Solutions for

homework assignment #4
Problem 1. Solve Laplaces equation inside a rectangle 0 x L, 0 y H, with the
following boundary conditions:
u
x
(0, y) = 0,
u
x
(L, y) = 0, u(x, 0) = 0, u(x, H) = f(x).
Solution:
u(x, y) = b
0
y
H
+

n=1
b
n
_
sinh
nH
L
_
1
sinh
ny
L
cos
nx
L
,
where
b
0
+

n=1
b
n
cos
nx
L
is the Fourier cosine series of the function f(x) on [0, L], that is,
b
0
=
1
L
_
L
0
f(x) dx, b
n
=
2
L
_
L
0
f(x) cos
nx
L
dx, n = 1, 2, . . .
Detailed solution: We search for the solution of the boundary value problem as a superposition
of solutions u(x, y) = (x)h(y) with separated variables of Laplaces equation that satisfy the three
homogeneous boundary conditions.
Substituting u(x, y) = (x)h(y) into Laplaces equation

2
u
x
2
+

2
u
y
2
= 0,
we obtain

(x)h(y) +(x)h

(y) = 0,

(x)
(x)
=
h

(y)
h(y)
.
Since the left-hand side does not depend on y while the right-hand side does not depend on x, it
follows that

(x)
(x)
=
h

(y)
h(y)
= ,
where is a constant. Then

= , h

= h.
Conversely, if functions and h are solutions of the above ODEs for the same value of , then
u(x, y) = (x)h(y) is a solution of Laplaces equation.
Substituting u(x, y) = (x)h(y) into the homogeneous boundary conditions, we get

(0)h(y) = 0,

(L)h(y) = 0, (x)h(0) = 0.
It is no loss to assume that neither nor h is identically zero. Then the boundary conditions are
satised if and only if

(0) =

(L) = 0, h(0) = 0.
To determine , we have an eigenvalue problem

= ,

(0) =

(L) = 0.
1
This problem has eigenvalues
n
=
_
n
L
_
2
, n = 0, 1, 2, . . .. The corresponding eigenfunctions are

0
= 1 and
n
(x) = cos
nx
L
, n = 1, 2, . . ..
The function h is to be determined from the equation h

= h and the boundary condition

h(0) = 0. We may assume that is one of the above eigenvalues so that 0. If = 0 then the
general solution of the equation is h(y) = c
1
+c
2
y, where c
1
, c
2
are constants. If > 0 then the general
solution is h(y) = c
1
cosh y + c
2
sinh y, where =

and c
1
, c
2
are constants. In both cases, the
boundary condition h(0) = 0 holds if c
1
= 0.
Thus we obtain the following solutions of Laplaces equation satisfying the three homogeneous
boundary conditions:
u
0
(x, y) = y, u
n
(x, y) = sinh
ny
L
cos
nx
L
, n = 1, 2, . . .
A superposition of these solutions is a series
u(x, y) = c
0
y +

n=1
c
n
sinh
ny
L
cos
nx
L
,
where c
0
, c
1
, c
2
, . . . are constants. Substituting the series into the boundary condition u(x, H) = f(x),
we get
f(x) = c
0
H +

n=1
c
n
sinh
nH
L
cos
nx
L
.
The right-hand side is a Fourier cosine series on the interval [0, L]. Therefore the boundary condition
is satised if the right-hand side coincides with the Fourier cosine series
b
0
+

n=1
b
n
cos
nx
L
of the function f(x) on [0, L]. Hence
c
0
=
b
0
H
, c
n
=
b
n
sinh
nH
L
, n = 1, 2, . . . ,
where
b
0
=
1
L
_
L
0
f(x) dx, b
n
=
2
L
_
L
0
f(x) cos
nx
L
dx, n = 1, 2, . . .
Problem 2. Solve Laplaces equation inside a semicircle of radius a (0 < r < a, 0 < < )
subject to the boundary conditions: u = 0 on the diameter and u(a, ) = g().
Solution:
u(r, ) =

n=1
b
n
_
r
a
_
n
sin n,
where

n=1
b
n
sin n
is the Fourier sine series of the function g() on [0, ], that is,
b
n
=
2

_

0
g() sin n d, n = 1, 2, . . .
Detailed solution: Laplaces equation in polar coordinates (r, ):

2
u
r
2
+
1
r
u
r
+
1
r
2

2
u

2
= 0.
2
The boundary condition u = 0 on the diameter gives rise to three boundary conditions in polar
coordinates:
u(r, 0) = u(r, ) = 0 (0 < r < a),
u(0, ) = 0 (0 < < )
(the latter condition means that u = 0 at the origin).
We search for the solution of the boundary value problem as a superposition of solutions u(r, ) =
h(r)() with separated variables of Laplaces equation that satisfy the above homogeneous boundary
conditions.
Substituting u(r, ) = h(r)() into Laplaces equation, we obtain
h

(r)() +
1
r
h

(r)() +
1
r
2
h(r)

() = 0,
r
2
h

(r) +rh

(r)
h(r)
=

()
()
.
Since the left-hand side does not depend on while the right-hand side does not depend on r, it follows
that
r
2
h

(r) +rh

(r)
h(r)
=

()
()
= ,
where is a constant. Then
r
2
h

(r) +rh

(r) = h(r),

= .
Conversely, if functions h and are solutions of the above ODEs for the same value of , then
u(r, ) = h(r)() is a solution of Laplaces equation in polar coordinates.
Substituting u(r, ) = h(r)() into the homogeneous boundary conditions, we get
h(r)(0) = 0, h(r)() = 0, h(0)() = 0.
It is no loss to assume that neither h nor is identically zero. Then the boundary conditions are
satised if and only if (0) = () = 0, h(0) = 0.
To determine , we have an eigenvalue problem

= , (0) = () = 0.
This problem has eigenvalues
n
= n
2
, n = 1, 2, . . .. The corresponding eigenfunctions are
n
() =
sin n, n = 1, 2, . . ..
The function h is to be determined from the equation r
2
h

+rh

= h and the boundary condition

h(0) = 0. We may assume that is one of the above eigenvalues so that > 0. Then the general
solution of the equation is h(r) = c
1
r

+c
2
r

, where =

and c
1
, c
2
are constants. The boundary
condition h(0) = 0 holds if c
2
= 0.
Thus we obtain the following solutions of Laplaces equation satisfying the three homogeneous
boundary conditions:
u
n
(r, ) = r
n
sin n, n = 1, 2, . . .
A superposition of these solutions is a series
u(r, ) =

n=1
c
n
r
n
sin n,
where c
1
, c
2
, . . . are constants. Substituting the series into the boundary condition u(a, ) = g(), we
get
g() =

n=1
c
n
a
n
sin n.
3
The right-hand side is a Fourier sine series on the interval [0, ]. Therefore the boundary condition is
satised if the right-hand side coincides with the Fourier sine series

n=1
b
n
sin n
of the function g() on [0, ]. Hence
c
n
= b
n
a
n
, n = 1, 2, . . . ,
where
b
n
=
2

_

0
g() sin n d, n = 1, 2, . . .
Problem 3. Solve Laplaces equation inside a 90
o
sector of a circular annulus (a < r < b,
0 < < /2) subject to the boundary conditions:
u(r, 0) = 0, u(r, /2) = 0, u(a, ) = 0, u(b, ) = f().
Solution:
u(r, ) =

n=1
b
n
(r/a)
2n
(a/r)
2n
(b/a)
2n
(a/b)
2n
sin 2n,
where

n=1
b
n
sin 2n
is the Fourier sine series of the function f() on [0, /2], that is,
b
n
=
4

_
/2
0
f() sin 2n d, n = 1, 2, . . .
Detailed solution: We search for the solution of the boundary value problem as a superposition
of solutions u(r, ) = h(r)() with separated variables of Laplaces equation that satisfy the three
homogeneous boundary conditions.
As shown in the solution of Problem 2, u(r, ) = h(r)() is a solution of Laplaces equation in
polar coordinates if functions h and are solutions of the equations
r
2
h

(r) +rh

(r) = h(r),

=
for the same constant .
Substituting u(r, ) = h(r)() into the homogeneous boundary conditions, we get
h(r)(0) = 0, h(r)(/2) = 0, h(a)() = 0.
It is no loss to assume that neither h nor is identically zero. Then the boundary conditions are
satised if and only if (0) = (/2) = 0, h(a) = 0.
To determine , we have an eigenvalue problem

= , (0) = (/2) = 0.
This problem has eigenvalues
n
= (2n)
2
, n = 1, 2, . . .. The corresponding eigenfunctions are
n
() =
sin 2n, n = 1, 2, . . ..
4
The function h is to be determined from the equation r
2
h

+rh

= h and the boundary condition

h(a) = 0. We may assume that is one of the above eigenvalues so that > 0. Then the general
solution of the equation is h(r) = c
1
r

+c
2
r

, where =

and c
1
, c
2
are constants. The boundary
condition h(a) = 0 holds if c
1
a

+ c
2
a

= 0, which implies that h(r) = c

0
((r/a)

(r/a)

), where
c
0
is a constant.
Thus we obtain the following solutions of Laplaces equation satisfying the three homogeneous
boundary conditions:
u
n
(r, ) =
_
_
r
a
_
2n

_
a
r
_
2n
_
sin 2n, n = 1, 2, . . .
A superposition of these solutions is a series
u(r, ) =

n=1
c
n
_
_
r
a
_
2n

_
a
r
_
2n
_
sin 2n,
where c
1
, c
2
, . . . are constants. Substituting the series into the boundary condition u(b, ) = f(), we
get
f() =

n=1
c
n
_
_
b
a
_
2n

_
a
b
_
2n
_
sin 2n.
The right-hand side is a Fourier sine series on the interval [0, /2]. Therefore the boundary condition
is satised if the right-hand side coincides with the Fourier sine series

n=1
b
n
sin 2n
of the function f() on [0, /2]. Hence
c
n
=
b
n
(b/a)
2n
(a/b)
2n
, n = 1, 2, . . . ,
where
b
n
=
4

_
/2
0
f() sin 2n d, n = 1, 2, . . .
Problem 4. Consider the heat equation in a two-dimensional rectangular region, 0 < x <
L, 0 < y < H,
u
t
= k
_

2
u
x
2
+

2
u
y
2
_
subject to the initial condition u(x, y, 0) = f(x, y).
Solve the initial-boundary value problem and analyze the temperature as t if the
boundary conditions are:
u
x
(0, y, t) = 0,
u
x
(L, y, t) = 0,
u
y
(x, 0, t) = 0,
u
y
(x, H, t) = 0.
Solution:
u(x, y, t) =

n=0

m=0
c
nm
exp
_

_
(n/L)
2
+ (m/H)
2
_
kt
_
cos
nx
L
cos
my
H
,
5
where
c
00
=
1
LH
_
L
0
_
H
0
f(x, y) dxdy,
c
n0
=
2
LH
_
L
0
_
H
0
f(x, y) cos
nx
L
dxdy, n 1,
c
0m
=
2
LH
_
L
0
_
H
0
f(x, y) cos
my
H
dxdy, m 1,
c
nm
=
4
LH
_
L
0
_
H
0
f(x, y) cos
nx
L
cos
my
H
dxdy, n, m 1.
As t , the temperature uniformly approaches the constant c
00
, the mean value of f(x, y) over
the rectangle.
Detailed solution: We search for the solution of the initial-boundary value problem as a
superposition of solutions u(x, y, t) = (x)h(y)G(t) with separated variables of the heat equation that
satisfy the boundary conditions.
Substituting u(x, y, t) = (x)h(y)G(t) into the heat equation, we obtain
(x)h(y)G

(t) = k
_

(x)h(y)G(t) +(x)h

(y)G(t)
_
,
G

(t)
kG(t)
=

(x)
(x)
+
h

(y)
h(y)
.
Since any of the expressions
G

(t)
kG(t)
,

(x)
(x)
, and
h

(y)
h(y)
depend on one of the variables x, y, t and does
not depend on the other two, it follows that each of these expressions is constant. Hence

(x)
(x)
= ,
h

(y)
h(y)
= ,
G

(t)
kG(t)
= ( +),
where and are constants. Then

= , h

= h, G

= ( +)kG.
Conversely, if functions , h, and G are solutions of the above ODEs for the same values of and ,
then u(x, y, t) = (x)h(y)G(t) is a solution of the heat equation.
Substituting u(x, y, t) = (x)h(y)G(t) into the boundary conditions, we get

(0)h(y)G(t) =

(L)h(y)G(t) = 0, (x)h

(0)G(t) = (x)h

(H)G(t) = 0.
It is no loss to assume that neither nor h nor G is identically zero. Then the boundary conditions
are satised if and only if

(0) =

(L) = 0, h

(0) = h

(H) = 0.
To determine , we have an eigenvalue problem

= ,

(0) =

(L) = 0.
This problem has eigenvalues
n
=
_
n
L
_
2
, n = 0, 1, 2, . . .. The corresponding eigenfunctions are

0
= 1 and
n
(x) = cos
nx
L
, n = 1, 2, . . ..
To determine h, we have another eigenvalue problem
h

= h, h

(0) = h

(H) = 0.
6
This problem has eigenvalues
m
=
_
m
H
_
2
, m = 0, 1, 2, . . .. The corresponding eigenfunctions are

0
= 1 and
m
(y) = cos
my
H
, m = 1, 2, . . ..
The function G is to be determined from the equation G

= ( +)kG. The general solution of

this equation is G(t) = c
0
e
(+)kt
, where c
0
is a constant.
Thus we obtain the following solutions of the heat equation satisfying the boundary conditions:
u
nm
(x, y, t) = e
(
n
+
m
)kt

n
(x)
m
(y)
= exp
_

_
(n/L)
2
+ (m/H)
2
_
kt
_
cos
nx
L
cos
my
H
, n, m = 0, 1, 2, . . .
A superposition of these solutions is a double series
u(x, y, t) =

n=0

m=0
c
nm
exp
_

_
(n/L)
2
+ (m/H)
2
_
kt
_
cos
nx
L
cos
my
H
,
where c
nm
are constants. Substituting the series into the initial condition u(x, y, 0) = f(x, y), we get
f(x, y) =

n=0

m=0
c
nm
cos
nx
L
cos
my
H
=

n=0

m=0
c
nm

n
(x)
m
(y).
To determine the coecients c
nm
, we multiply both sides by
N
(x)
M
(y) (N, M 0) and integrate
over the rectangle 0 x L, 0 y H. We assume that the series may be integrated term-by-term:
_
L
0
_
H
0
f(x, y)
N
(x)
M
(y) dxdy =

n=0

m=0
c
nm
_
L
0
_
H
0

N
(x)
M
(y)
n
(x)
m
(y) dxdy
=

n=0

m=0
c
nm
_
L
0

N
(x)
n
(x) dx
_
H
0

M
(y)
m
(y) dy.
Using the orthogonality relations
_
L
0

N
(x)
n
(x) dx = 0, N = n,
_
H
0

M
(y)
m
(y) dy = 0, M = m,
we obtain
_
L
0
_
H
0
f(x, y)
N
(x)
M
(y) dxdy = c
NM
_
L
0

2
N
(x) dx
_
H
0

2
M
(y) dy.
It remains to recall that
_
L
0

2
0
(x) dx = L,
_
L
0

2
N
(x) dx =
L
2
, N 1,
and, similarly,
_
H
0

2
0
(x) dx = H,
_
H
0

2
M
(x) dx =
H
2
, M 1.
In the double series expansion of u(x, y, t), each term contains an exponential factor e
(
n
+
m
)kt
,
which is decaying as t except for the case n = m = 0 when this factor is equal to 1. It follows
that, as t , the solution u(x, y, t) uniformly converges to the constant c
00
:
lim
t
u(x, y, t) = c
00
=
1
LH
_
L
0
_
H
0
f(x, y) dxdy.
7
Problem 5. Consider the wave equation for a vibrating rectangular membrane (0 < x < L,
0 < y < H)

2
u
t
2
= c
2
_

2
u
x
2
+

2
u
y
2
_
subject to the initial conditions u(x, y, 0) = 0 and
u
t
(x, y, 0) = f(x, y).
Solve the initial-boundary value problem if
u
x
(0, y, t) = 0,
u
x
(L, y, t) = 0,
u
y
(x, 0, t) = 0,
u
y
(x, H, t) = 0.
Solution:
u(x, y, t) =

n=0

m=0
b
nm
sin(
_
(n/L)
2
+ (m/H)
2
ct)
_
(n/L)
2
+ (m/H)
2
c
cos
nx
L
cos
my
H
,
where b
nm
are coecients of the expansion
f(x, y) =

n=0

m=0
b
nm
cos
nx
L
cos
my
H
.
The formulas for b
nm
are obtained in the solution of Problem 4.
Detailed solution: We search for the solution of the initial-boundary value problem as a
superposition of solutions u(x, y, t) = (x)h(y)G(t) with separated variables of the wave equation that
satisfy the boundary conditions.
Substituting u(x, y, t) = (x)h(y)G(t) into the wave equation, we obtain
(x)h(y)G

(t) = c
2
_

(x)h(y)G(t) +(x)h

(y)G(t)
_
,
G

(t)
c
2
G(t)
=

(x)
(x)
+
h

(y)
h(y)
.
Since any of the expressions
G

(t)
c
2
G(t)
,

(x)
(x)
, and
h

(y)
h(y)
depend on one of the variables x, y, t and does
not depend on the other two, it follows that each of these expressions is constant. Hence

(x)
(x)
= ,
h

(y)
h(y)
= ,
G

(t)
c
2
G(t)
= ( +),
where and are constants. Then

= , h

= h, G

= ( +)c
2
G.
Conversely, if functions , h, and G are solutions of the above ODEs for the same values of and ,
then u(x, y, t) = (x)h(y)G(t) is a solution of the wave equation.
Substituting u(x, y, t) = (x)h(y)G(t) into the boundary conditions, we get

(0)h(y)G(t) =

(L)h(y)G(t) = 0, (x)h

(0)G(t) = (x)h

(H)G(t) = 0.
It is no loss to assume that neither nor h nor G is identically zero. Then the boundary conditions
are satised if and only if

(0) =

(L) = 0, h

(0) = h

(H) = 0.
8
To determine , we have an eigenvalue problem

= ,

(0) =

(L) = 0.
This problem has eigenvalues
n
=
_
n
L
_
2
, n = 0, 1, 2, . . .. The corresponding eigenfunctions are

0
= 1 and
n
(x) = cos
nx
L
, n = 1, 2, . . ..
To determine h, we have another eigenvalue problem
h

= h, h

(0) = h

(H) = 0.
This problem has eigenvalues
m
=
_
m
H
_
2
, m = 0, 1, 2, . . .. The corresponding eigenfunctions are

0
= 1 and
m
(y) = cos
my
H
, m = 1, 2, . . ..
The function G is to be determined from the equation G

= ( + )c
2
G. We may assume that
and are eigenvalues of the above eigenvalue problems so that , 0. If = = 0 then the
general solution of the equation is G(t) = C
0
+D
0
t, where C
0
, D
0
are constants. If + > 0 then the
general solution of the equation is
G(t) = C
0
cos(
_
+ct) +D
0
sin(
_
+ct),
where C
0
, D
0
are constants.
Thus for any n, m 0 we have the following solutions of the wave equation satisfying the boundary
conditions:
u(x, y, t) =
_
C
0
cos(
_

n
+
m
ct) +D
0
sin(
_

n
+
m
ct)
_

n
(x)
m
(y)
=
_
C
0
cos(
_
(n/L)
2
+ (m/H)
2
ct) +D
0
sin(
_
(n/L)
2
+ (m/H)
2
ct)
_
cos
nx
L
cos
my
H
.
A superposition of these solutions is a double series
u(x, y, t) =

n=0

m=0
_
C
nm
cos(
_
(n/L)
2
+ (m/H)
2
ct)
+ D
nm
sin(
_
(n/L)
2
+ (m/H)
2
ct)
_
cos
nx
L
cos
my
H
,
where C
nm
, D
nm
are constants. Substituting the series into the initial conditions u(x, y, 0) = 0 and
u
t
(x, y, 0) = f(x, y), we get

n=0

m=0
C
nm
cos
nx
L
cos
my
H
= 0,
f(x, y) =

n=0

m=0
D
nm
_
(n/L)
2
+ (m/H)
2
c cos
nx
L
cos
my
H
.
It follows that C
nm
= 0 while D
nm
=
b
nm
_
(n/L)
2
+ (m/H)
2
c
, where b
nm
are coecients of the
expansion
f(x, y) =

n=0

m=0
b
nm
cos
nx
L
cos
my
H
.
9