDE in (1.1) can be solved by either d'Alembert's solution, using a Fourier transform method or by applying
method of separation of variables. DAlemberts solution of (1.1) is given by , = + + ( ),
where ,: are arbitrary twice differentiable functions. The dAlemberts solution to this PDE is presented
1
below to show that the initial conditions in (1.1) imply that = + and =
Rearranging (1.1) as
=
=
c2
+
+
(1.3)
(1.4)
+2
2
2
2
1
2
2
=
=
2
+
c2
2
2
=
1
c2
indicates that
(1.5)
(1.6)
, = + + ( )
(1.7)
(1.8)
(1.9)
= +
(1.10)
= = ( )
1
2
1
2
1
2
1
(1.11)
(1.12)
Plugging (1.11) and (1.12) into (1.7) yields the PDE solution with the given initial conditions:
1
1
1
, = + + +
2
2
2 +
(1.13)
Since both PDE and boundary conditions are linear and homogenous, the method of separation of variables is also
attempted to show that the solution of the PDE can be written in the for m of (1.2).
, = .
2
= . 2
2
= . 2
. = 2 . ()
1
1 2
.
=
.
=
2 () 2
2
2
= 2 ()
1 = 0
2
= 0,
=
, = 1,2,3,
= 2
There are two families of product solutions; the principle of superposition implies that we should be able to solve
the initial value problem by considering a linear combination of all product solutions:
, =
=1
(1.14)
() =
=1
() =
=1
(1.15)
(1.16)
=
sin
0
=
sin
0
Using the sum to product of sine and cosine ( . = [ + + ( )]), equation (1.14) can
2
be written as:
, =
=1
1
2
+ +
=1
(1.19)
+ =
sin
=1
sin
=1
(1.20)
(1.21)
=1
1
2
= 2
=1
cos
sin
=
=1
cos
=1
sin
[ ( + )]
sin
=1
sin
=1
. sin
. sin
1
2
Equation (1.20) can now be written as the combination of equations (1.20), (1.21) and (1.22):
1
1
1 +
, = + + +
2
2
2
(1.22)
(1.23)
(1.24)
Using dAlemberts formula to find a solution to the initial value problem in (1.1), we need to choose the functions
and such that (1.1) is satisfied.
, 0 = + = ()
, 0 = () = ()
(1.25)
(1.26)
(1.27)
= +
+
2
2
2
1
1
=
2
2
2
Using these functions in (1.25) will give us the result:
, =
1
2
+ +
1
2
or
(1.28)
, =
1
2
+ +
1
2
Suppose that c=1, = if , = otherwise, and g=0. Plot snapshots of the string
at = , . , . , . , , , , , and verbally describe the motion.
Solution:
1 ,
0
0,
Assigning different values for t0 give the plot of snapshots of the string at different t. These plots are shown in the
following page. The plots show a wave at its maximum magnitude equal to 1 at t 0 . The magnitude of wave reduces
and the wave forms two traveling wave going away from each other with speed 2c and each with magnitude 0.5.
It can also be seen that the magnitude of the two waves remains constant when they are moving and start to add
up when they reach each other. In the other words, the energy of the wave remains as time goes by and the
motion continue forever. From the solution, it can be seen why the waves are traveling to the right and left. This is
because of the replacement of with + and which will shift the function to the right and left.
Snapshots at t=[0,0.25,0.5,0.75,1,4,7,9]
Snapshot at t=0
Snapshot at t=0.25
Snapshot at t=0.5
Snapshot at t=0.75
Snapshot at t=1
Snapshot at t=2
Snapshot at t=3
Snapshot at t=4
Snapshot at t=5
Solution:
The below Maple program was used to solve the problem.
> res tart: with(plots ):
> (1 /c ^2 )* u[tt]=u[xx];
> u(x,0 )=f(x); u[t](x,0 )=g(x);
> u(x,t) =(f(x- c* t)+f(x+c* t))/2 +(1 /(2*c ))*int(g(y),y=x-c* t..x+c*t);
> c :=1 ; f(x):=0 ; g(x):=piec ewise(abs (x)<=1 ,1 ,0 );
>`C omputing u(x,t)...`;
f(x):=s implify(f(x)): g(x):=s implify(g(x)): G (x):=int(g(x),x):
f0 :=unapply(f(x),x): G 0 :=unapply(G (x),x): u:=(x,t)- >(f0 (x+c* t)+f0 (x-c* t))/2 +(G0 (x+c*t)-G0 (x-c*t))/(2*c ):
`Finis hed.`;
>t0 :=0 :
plot(u(x,t0 ),x=-5 ..5 ,0 ..1 ,numpoints =250 ,
title=cat(`Snaps hot at `,c onvert('t'=t0 ,s tring)));
Assigning different values for t0 give the plot of snapshots of the string at different t. These plots are shown in the
following page. The initial magnitude of the wave is zero, but wave starts being built by the speed of 1 up
the maximum value of 1. After that point, the wave starts to travel and become wider and wider.
Snapshots at t=[0,0.5,1,2.5,5,7.5,9]
Snapshot at t=0
Snapshot at t=0.25
Snapshot at t=0.5
Snapshot at t=0.75
Snapshot at t=1
Snapshot at t=2
Snapshot at t=3
Snapshot at t=4
Snapshot at t=5
10
2.
Let = (, ) denote the temperature in a homogenous rod of length L with perfectly insulated
lateral surface. Assume that the ends of the rod at = and = are held temperature zero, and the
initial temperature distribution is a given function = (). Then u solves the initial boundary value
problem;
= > ,
< < ,
, = , = > 0 ,
, = < < ,
(2.1)
(2.2)
(2.3)
Solution:
, =
, ,
0 < < , > 0,
2
0, = , = 0,
> 0,
, 0 = ,
0 < < .
(2.4)
(2.5)
(2.6)
We expect that equations (2.4)-(2.6) completely and unambiguously specify the temperature in the rod. Once we
have found a function (, ) that meets all three of these conditions, we can be assured that is the
temperature. One classical technique for solving the initial -boundary value problem for the heat equation (2.4)(2.6) is the method of separation of variables which allow us to replace the partial derivatives by ordinary
derivatives. The idea is to think of a solution (, ) to a partial differential equation as being an infinite linear
combination of simple component function , , = 0,1,2, , which also satisfy the equation and certain
boundary conditions. To deter mine , , we assume it can be written with its variable separated as:
, = . .
Substituting this into the PDE and using boundary conditions leads to two ODE for the unknown functions
and .
, = .
= ()
2
= ()
2
Substituting these into (2.4) and separating variables gives:
= ()
Since x and t are variables that are independent of one another, the two ratios must be constant.
=
=
11
(2.7)
(2.8)
= 0,
&
> 0.
0 = =0
(2.9)
We can write the differential equation in (2.9) as ( 2 ) = 0 which is similar with the matri x eigenvector
equation = . The function 0 0 is a solution of (2.9) for every . To find solution, we must first
determine those values of for which the boundary value problem in (2.9) has nontrivial solution. These specific
values are called eigenvalues and the corresponding nontrivial solutions to (2.9) are the eigenfunctions. To solve
the equation in (2.9), we try = , derive the auxiliary equation 2 = 0. Three cases can be considered:
Case 1: > 0
= , so a general solution to (2.9) is:
= 1 + 2
0 = 1 + 2 = 0
= 1 + 2
Since > 0, ( 2
1 = 2
= 1( 2 1) = 0
Case 2: = 0
= 0 is a repeated root to the auxiliary equation, so a general solution to (2.9) is:
= 1 + 2
0 = 1 + 2 = 0
= 1 + 2 = 0
1 = 0
2 = 0
= 1 + 2 ,
0 = 1 + 2 = 0
= 2 = 0
1 =
(2.10)
, = 1,2,3,
The corresponding eigenfunctions to the eigenvalues are the nontrivial solution to (2.9) and are given by:
=
,
(2.11)
12
=0
For each n=1,2,3,, a general solution to this linear first order equation is:
=
(2.12)
(2.13)
Since / 2 / 2 is a linear operator and the boundary conditions in (2.5) are homogenous, we can
say that if and are solution to (2.4)-(2.5), then so is any linear combination + . Taking an infinite
sum of (2.13) and replacing with its equivalent
, =
, =
=1
=1
(2.14)
, 0 =
=1
= ()
(2.15)
Equation (2.15) is a Fourier since series. According to the definition of Fourier sine series of f(x) on [0,L], the
coefficient is given by:
=
.
0
, = 1,2,
(2.16)
Equation (2.14) with the coefficient in (2.16) is the solution to (2.1)-(2.3). This solution is claimed to be unique.
From the theor em of uniqueness of solution, the initial -boundary value problem in (2.4)-(2.6) has at most one
continuously differentiable solution. This can be proved by assuming (, ) and (, ) are continuously
differentiable functions that satisfy this initial-boundary value problem. Let = . Now is continuously
differentiable solution to the boundary value problem (2.4)-(2.5). By the maximum principle, must attain its
maximum at t=0, and since
, 0 = , 0 , 0 = = 0
We have , 0. Hence (, ) (, ) for all 0 , 0. A similar argument using =
yields (, ) (, ). Therefore we have , = (, ) for all 0 , 0. Thus there is at most one
continuously differentiable solution to the problem (2.4)-(2.6).
13
Solution:
The Fourier coefficient is given by (2.16). Substituting with 1 in the formula will result:
=
=
1 1
= ,
0,
= 1,2,3,
(2.17)
Therefore the solution to the PDE (2.1)-(2.3) with f(x)=1 is given by:
, =
1 ()
2
=1
2 2
(2.18)
k 2u
c x2
This answer can be verified. For the sake of simplicity, let = 1. If Equation (2.18) is a solution to
u
t
=0
then:
2
2
=1
1 ()
1 ()
2 2 ( ( ]
2 2 ( 2 2 ( ]
(2.20)
[ 2 2 ( 2 2 )]()
(2.21)
=0
1 ()
2
=0
(2.19)
( 2 2 )
1 ()
=0
2 2
=0
2 2 (
1 ()
2
2 2 () = 0
14
Plot a sequence of snapshots of an approximate solution obtained by suitably truncating the Fourier
series.
Solution:
, =
1 ()
2
=0
2 2
(2.22)
First of all, from physical interpretation of the problem, the boundary conditions state that temperature,
regardless of the change in time, at the two ends of the rod is zero. It means that at initial condition which requires
f=1, we have zero temperature at the endings. This is a very ideal theoretical assumption which require perfect
isolation of the materials at those points because according to the initial condition, the temperature at t0 =0 should
be equal to 1 for any x in 0 < < . In the other words, the temperature should be zero at x=0 and suddenly
becomes 1 at x=0+ or x=L-
Based on the rules of nature, any step change will always result overshoots and corresponding undershoots.
Mathematically this is called Gibbs phenomenon and occurs only when a finite series of eigenfunctions
approximates a discontinuous functions. In general, there is an overshoot and undershoot of approximatel y 9% of
the jump discontinuity.
If we consider the temperature at the middle of the rod at time zero, it would make sense if the temperature
converges to f=1 as n increases in (2.22). If the length of the rod is assumed L=1, then at x =0.5, we will have:
( )
2
2
=0
4 1 (0)
= 0 =>
0
0.5,0 =
1 ()
= 1 =>
= 3 =>
= 5 =>
= 7 =>
0.5,0 =
[1
1
3
4 1 ()
0.5
4 1 (3)
4 1 (5)
10
4 1 (7)
14
1
5
1
7
=0
=
3 0.5
5 0.5
7 0.5
4 1
3
4 1
5
4 1
7
. [1] =
. [+1] =
. [1] =
+]
15
Using Delphi programming, for the first 36 terms, the series will be equal to 0.9912 and therefore shows
convergence. This is shown as below:
0.5,0 =
1
3
7
1
1
9
1
11 13 15 17 19 21 23 25 27 29 31 33 35 37
1
1
1
1
1
1
1
1
1
1
1
1
1
1
39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71
1
+ = 0.9912
If we divide the rod into 10 parts such that we can measure temperature at x = 0.1, using the procedure of Riley
et al. (1997, p.952) for Thomas algorithm, we will have I = 9. Choosing
x 2
k t
= 0 ,+1 + 0, + 1, + 2, = 0 + 0 + 0 + 1 = 1
= 1, + 2, + 3, = 1 + 0 + 1 = 2
= 2, + 3, + 4, = 1 + 0 + 1 = 2
= 3, + 4, + 5, = 1 + 0 + 1 = 2
= 4, + 5, + 6, = 1 + 0 + 1 = 2
= 5, + 6, + 7, = 1 + 0 + 1 = 2
= 6, + 7, + 8, = 1 + 0 + 1 = 2
= 7, + 8, + 9, = 1 + 0 + 1 = 2
= 8, + 9, + 10, + 10 ,+1 = 1 + 0 + 0 + 0 = 1
4 1
1 4
0 1
0
0
0
0
0
0
0
0
0
0
0
0
1 =
=
1
1
0
1
4
1
0
0
0
0
0
0
0
0
0
1 0
4 1
1 4
0 1
0
0
0
0
0
0
0
0
0
0
1
4
1
0
0
0
0
0
0
0
0
0
0
0
0
1 0
4 1
1 4
0 1
1 =
1 +
1
0
2
0
3
0
4
0
0 . 5
6
0
7
0
1 8
9
4
+1
1
2
3
4
= 5
6
7
8
9
1
1
=
1 +
1 = 1 + 1
=
Using Delphi, this algorithm was implemented for different values of j in excel and results are shown in tables
presented in the two following pages.
16
j=1
0
-1
-1
-1
-1
-1
-1
-1
-1
4
4
4
4
4
4
4
4
4
-1
-1
-1
-1
-1
-1
-1
-1
0
j=2
0
-1
-1
-1
-1
-1
-1
-1
-1
4
4
4
4
4
4
4
4
4
-1
-1
-1
-1
-1
-1
-1
-1
0
j=3
0
-1
-1
-1
-1
-1
-1
-1
-1
4
4
4
4
4
4
4
4
4
-1
-1
-1
-1
-1
-1
-1
-1
0
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
j=4
0
-1
-1
-1
-1
-1
-1
-1
-1
j=5
0
-1
-1
-1
-1
-1
-1
-1
-1
+
3.75
3.733333333
3.732142857
3.732057416
3.732051282
3.732050842
3.73205081
3.732050808
+
3.75
3.733333333
3.732142857
3.732057416
3.732051282
3.732050842
3.73205081
3.732050808
+
3.75
3.733333333
3.732142857
3.732057416
3.732051282
3.732050842
3.73205081
3.732050808
+
-1
-1
-1
-1
-1
-1
-1
-1
0
3.75
3.733333333
3.732142857
3.732057416
3.732051282
3.732050842
3.73205081
3.732050808
+
-1
-1
-1
-1
-1
-1
-1
-1
0
3.75
3.733333333
3.732142857
3.732057416
3.732051282
3.732050842
3.73205081
3.732050808
0.25
0.266666667
0.267857143
0.267942584
0.267948718
0.267949158
0.26794919
0.267949192
0
1
2
2
2
2
2
2
2
1
0.25
0.266666667
0.267857143
0.267942584
0.267948718
0.267949158
0.26794919
0.267949192
0
0.856354
1.425414
1.845304
1.955801
1.977901
1.955801
1.845304
1.425414
0.856354
0.25
0.266666667
0.267857143
0.267942584
0.267948718
0.267949158
0.26794919
0.267949192
0
0.667898
1.246177
1.615152
1.833216
1.894509
1.833216
1.615152
1.246177
0.667898
0.25
0.266666667
0.267857143
0.267942584
0.267948718
0.267949158
0.26794919
0.267949192
0
0.581897
1.081411
1.460699
1.681991
1.757604
1.681991
1.460699
1.081411
0.581897
0.25
0.266666667
0.267857143
0.267942584
0.267948718
0.267949158
0.26794919
0.267949192
0
0.512339
0.967946
1.316847
1.536041
1.609016
1.536041
1.316847
0.967946
0.512339
-0.25
-0.6
-0.69643
-0.72249
-0.72949
-0.73136
-0.73187
-0.732
-0.21409
-0.4372
-0.61139
-0.68786
-0.71429
-0.71545
-0.68615
-0.56579
-0.16697
-0.37684
-0.53357
-0.63416
-0.67755
-0.67276
-0.61304
-0.49818
-0.14547
-0.32717
-0.47889
-0.57899
-0.62609
-0.61845
-0.55711
-0.43904
-0.12808
-0.29227
-0.43101
-0.52706
-0.57236
-0.56494
-0.50422
-0.39447
0.25
0.6
0.696429
0.722488
0.729487
0.731364
0.731867
0.732001
0.464088
0.464088
0.856354
0.961326
0.98895
0.994475
0.98895
0.961326
0.856354
0.464088
0.214088
0.437201
0.611385
0.687859
0.714287
0.715448
0.686151
0.565792
0.381063
0.381063
0.667898
0.865114
0.947254
0.968102
0.947254
0.865114
0.667898
0.381063
0.166974
0.37684
0.533569
0.634163
0.677554
0.672759
0.613044
0.498177
0.312449
0.312449
0.581897
0.768963
0.878802
0.913028
0.878802
0.768963
0.581897
0.312449
0.145474
0.32717
0.478893
0.578993
0.626088
0.618448
0.557106
0.439039
0.273559
0.273559
0.512339
0.694387
0.804508
0.841655
0.804508
0.694387
0.512339
0.273559
0.128085
0.292275
0.431015
0.527058
0.572358
0.564944
0.504224
0.394467
0.242978
0.242978
0.459573
0.627367
0.733046
0.768777
0.733046
0.627367
0.459573
0.242978
17
j=6
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
0
-1
-1
-1
-1
-1
-1
-1
-1
j=10
0
-1
-1
-1
-1
-1
-1
-1
-1
4
4
4
4
4
4
4
4
4
0
-1
-1
-1
-1
-1
-1
-1
-1
j=7
0
-1
-1
-1
-1
-1
-1
-1
-1
j=8
0
-1
-1
-1
-1
-1
-1
-1
-1
j=9
+
-1
-1
-1
-1
-1
-1
-1
-1
0
3.75
3.733333333
3.732142857
3.732057416
3.732051282
3.732050842
3.73205081
3.732050808
+
-1
-1
-1
-1
-1
-1
-1
-1
0
3.75
3.733333333
3.732142857
3.732057416
3.732051282
3.732050842
3.73205081
3.732050808
+
-1
-1
-1
-1
-1
-1
-1
-1
0
3.75
3.733333333
3.732142857
3.732057416
3.732051282
3.732050842
3.73205081
3.732050808
+
-1
-1
-1
-1
-1
-1
-1
-1
0
3.75
3.733333333
3.732142857
3.732057416
3.732051282
3.732050842
3.73205081
3.732050808
+
-1
-1
-1
-1
-1
-1
-1
-1
0
3.75
3.733333333
3.732142857
3.732057416
3.732051282
3.732050842
3.73205081
3.732050808
0.25
0.266666667
0.267857143
0.267942584
0.267948718
0.267949158
0.26794919
0.267949192
0
0.459573
0.870345
1.192619
1.396144
1.466093
1.396144
1.192619
0.870345
0.459573
0.25
0.266666667
0.267857143
0.267942584
0.267948718
0.267949158
0.26794919
0.267949192
0
0.414252
0.786665
1.080217
1.267714
1.331929
1.267714
1.080217
0.786665
0.414252
0.25
0.266666667
0.267857143
0.267942584
0.267948718
0.267949158
0.26794919
0.267949192
0
0.374683
0.712068
0.979121
1.150035
1.208875
1.150035
0.979121
0.712068
0.374683
0.25
0.266666667
0.267857143
0.267942584
0.267948718
0.267949158
0.26794919
0.267949192
0
0.33931
0.645172
0.887575
1.043025
1.09653
1.043025
0.887575
0.645172
0.33931
0.25
0.266666667
0.267857143
0.267942584
0.267948718
0.267949158
0.26794919
0.267949192
0
0.307484
0.584765
0.804692
0.945803
0.994414
0.945803
0.804692
0.584765
0.307484
-0.11489
-0.26273
-0.38983
-0.47854
-0.52106
-0.51371
-0.45721
-0.35572
-0.10356
-0.23739
-0.35293
-0.43424
-0.47324
-0.46649
-0.41444
-0.32183
-0.09367
-0.21486
-0.31982
-0.39384
-0.42944
-0.42322
-0.37576
-0.29148
-0.08483
-0.19467
-0.28989
-0.35714
-0.38951
-0.38385
-0.34068
-0.26416
-0.07687
-0.17644
-0.2628
-0.32384
-0.35322
-0.34807
-0.30888
-0.23945
0.114893
0.26273
0.389826
0.478537
0.521061
0.513713
0.45721
0.355717
0.218456
0.218456
0.414252
0.568209
0.665965
0.699505
0.665965
0.568209
0.414252
0.218456
0.103563
0.237394
0.352932
0.43424
0.473243
0.466488
0.414438
0.321835
0.197234
0.197234
0.374683
0.514834
0.604437
0.635201
0.604437
0.514834
0.374683
0.197234
0.093671
0.214864
0.319817
0.393836
0.429444
0.42322
0.375756
0.291482
0.178498
0.178498
0.33931
0.466674
0.548265
0.576351
0.548265
0.466674
0.33931
0.178498
0.084828
0.194667
0.289886
0.357144
0.38951
0.383847
0.340677
0.264157
0.161699
0.161699
0.307484
0.423067
0.497207
0.522736
0.497207
0.423067
0.307484
0.161699
0.076871
0.176436
0.262802
0.323837
0.353224
0.348073
0.308882
0.239452
0.146551
0.146551
0.27872
0.383565
0.450849
0.474028
0.450849
0.383565
0.27872
0.146551
18
x
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
j=1
0
0.464088
0.856354
0.961326
0.98895
0.994475
0.98895
0.961326
0.856354
0.464088
0
j=2
0
0.381063
0.667898
0.865114
0.947254
0.968102
0.947254
0.865114
0.667898
0.381063
0
j=3
0
0.312449
0.581897
0.768963
0.878802
0.913028
0.878802
0.768963
0.581897
0.312449
0
j=4
0
0.273559
0.512339
0.694387
0.804508
0.841655
0.804508
0.694387
0.512339
0.273559
0
j=5
0
0.242978
0.459573
0.627367
0.733046
0.768777
0.733046
0.627367
0.459573
0.242978
0
j=6
0
0.218456
0.414252
0.568209
0.665965
0.699505
0.665965
0.568209
0.414252
0.218456
0
j=7
0
0.197234
0.374683
0.514834
0.604437
0.635201
0.604437
0.514834
0.374683
0.197234
0
j=8
0
0.178498
0.33931
0.466674
0.548265
0.576351
0.548265
0.466674
0.33931
0.178498
0
j=9
0
0.161699
0.307484
0.423067
0.497207
0.522736
0.497207
0.423067
0.307484
0.161699
0
j=10
0
0.146551
0.27872
0.383565
0.450849
0.474028
0.450849
0.383565
0.27872
0.146551
0
19
To plot a sequence of snapshots, equation (2.18) was programmed in Delphi. Results were then plotted in
MATLAB. = 0.002 and = 50 was used in the calculations.
proc edure T Form1 .Button1Clic k(Sender: TO bject);
var L ,beta,c n,en,t,s in_n,x,fx:real;
n,x_inc :integer;
label s tart;
begin
memo1 .C lear;
//D efine L and beta here
L :=1 ; beta:=s trtofloat(edit2 .T ext);
//D efine time s naps hot here
t:=s trtofloat(edit1 .T ext);
s tart:
fx:=0 ;
for n:=1 to s pinedit1 .Value do
begin
fx:=fx+( (2 /(n* pi))* (1-c os (n*pi))* (exp(- (beta* n* n*pi* pi* t/(L*L ))))* (sin(n* pi*x/L )));
end;
memo1 .Lines .Add(floattostr(fx));
x:=x+(L /500 );
if x<=L then goto start;
end;
end.
20
What can you say about the solutions behavior as ? Does this behavior depend on the initial
data?
Answer:
It can be seen from the plot of the solution that , (, ) starts decreasing rapidly with time. Since the temperature
at the boundaries is zero, it is expected that the temperature of the rod at any point x in (0,L) becomes equal to
the temperature at the boundaries as time goes to infinity. To answer that whether this behavior depend on the
initial data, solution in (2.18) was plotted for different values of = , and for t=0.001 and t=0.1 as shown in
figure (2.3) and (2.4). It can be seen that (, ) decreases for the smaller values of . Therefore the shape of the
solution depends on the value of , however the behavior is still same and (, ) starts vanishing away with time.
21
We can also change the initial condition; say = 2 to see how the shape of the solution changes. It can be
seen that the behavior is still same and (, ) decreases with time.
= ,
0, = ()
2
= 1, () =
, =
, =
( ) ()
2
4
4
Snapshot at t=[0,0.1,0.5,2.5,10]
= 2
22
3.
(3.1)
subject to , = , = and , = () is not well-posed. Show that if the data are changed an
arbitrary sm all amount, for example,
, =
, =
=1
=1
, = 1,2,
(3.2)
(3.3)
To show that the solution will change by large amount if data are changed by small amount, we can compare the
1
Let k=-1, L=1, with u x, 0 = f x = 1 , the solution in (3.2)-(3.3) can be written as:
, =
=1
1 ()
2
22
(3.4)
23
24
n x
below:
=
=
=
sin
).
, = 1,2,
+
sin2
0
0
1
1
1 +
2
.
0
(1 +
1 +
, = 1,2,
(3.5)
=1
2 1 +
22
(3.6)
Discussion:
1
For large , +
is equal to + , since > 0,
1
1 <
1, thus
< 1. Comparing plot of equation (3.4) with the plot of equation (2.22), we can see that for same
initial condition, the magnitude of (3.4) is larger which is due to its positive exponential term (
2 2
lim = ) which makes the system to oscillate and become unstable for small change in the input. It can
be seen from Figure (3.3) that the magnitude of solution for the same time snap is almost 37 for = 1 +
while the corresponding magnitude for = 1 is almost 1.5. This can also be seen for larger ts. For t=0.01, we
see a big change in the magnitude of solution, from 450 in Figure (3.2) to 27000 in Figure (3.4).
25
26
4.
Let be a positive number and let and be nonnegative numbers, not both equal to zero. Find
eigenvalues and a complete orthogonal system of eigenfunctions for BVP
+ = ,
= ,
+ = .
(4.1)
Solution:
Solving the characteristic equation of BVP in (4.1):
2 + = 0
1 ,2 =
General solution:
= 1 + 2
(4.2)
(4.3)
(4.4)
From the problem statement, and are nonnegative numbers, not both equal to zero. We will have three cases
associated with equation (4.4).
Case 1: > 0 , > 0
In this case, equation (4.4) can be written as:
tan =
Let
(4.5)
=
tan =
(4.6)
The eigenvalues are therefor e the roots of equation (4.6). These eigenvalues are the values for which 1 = tan
intercepts with 2 = . Plot of y1 and y2 for different values of c is shown in figure 4.1.
27
(4.7)
Case 2: > 0 , = 0
In this case, equation (4.4) can be written as:
2 = 0
(4.8)
Since 2 cannot be zero in order to get nontrivial solution, should be set to zero. Ther efore the eigenvalues
for this case are equal to:
=
= 0,1,2,
(4.9)
28
Case 3: = 0 , > 0
In this case, equation (4.4) can be written as:
2 = 0
(4.10)
Since 2 cannot be zero in order to get nontrivial solution, should be set to zero. Ther efore the eigenvalues
for this case are equal to:
=
2 + 1
,
= 0,1,2,
2
The sets of eigenfunctions corresponding to these eigenvalues are then:
2 + 1
=
(4.11)
29
Find numerical approxim ations of the first ten eigenvalue s. What can be said in general about the
number and approximate location of the eigenvalues?
Solution:
Case 1: > 0 , > 0
Since solution to equation (4.4) depends on c, if we fix c=1, we will have:
tan =
(4.5)
Letting = , then = 2 . An iteration algorithm was written in Delphi to calculate roots of the equation
tan = in the interval [7, 7]. Results are given in table 4.1 and plotted in Figure 4.2.
X=roots
-17.33607346
-14.20792654
-11.08592654
-7.979926536
-4.916926536
-2.044926536
0
2.013073464
4.910073464
7.978073464
11.08507346
14.20707346
17.33607346
20.46907346
=
17.42867
14.10876
11.03766
7.897954
4.820708
1.948661
0
-2.11164
-4.9925
-8.01714
-11.1434
-14.2815
-17.4287
-20.5087
=
17.33607
14.20793
11.08593
7.979927
4.916927
2.044927
0
-2.01307
-4.91007
-7.97807
-11.0851
-14.2071
-17.3361
-20.4691
error
0.092601501
-0.099161879
-0.048267151
-0.081972649
-0.096218943
-0.096265094
0
-0.098567532
-0.082425351
-0.039068311
-0.058364692
-0.074433508
-0.092601501
-0.039622929
=
300.5394
201.8652
122.8978
63.67923
24.17617
4.181725
0
4.052465
24.10882
63.64966
122.8789
201.8409
300.5394
418.983
30
n
0
1
2
3
4
5
6
7
8
9
10
> 0, = 0
=
0 = 0
2
1 =
2 2
2 =
3 2
3 =
4 2
4 =
5 2
5 =
6 2
6 =
7 2
7 =
8 2
8 =
9 2
9 =
10 2
10 =
= , > 0
+
=
0 = 0
3 2
1 =
2
5 2
2 =
2
7 2
3 =
2
9 2
4 =
2
11 2
5 =
2
13 2
6 =
2
15 2
7 =
2
17 2
8 =
2
19 2
9 =
2
21 2
10 =
2
and =
2 +1
31
5.
Using the maximum principle for Laplace's equation, prove that the solution of Poisson's equation
= (), subject to = () on the boundary is unique.
Review:
The Laplaces equation is of the form:
u=0
(5.1)
where is the Laplace operator and is defined as the divergence of the gradient (dot product of with itself ),
thus, equation (5.1) can also be written as:
u = 2 = . = 0
(5.2)
The Laplace equation in 3 dimensions is written as:
2 2 2
u= 2+ 2+ 2 =0
(5.3)
The Poissons equation in 2 dimensions is written as:
2 2
+
= (, )
(5.4)
2 2
In polar coordinates, equation (5.4) will be written as:
2 1
1 2
+
+
= (, )
(5.5)
2 2 2
Solution:
According to the maximum principle, if (, , ) is a solution to Laplaces equation in a bounded domain with
(, , ) continuous in , then (, , ) attain its maximum value on .( = , where denotes the
boundary of ).
Solutions to Laplaces equation in two variables are called harmonic functions. The uniqueness of the solution
follows from the maximum principle. Just like the heat equation, (, ) attains its maximum value at initial
condition, = 0 for some in [0, ]:
, 0 = (, 0)
For the Poisson Equation, =
2
2
2
2
are continuously differentiable functions that satisfy the equation with initial conditions 1 = 1 () and
2 = 2 (), then a combination, = 1 (, ) 2 (, ) is also continuously differentiable solution to the
problem and by the maximum principle, hence must attain its maximum at the initial condition , 0 .
, 0 = , 0 , 0 = 1 2 = 0
The r esult shows that , 0. Hence (, ) (, ) for all 0 , 0. Therefor e 1 , =
2 (, ) for all 0 , 0.
32
6.
Write a technical report about the qualitative differences between wave propagation, as governed by the wave
equation, and diffusion processes, as governed by the heat equation. Address issues such as finite vs. infinite
speed propagation, propagation of singularities vs. instantaneous smoothing, conservation of wave profile vs.
averaging, and time reversibility vs. irreversibility.
Support your arguments with examples and graphical evi dence from suitable computer experiments.
Answer:
Wave propagation is described by wave equation which is a hyperbolic partial differential equation in the form of:
= 2
(6.1)
where c is the speed of propagation. Equation (6.1) is one-dimensional wave equation.
Heat distribution is described by heat equation which is an example of parabolic equation and is written as below
in three dimensions:
= k2 u = k( + + ) , ,
(6.2)
Where k is a constant and is related to material property.
will have two parts. The first part of the solution consists of sinusoidal function sin multiplied by a time
varying amplitude. The second term of the solution is also sinusoidal function multiplied by sinusoidal time-varying
amplitude which will present a node in = that never moves. For the n-th term, we will have (n-1) nodes. The
2
1
2
+ +
1
2
(1.2)
33
Figure 6.1. Shifting sin(x) to right and left, sin(x+h) and sin(x-h)
In wave equation, the parameter t is the time. Therefore the function ( + ) will represent sets of functions
that have same shape, but shifted more and mor e to the left as time passes. The function ( + ) is called a
travelling wave moving to the left with speed c. In a similar manner, the function ( )is also a travelling wave
moving to the right with speed c.
The integral term in dAlemberts formula comes from the initial velocity. In the special case when the initial
velocity is zero, (that is , 0 = = 0), equation (2.1) will be written as in (6.3) which is the sum of
travelling waves. The waves are initially superimposed. As time increases, the two waves move away from each
other with speed 2c, (Each with speed c).
, =
1
2
+ +
(6.3)
Examples of travelling waves equation are model of stri ng motion as shown in figure 6.2 and 6.3. More plots of
wave equation with different boundary values are shown in figure 6.4 and 6.5. An important observation from the
snapshots of these plots is that wave travel s with a constant speed and continue propagating forever as time goes
to infinity. Depending on the physic of the problem, wave equation transfers a form of energy, like a vibrating
string of a guitar which transfers sound energy or electromagnetic waves which transfer light. According to the
rule of conservation of energy, the total energy of a system is constant, thus theoretically wave should keep
travelling forever. We can also use this argument to show the uniqueness of the solution to the wave equation.
34
35
Figure 6.5. Wave Equation with Damping and External Force on Interval (0, L)
36
Heat distribution:
Generally heat is transferred by conduction, convection and radiation. Conduction occurs in solid and is because of
vibrating hot molecules interacting with neighboring atoms and molecules. This vibration of hot atoms transfers
their energy (heat) to lower energy level atoms (colder), thus heat also transfer energy. Convection is the transfer
of heat in fluid (liquid and gas). Radiation heat transfer, despite of two previous methods of heat transfer which
require a form of substance, is the only form that can occur in the absence of any form of medium. Since atoms
and molecules are composed of charged particles (protons and electrons), their movement result in the emission
of electromagnetic radiation which carries energy away from the surface. Like wave equation, the heat equation
also transfer energy, but it has energy dissipation built into the high order terms, however ther e are special
nonlinear forms of heat distribution (radiation) that can have travelling wave solution. The heat equation is also an
example of parabolic equation.
For a simple linear heat transfer equation, lets consider the problem in (2.1)-(2.3). The solution was shown to be
unique and of the for m (2.14) with coefficient as in (2.16). According to (2.1), although physically not valid, but
theoretically there is no limit to the speed which heat distribute. It can be said that a hot spot in theory will warm
C
its surroundings immediately. As an example, if the rod in (2.1) has initial condition f=100 , it takes some time for
x=L/2 to balance with the boundary points x(0) and x(L), but according to heat equation, this happens in a split
second as was observed in problem 2. In addition to this, the temperature at the no point of the rod will ever
C
become greater than the initial value, f=100 , therefore the energy is not conserved. This is not in contrast with
the rule of conservation of energy since the energy from the hot rod is transferred to its soundings , so depending
on the boundaries of our system, the total energy is still constant.
, =
=
=1
, =
.
=1
(2.14)
, = 1,2,
(2.16)
One mor e interesting observation from the solution to heat equation given in (2.14) -(2.16) is that since we have
solved the heat equation to find the temperature at any point of the rod, (0 ) , therefore 0 1 and
since n is a positive integer, then all the ter ms in (2.14) are always nonnegative.
2
0,
0,
lim
= 0, 0. This means that the solution to heat equation, , cannot be negative unless the
initial condition is negative. This makes sense since temperature cannot hold any negative value. Figure 6.6
explains all these properties.
Snapshots at t=[0.1, 0.2, 0.3, 0.4, 0.5, 1, 2, 3]
37
Heat equation:
o Distribution speed is very fast (infinite)
o Does not conserve energy (loses its energy)
o Cannot exceed the value at initial condition
o Cannot become negative if the initial condition is not negative
o Non reversible equation
o Uniformly distributed and dies down as t goes to infinity
Wave equation:
o Propagation speed is finite
o conserving energy
o Can exceed the value of initial condition (when two waves coincides)
o Does not depend on initial condition to have negative value.
o Reversible equation
o Bouncing up and down and distribute forever.
The End.
Ramin Shamshiri
17 March 2009,