1

MAP 5345
Partial Differential Equations
PROJECT I

RAMIN SHAMSHIRI
Spring 2009

Due March 18, 2009

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

1.

Consider the IVP

= ,

, = ,

, = ,

(1.1)

where is a positive constant and , are given functions

Show that the unique solution of the IVP is given by:

, =

+ +

(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 =

where G(x) is an antiderivative of g(x)

1

Rearranging (1.1) as

=
=

c2

= , let = , = + . Using the chain rules, we get;

+
+

(1.3)

(1.4)

Plugging (1.3), (1.4) into (1.1):

=

+2

2
2
2
1


2
2

=
=

2
+

c2

2
2
=

1
c2

indicates that

(1.5)
(1.6)

= 0. This yield the general solution of the form:

, = + + ( )

(1.7)

Applying the initial conditions, , 0 = and , 0 = :

, 0 = = +

+
, 0 =
+
= + = ()

(1.8)
(1.9)

Integrating (1.9) gives:

= +

(1.10)

Multiply equation (1.10) by a negative sign, we get:

= = ( )

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

Solving (1.7) and (1.10) for and yields:

=
=

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 ()

If > 0, the general solution is a linear combination of sines and cosines:

= 1 ( ) + 2 ( )
If = 0, = 1 + 2
If < 0, T(t) is a linear combination of exponentially growing and decaying solution in time.
The boundary problem:
2
=
2
0 = =0
If > 0, ( 0 not eigenvalues for this problem), exponential solutions have imaginary exponents, . If we
desire real independent solutions, the choices and are usually made, ( and are
each linear combinations of ).
= 1 + 2
0 = 0,

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

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

(1.14)

Due 18 Ma rch 2009

The initial conditions are satisfied if:

() =

=1

() =

=1

(1.15)

(1.16)

From Fourier since series, we know that

2

=
sin

0

=
sin

0

forms an orthogonal set, therefor e:

(1.17)
(1.18)
1

Using the sum to product of sine and cosine ( . = [ + + ( )]), equation (1.14) can
2

be written as:

, =
=1

1
2

+ +

=1

(1.19)

Substituting with + and in (1.15) we get:

1
2
1
2

+ =

sin
=1

sin

=1

(1.20)

(1.21)

Integrating (1.16) from to + :

+

=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)

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

The dAlemberts solution to (1.1) is of the form:

, = + +

(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)

Integrating equation (1.26) from to and dividing by gives:

1
=
+

(1.27)

Solving (1.25) and (1.27) yields:

1
1

= +
+
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

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

Suppose that c=1, = if , = otherwise, and g=0. Plot snapshots of the string
at = , . , . , . , , , , , and verbally describe the motion.

Solution:

1 ,
0
0,

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):=piec ewise(abs (x)<=1 ,1 -abs (x),0 ); g(x):=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 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]

Sequenc e of s naps hots at times [t1 , t2 , t3 , ..., tn]:

>ts eq:=[0 ,0 .25 ,0 .5 ,0 .75 ,1 ,4 ,7 ,9 ]:
plot([s eq(u(x,t),t=ts eq)],x=- 10 ..10 ,0 ..1 ,numpoints =250 ,
c olor=[black,red,magenta,blue,green],
title=cat(`Snaps hots at `,c onvert('t'=tseq,s tring)));

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

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

C reate movie (t = t0 .. tn, n + 1 frames ):

>t0 :=0 : tn:=12 : n:=60 :
h:=(tn- t0 )/n: `G enerating graphs ... Be patient! `;
s naps :=seq(plot(u(x,t0 +i* h),x=-10 ..10 ,0 ..1 ,numpoints =250 ),i=0 ..n): `Finished.`;
> dis play(s naps [1 ..n+1 ],ins equenc e=true);

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

Repeat previous question, but with = and = , = otherwise.

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]

Sequenc e of s naps hots at times [t 1, t 2, t 3, ..., t n]:

>ts eq:=[0 ,0 .5 ,1 ,2 .5 ,5 ,7 .5 ,9 ]:
plot([s eq(u(x,t),t=ts eq)],x=- 10 ..10 ,0 ..1 ,numpoints =250 ,
c olor=[black,red,magenta,blue,green],
title=cat(`Snaps hots at `,c onvert('t'=tseq,s tring)));

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

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

Create movie (t = t0 .. tn, n + 1 frames):

>t0:=0: tn:=12: n:=60:
h:=(tn-t0)/n: `Generating graphs... Be patient!`;
snaps:=seq(plot(u(x,t0+i*h),x=-10..10,0..1,numpoints=250),i=0..n): `Finished.`;
> display(snaps[1..n+1],insequence=true);

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

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)

Write the solution of the IBVP as a Fourier series.

Solution:

Let = , Equation (2.1) can be written as:

, =

, ,
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.

Ra min Sha mshi ri

=
=

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

11

Now we have two ODEs as follow:

= 0
= 0

(2.7)
(2.8)

Considering the boundary conditions in (2.5);

0 =0

= 0,

&

> 0.

This implies , 0 or 0 = = 0. Ignoring the trivial solution , 0 and combining these

boundary conditions with the ODE in (2.7), we get the boundary value problem as follow:
= 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

1) > 0. Ther efore 1 = 2 = 0 and there is no nontrivial solution for > 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

Thus for = 0, there is no nontrivial solution to (2.9).

Case 3: < 0

= , thus a general to (2.9) is

= 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)

where are arbitrary nonzero constants.

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

12

Having determined that =

+

for any positive integer n, we can substitute it in (2.8):

=0

For each n=1,2,3,, a general solution to this linear first order equation is:
=

(2.12)

Substituting (2.11) and (2.12) in , = . ;

, =

where = is also an arbitrary constant.

(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

, =

will give us the solution to (2.1)-(2.2):

, =
=1

=1

(2.14)

To satisfy the initial condition in (2.3),

, 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).

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

13

Compute the Fourier coefficients explicitly for the special case f = 1 .

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)

Rearranging (2.20) and (2.21) will confirm the solution.

2
2

( 2 2 )

1 ()

=0

2 2
=0

2 2 (

1 ()
2

2 2 () = 0

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

14

Plot a sequence of snapshots of an approximate solution obtained by suitably truncating the Fourier
series.
Solution:

Let = 1, then (2.18) becomes:

, =

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 =

Ra min Sha mshi ri

[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] =

+]

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

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

= 1 so that R = 4 and S = 0. The

initial condition gives:

1
2
3
4
5
6
7
8
9

= 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.

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

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

Ra min Sha mshi ri

+
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

MAP 5345, Pa rtial Differential Equati ons

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

Due 18 Ma rch 2009

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

Ra min Sha mshi ri

+
-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

MAP 5345, Pa rtial Differential Equati ons

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

Due 18 Ma rch 2009

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

Figure 2.1. Numerical solution to heat equation.

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

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.

Figure 2.2. Solution to the heat equation in 2.1.

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

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.

Figure 2.3, t=0.001

Figure 2.4, t=0.1

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

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

res tart: with(plots ):

u[t]=k* u[xx];
u(0 ,x)=f(x);
u(t,x)=int(G [k](x- y)* f(y),y=- infinity..infinity);
G [k](t,x)=exp(- x^2 /(4*k* t))/'s qrt(4*Pi*k* t)';
k:=1 ; f(x):=exp(- x^2 );
> `C omputing u(t,x)...`;
f(x):=s implify(f(x)): f0 :=unapply(f(x),x): fudge:=t - >piecewis e(t=0 ,1 ,t):
G00 :=(t,x)- >exp(- x^2 /(4*k* t))/sqrt(4*Pi* k*t): G0 :=(t,x)- >G00 (fudge(t),x):
u:=(t,x)- >piecewis e(t>0 ,int(G0 (t,z)* f0 (x- z),z=- infinity..infinity),f0 (x)):
`Finis hed.`;
> t0 :=0 .2 :
plot(u(t0 ,x),x=-10 ..10 ,0 ..1 ,numpoints =250 ,
title=cat(`Snaps hot at `,c onvert('t'=t0 ,s tring)));
> ts eq:=[0 ,0 .1 ,0 .5 ,2 .5 ,10 ]:
plot([s eq(u(t,x),t=ts eq)],x=- 10 ..10 ,0 ..1 ,numpoints =250 ,
c olor=[black,red,magenta,blue,green],
title=cat(`Snaps hots at `,c onvert('t'=tseq,s tring)));
> t0 :=0 : tn:=10 : n:=50 :
h:=(tn- t0 )/n: `G enerating graphs ... Be patient! `;
s naps :=seq(plot(u(t0 +i* h,x),x=-10 ..10 ,0 ..1 ,numpoints =250 ),i=0 ..n): `Finished.`;
dis play(s naps [1 ..n+1 ],ins equenc e=true);

Snapshot at t=[0,0.1,0.5,2.5,10]

Figure 2.5. Solution to heat equation with initial condition

= 2

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

22

3.

Backward Heat Equation. Show that the backward heat equation

=

(3.1)

subject to , = , = and , = () is not well-posed. Show that if the data are changed an
arbitrary sm all amount, for example,

for large n, then solution changes by a large amount.

Solution:
According to the previous problem, the solution to equation (3.1) can be written as:

, =

, =
=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

solution for f x = 1 with = 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)

Implementing (3.4) in Delphi and plotting the results:

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); //beta=-1 for this problem
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 -cos (n* pi))+(1 /n))* (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.

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

23

Figure 3.1. Comparing equation (2.) with equation (3.) at t=0.001

Figure 3.2. Plot of equation (3.4) at t=0.01.

(The response is highly oscillatory and unstable.)

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

24

n x

For the same k=-1 and L=1, if we let u x, 0 = f x = 1 + sin

, then the coefficient cn can be calculated as

below:
=
=
=

sin

).

, = 1,2,

+
sin2

0
0

1
1

1 +

2
.
0

(1 +

1 +

, = 1,2,

(3.5)

and the solution to the PDE (3.1) will be given by:

, =

=1

2 1 +

22

(3.6)

Implementing (3.6) and plotting the results:

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); //beta=-1 for this problem
t:=s trtofloat(edit1 .T ext);
s tart:
fx:=0 ;
for n:=1 to s pinedit1 .Value do
begin
fx:=fx+ (1 /n* pi) * (2* (1-c os (n* pi))+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.

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).

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

25

Figure 3.3. Plot of equation (3.6) at t=0 and t=0.001.

(Big change in u(x,t) compared with Figure 3.1)

Figure 3.4. Plot of equation (3.6) at t=0.01.

(Big change in u(x,t) compared with Figure 3.2)

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

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)

Applying the first boundary condition in (4.2):

0 =0
: 1 = 0
Therefore equation (4.2) becomes:
= 2

(4.3)

Applying the second boundary condition, + = 0

2 + 2 = 0

(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.

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

27

Figure 4.1. Plot of = for different c =[0.1, 0.5, 1, 2, 3, 10]

To solve tan = c , let = x and c = 1 The problem is now to solve tan x = x . From figure, it can be
2n 1 2n +1
seen that this equation has a solution in each interval
,
, n = 0,1,2, . Therefore ther e would be
2

infinite number of x points to satisfy this equation.

Since it is difficult to find an explicit term for that satisfy equation (4.6), we can set
sets of eigenfunctions as:
= 2 n

= and then write the

(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,

The sets of eigenfunctions corresponding to these eigenvalues are then:

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

(4.9)

Due 18 Ma rch 2009

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)

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

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

Figure 4.2. Plot of tan = for c=1.

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

30

Case 2, 3: > 0 , = 0 & = 0 , > 0

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

Figure 4.3.Plot of the first 10 term of =

Ra min Sha mshi ri

and =

MAP 5345, Pa rtial Differential Equati ons

2 +1

Due 18 Ma rch 2009

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)

In polar coordinates, equation (5.3) will be written as:

2 1
1 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

= , with initial condition = (), If we assume 1 (, ) and 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.

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

32

6.

A Comparison of Wave Propagation and Diffusion Processes.

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.

Properties of wave propagation:

A model for the motion of a vibrating string is an example of wave equation. In this case, the boundary conditions
for equation (6.1) will dictate the condition of the string at the ending points, (whether it is held fixed or not). The
initial conditions usually specify the initial displacement and the initial velocity of each point.
Considering a standing wave ( a wave that vibrates in place without lateral motion along the string), the solution

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

constant c depends on the linear density and tension of the string.

When the boundary conditions in wave equations are not given, as in problem 1, the solution will give the infinite
string problem consisting travelling waves. Wave equation travels at a finite speed of c to the left and right of the
one space dimension (the x axis) as time goes by. It can be seen that the constant in the wave equation is always
2
positive ( ^ ). Looking at dAlemberts formula in (1.2), let () be a function on . The function ( + ) where c
is a positive constant is the same shape as () with its position shifted to the left by c. For example, let
= sin(), then plot of + = sin( + ) versus sin(x) will look like in figure 6.1.
, =

Ra min Sha mshi ri

1
2

+ +

1
2

MAP 5345, Pa rtial Differential Equati ons

(1.2)

Due 18 Ma rch 2009

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.

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

34

Snapshots at t=[0, 0.5, 1, 2.5, 5, 10]

Figure 6.2 . Infinite String.

Snapshot at t=[0, 0.5, 1, 5,9, 9.6, 9.8]

Figure 6.3 .Finite String (Dirichlet).

Snapshots at t=[0, 0.25, 0.5, 1, 5, 9, 9.75, 10]

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

35

Figure 6.4 . Finite String (Neumann).

Snapshots at t=[0.1, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10]

Figure 6.5. Wave Equation with Damping and External Force on Interval (0, L)

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

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]

Figure 6.6 . Heat Equation with Source on Interval (0, L)

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009

37

Summary of the difference between heat and wave equation:

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,

Ra min Sha mshi ri

MAP 5345, Pa rtial Differential Equati ons

Due 18 Ma rch 2009