Laplacian and Laplace’s Equation

Laplacian
∂2u ∂2u
Let u(x, y) be a scalar function of two variables such that the second order partial derivatives and
∂x2 ∂y 2
exist. Then, the Laplacian of the function u is

∂2u ∂2u
∇2 u = + 2 . (Two-dimensional Laplacian)
∂x2 ∂y
∂2u ∂2u
If u(x, y, z) is a scalar function of three variables such that the second order partial derivatives ,
∂x2 ∂y 2
∂2u
and exist, then the Laplacian of the function u is
∂z 2
∂2u ∂2u ∂2u
∇2 u = + 2 + 2 . (Three-dimensional Laplacian)
∂x2 ∂y ∂z

Laplace’s Equation
Definition. The partial differential equation ∇2 u = 0 is called Laplace’s equation.

In two dimensions, Laplace’s equation is

∂2u ∂2u
+ 2 =0 (or uxx (x, y) + uyy (x, y) = 0)
∂x2 ∂y
and in three dimensions Laplace’s equation is
∂2u ∂2u ∂2u
+ 2 + 2 =0 (or uxx + uyy + uzz = 0).
∂x2 ∂y ∂z
Definition. A function satisfying Laplace’s equation in a region in plane or in space is said to be harmonic
on that region.

Example. Let u(x, y) = x2 − y 2 . Then ux = 2x, uy = −2y, uxx = 2 and uyy = −2. So, we have
uxx + uyy = 2 − 2 = 0 for all (x, y) ∈ R2 . Therefore, u(x, y) = x2 − y 2 is harmonic over the entire plane.

Dirichlet problem
A Dirichlet problem for a region D consists of finding a function that is harmonic on D and assumes specified
values on the boundary of D. We will denote by ∂D the boundary of the region D.

1
The Dirichlet problem for D is to solve

2
 ∇ u = 0 on D



u(x, y) = f (x, y) for (x, y) in ∂D


with f (x, y) as a given function. The function f (x, y) is called boundary data for D.

In order to solve the Dirichlet problem, we will use the method of separation of variables, and in this tech-
nique we will need solutions of eigenvalue problems.

Eigenvalue Problems
Definition. An ordinary differential equation (ODE)

y 00 + p(x)y 0 + q(x)y = h(x)

together with the conditions

y(α) = y0 and y(β) = y1 ,
where α, β, y0 , y1 are real numbers, α 6= β, and p(x), q(x), h(x) are continuous functions, is called a
two-point boundary value problem (BVP), and the conditions y(α) = y0 and y(β) = y1 are called boundary
conditions.
d2 y
 
00 0 dy
Here, y = 2 and y = .
dx dx
Let us consider the two-point boundary value problem

y 00 + λy = 0, y(0) = 0, y(L) = 0 (1)

where λ is a real number and L is a positive number.

The solution y = 0 of the BVP (1) is called trivial solution. Our purpose is to find non-trivial solutions
(solutions that are different from zero). We need to consider separately the cases λ > 0, λ = 0, and λ < 0.

Case i : λ > 0

The characteristic equation of the ODE y 00 + λy = 0 is

r2 + λ = 0
√
with roots r = ±i λ (complex conjugate roots). So, the general solution is
√
√

y = c1 cos λ x + c2 sin λ x ,

where c1 and c2 are arbitrary real constants.

The boundary condition y(0) = 0 implies that c1 cos 0 + c2 sin 0 = 0. Therefore, c1 = 0 so that
√

y = c2 sin λ x .

2
 √

Now, using y(L) = 0, we obtain that c2 sin λ L = 0. Since we want to find non-trivial solutions, we must
√
√
require that c2 6= 0. Therefore, sin λ L = 0 so that λ L = nπ, where n is a positive integer. Thus, the
possible values of λ are
 nπ 2 n2 π 2
λn = = , n = 1, 2, 3, . . .
L L2
The numbers λn , n = 1, 2, 3, . . ., are called eigenvalues of the BVP (1). So, any constant multiple of the
functions p
 nπx 
yn = sin λn x = sin , n = 1, 2, 3, . . .
L
  nπ      
2nπ 3nπ
are non-trivial solutions of the BVP (1). y1 = sin , y2 = sin , y3 = sin ,...
L L L
The non-trivial solutions yn , n = 1, 2, 3, . . ., are called eigenfunctions.

Note that any constant multiple of these eigenfunctions is again an eigenfunction.

Case ii : λ < 0.
√
The characteristic equation r2 + λ = 0 of y 00 + λy = 0 has the roots r = ± −λ. Thus, the general solution is
√
√

y = c1 cosh −λ x + c2 sinh −λ x ,

where c1 and c2 are arbitrary real constants.

ex + e−x ex − e−x
 
Recall that cosh x = and sinh x =
2 2
√

Since y(0) = 0, we have c1 cosh 0 + c2 sinh 0 = 0. Thus, c1 = 0 so that y = c2 sinh −λ x .
√
√

Next, using y(L) = 0, we obtain that c2 sinh −λ L = 0. Since λ 6= 0, it follows that sinh −λ L 6= 0,
and therefore, we must have c2 = 0.

Consequently, y = 0 is the only solution and there are no trivial solutions for λ < 0. (i.e., the problem has
no negative eigenvalues)

Case iii : λ = 0.

If λ = 0, then we have the ODE y 00 = 0. The general solution is y = c1 x + c2 . Using y(0) = 0, we have
c2 = 0. The condition y(L) = 0 implies that c1 L = 0 so that c1 = 0. Thus, the only solution is the trivial
solution y = 0. ( that is, λ = 0 is not an eigenvalue. )

n2 π 2
Summary : λn = , n = 1, 2, 3, . . . , are the eigenvalues and the corresponding eigenfunctions are
 nπx  L2
yn = sin , n = 1, 2, 3, . . .
L
 nπx 
There are no other real eigenvalues. For any real constant c, y = c sin is a non-trivial solution of the
00
L
BVP y + λy = 0, y(0) = 0, y(L) = 0.

3
Remark : Let L be a positive real number.

(i) If m is an integer, then

ZL  mπx  ZL  !
1 2mπx
sin2 dx = 1 − cos dx
L 2 L
0 0
 ! L
1 L 2mπx
= x− sin
2 2mπ L
0
 
1 L
= L− sin (2mπ)
2 2mπ
L
= .
2
(ii) If m and n are integers such that m 6= n, then
ZL ZL "    #
 mπx   nπx  1 (m − n)πx (m + n)πx
sin sin dx = cos − cos dx
L L 2 L L
0 0
"    # L
1 L (m − n)πx L (m + n)πx
= sin − sin
2 (m − n)π L (m + n)π L
0
 
1 L   L 
= sin (m − n)π − sin (m + n)π
2 (m − n)π (m + n)π
= 0.
Consequently, if m and n are integers, then we have
ZL  mπx   nπx 
(
0, if m 6= n
sin sin dx = L
L L , if m = n
0 2

Dirichlet Problem for a Rectangle

Let us consider the problem of finding the function u(x, y) satisfying the Laplace’s equation
uxx + uyy = 0 (2)
in the rectangle D = (x, y) ∈ R2 | 0 < x < L, 0 < y < K and also satisfying the boundary conditions


u(x, 0) = 0 for 0 ≤ x ≤ L, u(0, y) = 0 for 0 ≤ y ≤ K, u(L, y) = 0 for 0 ≤ y ≤ K, and u(x, K) = f (x) for
0 ≤ x ≤ L, where f (x) is a given nonzero function defined for 0 ≤ x ≤ L.

To solve this problem, we will use the method of separation of variables to construct a fundamental set of
solutions satisfying the Laplace’s equation and homogeneous boundary conditions, then we will superpose
these solutions so as to satisfy the remaining boundary condition u(x, K) = f (x).

Let us assume that u(x, y) = X(x)Y (y), where X(x) and Y (y) are both nonzero functions. Then,
2X 2Y
 
d d
uxx = X 00 (x)Y (y) and uyy = X(x)Y 00 (y). Here, X 00 (x) = and Y 00 (y) =
dx2 dy 2

4
Substituting these derivatives in the Laplace’s equation (2), we obtain that X 00 (x)Y (y) + X(x)Y 00 (y) = 0.
So,
X 00 (x) Y 00 (y)
=− .
X(x) Y (y)
X 00 (x) Y 00 (y)
is a function of x only and − is a function of y only. Therefore,
X(x) Y (y)
X 00 (x) Y 00 (y)
=− = −λ
X(x) Y (y)
for some real constant λ. Hence,
X 00 + λX = 0, (3)
and
Y 00 − λY = 0. (4)
The given boundary conditions u(x, 0) = 0, u(0, y) = 0 and u(L, y) = 0 imply that
X(x)Y (0) = 0, X(0)Y (y) = 0, and X(L)Y (y) = 0.
So, we have X(0) = 0, X(L) = 0 and Y (0) = 0 (since we want to determine non-trivial solutions).

We will first determine the solution of ODE (3) subject to the boundary conditions X(0) = 0 and X(L) = 0.

We have previously verified that this BVP has non-trivial solutions if and only if λ is an eigenvalue, namely,
n2 π 2
λ= , n = 1, 2, 3, . . .
L2
 nπx 
and only constant multiple of the corresponding eigenfunction Xn (x) = sin is a non-trivial solution
L 
00 nπx 
of the BVP X + λX = 0, X(0) = 0, X(L) = 0. (i.e. for any c ∈ R, X(x) = c sin , n = 1, 2, 3, . . . ,
L
is a non-trivial solution. )

n2 π 2
Next, in ODE (4) we substitute λ = to obtain
L2
n2 π 2
Y 00 − Y = 0. (5)
L2
n2 π 2 nπ nπ
The characteristic equation of the ODE (5) is r2 − = 0 with roots r1 = and r2 = − .
L2 L L
 nπy   nπy 
The general solution of the ODE (5) is Y (y) = k1 cosh +k2 sinh , where k1 and k2 are arbitrary
L L
real constants.
 nπy 
Using the condition Y (0) = 0, we obtain k1 = 0. So, Y (y) = k2 sinh , i.e., Y (y) is proportional to the
 nπy  L
function sinh . Thus, we obtain the fundamental solutions
L
 nπx   nπy 
un (x, y) = Xn (x)Yn (y) = sin sinh , n = 1, 2, 3, . . .
L L
To satisfy the boundary condition u(x, K) = f (x), 0 ≤ x ≤ L, we use a superposition to represent the
solution u(x, y) in the form
∞
X ∞
X  nπx   nπy 
u(x, y) = cn un (x, y) = cn sin sinh
L L
n=1 n=1

5
where cn , n = 1, 2, 3, . . . , are real constants.

Using u(x, K) = f (x), we have

∞  
X nπK  nπx 
f (x) = cn sinh sin .
L L
n=1

For each positive integer m, the equation

ZL  mπx  ∞    ZL 
X nπK  mπx   nπx 
f (x) sin dx = cn sinh sin sin dx
L L L L
0 n=1 0

is satisfied. Using
ZL  mπx   nπx 
(
0, if m 6= n
sin sin dx = L
L L , if m = n
0 2
we obtain that
ZL  mπx   
L mπK
f (x) sin dx = cm sinh .
L 2 L
0

So,
ZL
2  nπx 
cn =  nπK  f (x) sin dx.
L
L sinh 0
L
Consequently, the solution is
∞
X  nπx   nπy 
u(x, y) = cn sin sinh ,
L L
n=1

where
ZL
2  nπx 
cn =   f (x) sin dx, n = 1, 2, 3, . . .
nπK L
L sinh 0
L
Example. Let D = (x, y) ∈ R2 | 0 < x < π, 0 < y < 2π . Find the solution of the Laplace’s equation


uxx + uyy = 0 for (x, y) in D subject to the boundary conditions

u(0, y) = 0, u(π, y) = 0, for 0 ≤ y ≤ 2π,

u(x, 0) = 0, u(x, 2π) = sin x + sin(3x), for 0 ≤ x ≤ π.

Solution. Let u(x, y) = X(x)Y (y). Then,

X 00 (x)Y (y) + X(x)Y 00 (y) = 0

So,
X 00 (x) Y 00 (y)
=− = −λ
X(x) Y (y)
for some real constant λ. Thus,
X 00 + λX = 0,

6
and
Y 00 − λY = 0.
√
If λ > 0,√then the characteristic equation of the ODE X 00 + λX = 0 is r2 + λ = 0 with roots r1 = i λ and
r2 = −i λ. The general solution of X 00 + λX = 0 is
√
√

y = α1 cos λ x + α2 sin λ x .
√

Since X(0) = 0, we obtain that α1 = 0, and similarly, since X(π) = 0, we obtain that α2 sin λ π = 0.
√
To obtain a non-trivial solution, we must have sin λ π = 0. So, λ = n2 , n = 1, 2, 3, . . . , (these are

eigenvalues), and any constant multiple of the functions Xn (x) = sin(nx) are solutions of X 00 + λX = 0,
X(0) = 0, X(0) = 0.

There are no non-trivial solutions for λ ≤ 0.

Next, consider the ODE Y 00 − n2 Y = 0. The characteristic equation is r2 − n2 = 0 with roots r1 = n and
r2 = −n. The general solution is Y (y) = β1 cosh(ny) + β2 sinh(ny). Using Y (0) = 0, we have β1 = 0. So,
the solutions are proportional to Yn (y) = sinh(ny). Using superposition, we have
∞
X ∞
X
u(x, y) = cn Xn (x)Yn (y) == cn sin(nx) sinh(ny).
n=1 n=1

The condition u(x, 2π) = sin x + sin(3x) implies that

∞
X
sin x + sin(3x) = cn sinh(2πn) sin(nx).
n=1

Thus, for any positive integer m we have

Zπ Zπ ∞
X  Zπ 
sin x sin(mx)dx + sin(3x) sin(mx)dx = cn sinh(2πn) sin(mx) sin(nx)dx .
0 0 n=1 0

Using
Zπ (
0, if m 6= n
sin(mx) sin(nx)dx = π
, if m = n
0 2
we obtain
Zπ Zπ
π
sin x sin(mx)dx + sin(3x) sin(mx)dx = cm sinh(2πm).
2
0 0

If m = 1, then
π π
c1 sinh(2π) =
2 2
1
so that c1 = .
sinh(2π)
If m = 3, then
π π
c3 sinh(6π) =
2 2
1
so that c3 = .
sinh(6π)

7
If m 6= 1 and m 6= 3, then
π
cm sinh(2πm) = 0
2
so that cm = 0.

Thus, the solution is

1 1
u(x, y) = sin x sinh y + sin(3x) sinh(3y).
sinh(2π) sinh(6π)