Lecture Notes for Math 251: ODE and PDE.

Lecture 35:
10.8 Laplaces Equation
Shawn D. Ryan
Spring 2012
Last Time: We studied another fundamental equation in the study of partial differential equations, which is the wave equation. Today we will look at the final fundamental equation, which is
Laplaces Equation.

1 Laplaces Equation
We will consider the two-dimensional and three-dimensional Laplace Equations
(2D) : uxx + uyy = 0,
(3D) : uxx + uyy + uzz = 0.

(1)
(2)

1.1 Dirichlet Problem for a Rectangle

We want to find the function u satisfying Laplaces Equation
uxx + uyy = 0

(3)

in the rectangle 0 < x < a, 0 < y < b, and satisfying the boundary conditions
u(x, 0) = 0,
u(0, y) = 0,

u(x, b) = 0, 0 < x < a,

u(a, y) = f (y), 0 y b.

(4)
(5)

We need four boundary conditions for the four spatial derivatives.

Start by using Separation of Variables and assume u(x, y) = X(x)Y (y). Substitute u into
Equation (??). This yields
Y 00
X 00
=
= ,
(6)
X
Y
where is a constant. We obtain the following system of ODEs
X 00 X = 0
Y 00 + Y = 0.
1

(7)
(8)

From the boundary conditions we find

X(0) = 0
Y (0) = 0, Y (b) = 0.

(9)
(10)

We first solve the ODE for Y , which we have seen numerous times before. Using the BCs we find
there are nontrivial solutions if and only if is an eigenvalue
=

n
2
,
b

n = 1, 2, 3, ...

(11)

and Yn (y) = sin( ny

), the corresponding eigenfunction. Now substituting in for we want to
b
solve the ODE for X. This is another problem we have seen regularly and the solution is
Xn (x) = c1 cosh

nx

nx

+ c2 sinh
b
b

(12)

The BC implies that c1 = 0. So the fundamental solution to the problem is

un (x, y) = sinh

ny

nx

sin
.
b
b

(13)

By linear superposition the general solution is

u(x, y) =

cn un (x, y) =

cn sinh

n=1

n=1

nx

ny

sin
.
b
b

(14)

Using the last boundary condition u(a, y) = f (y) solve for the coefficients cn .
u(a, y) =

cn sinh

n=1

na

ny

sin
= f (y)
b
b

Using the Fourier Since Series coefficients we find

Z b
2
ny

dy.
cn =
f (y) sin
na
b sinh( b ) 0
b

(15)

(16)

1.2 Dirichlet Problem For A Circle

Consider solving Laplaces Equation in a circular region r < a subject to the boundary condition
u(a, ) = f ()

(17)

where f is a given function on 0 2. In polar coordinates Laplaces Equation becomes

1
1
urr + ur + 2 u = 0.
r
r
2

(18)

Try Separation of Variables in Polar Coordinates

u(r, ) = R(r)(),

(19)

plug into the differential equation, Equation (??). This yields

1
1
R00 + R0 + 2 R00 = 0
r
r

(20)

or

R00
R0
00
+r =
=
R
R

where is a constant. We obtain the following system of ODEs

r2

(21)

r 2 R00 + rR0 R = 0,
00 + = 0.

(22)
(23)

Since we have no homogeneous boundary conditions we must use instead the fact that the solutions
must be bounded and also periodic in with period 2. It can be shown that we need to be real.
Consider the three cases when < 0, = 0, > 0.
If < 0, let = 2 , where > 0. So we find the equation for becomes 00 2 = 0.
So
() = c1 e + c2 e
(24)
can only be periodic if c1 = c2 = 0, so cannot be negative (Since we do not get any nontrivial
solutions.
If = 0, then the equation for becomes 00 = 0 and thus
() = c1 + c2

(25)

For to be periodic c2 = 0. Then the equation for R becomes

r 2 R00 + rR0 = 0.

(26)

This equation is an Euler equation and has solution

R(r) = k1 + k2 ln(r)

(27)

Since we also need the solution bounded as r , then k2 = 0. So u(r, ) is a constant, and thus
proportional to the solution u0(r, ) = 1.
If > 0, we let = 2 , where > 0. Then the system of equations becomes
r 2 R00 + rR0 2 R = 0
00 + 2 = 0

(28)
(29)

The equation for R is an Euler equation and has the solution

R(r) = k1 r + k2 r
3

(30)

and the equation for has the solution

() = c1 sin() + c2 cos().

(31)

For to be periodic we need to be a positive integer n, so = n. Thus the solution r is

unbounded as r 0. So k2 = 0. So the solutions to the original problem are
un (r, ) = r n cos(n),

vn (r, ) = r n sin(n),

n = 1, 2, 3, ...

(32)

Together with u0 (r, ) = 1, by linear superposition we find

c0 X n
+
r (cn cos(n) + kn sin(n)).
u(r, ) =
2
n=1

(33)

Using the boundary condition from the beginning

c0 X n
u(a, ) =
+
a (cn cos(n) + kn sin(n)) = f ()
2
n=1
for 0 2. We compute to coefficients by using our previous Fourier Series equations
Z 2
1
f () cos(n)d, n = 1, 2, 3, ...
cn =
an 0
Z 2
1
kn =
f () sin(n)d, n = 1, 2, 3, ...
an 0

(34)

(35)
(36)

Note we need both terms since sine and cosine terms remain throughout the general solution.

1.3 HW 10.8 # 2
Find the solution u(x, y) of Laplaces Equation in the rectangle 0 < x < a, 0 < y < b, that
satisfies the boundary conditions
u(0, y) = 0, u(a, y) = 0, 0 < y < b
u(x, 0) = h(x), u(x, b) = 0, 0 x a

(37)
(38)

Answer: Using the method of Separation of Variables, write u(x, y) = X(x)Y (y). We get the
following system of ODEs
X 00 + X = 0,
Y 00 Y = 0,

X(0) = X(a) = 0
Y (b) = 0

(39)
(40)

)2 and Xn (x) = sin( nx

). The solution of the second ODE gives
It follows that n = ( n
a
a
Y (y) = d1 cosh((b y)) + d2 sinh((b y)).
4

(41)

Using y(b) = 0, we find that d1 = 0. Therefore the fundamental solutions are

un (x, y) = sin(

nx
) sinh(n (b y)),
a

(42)

and the general solution is

u(x, y) =

cn sin(

n=1

n(b y)
nx
) sinh(
).
a
a

(43)

Using another boundary condition

h(x) =

cn sin(

n=1

nx
nb
) sinh(
).
a
a

The coefficients are calculated using the equation from the Fourier Sine Series
Z a
nx
2
)dx.
h(x) sin(
cn =
nb
a
a sinh( a ) 0

(44)

(45)

1.4 HW 10.8 # 10a

Consider the problem of finding a solution u(x, y) of Laplaces Equation in the rectangle 0 < x <
a, 0 < y < b, that satisfies the boundary conditions
ux (0, y) = 0,
uy (x, 0) = 0,

ux (a, y) = f (y), 0 < y < b,

uy (x, b) = 0, 0 x a

(46)
(47)

This is an example of a Neumann Problem. We want to find the fundamental set of solutions.
X 00 X = 0, X 0 (0) = 0
Y 00 + Y = 0, Y 0 (0) = Y 0 (b) = 0.

(48)
(49)

The solution to the equation for Y is

Y (y) = c1 cos(1/2 y) + c2 sin(1/2 y),

(50)

with Y 0 (y) = c1 1/2 sin(1/2 y) + c2 1/2 cos(1/2 y). Using the boundary conditions we find
2 2
c2 = 0 and the eigenvalues are n = nb2 , for n = 1, 2, 3, .... The corresponding Eigenfunctions
are Y (y) = cos( ny
) for n = 1, 2, 3, ... The solution of the equation for X becomes X(x) =
b
nx
d1 cosh( nx
)
+
d
sinh(
), with
2
b
b
X 0 (x) = d1

n
nx
n
nx
sinh(
) + d2
cosh(
).
b
b
b
b
5

(51)

).So the fundamental set of solutions is

Using the boundary conditions, X(x) = d1 cosh( nx
b
un (x, y) = cosh(

nx
ny
) cos(
),
b
b

n = 1, 2, 3, ...

(52)

The general solution is given by

a0 X
nx
ny
u(x, y) =
+
an cosh(
) cos(
)
2
b
b
n=1

(53)