Application of PDE: Solving Laplace’s

equation
José Rafael Ortega
Septiembre 2024

Contain an exercise fully explained solving Laplace’s

equation

KeyWords:
Orthogonal Systems,Method of separation of variables, Bound-
ary Conditions,Laplace’s equation, Coupled equations, Fourier
Series

1 Introduction
Remember that a second order linear homogeneous equation ax′′ +
bx′ + c has solutions:

1
 roots of ar2 + br + c General solution
r1, r2 real and distints x = Aer1t + Ber2t
r = r1 = r2 x = Aert + Btert
r1, r2 complex: α ± βi x = eαt(A cos(βt) + iB sin(βt))
Table 1: Roots of second order equation

Other important facts are the orthogonality relationships of the

sine and cosine functions that are consequence of the following rela-
tions:
1
sin α cos β = [sin (α − β) + sin (α + β)]
2
1
sin α sin β = [cos (α − β) − cos (α + β)]
2
1
cos α cos β = [cos (α − β) + cos (α + β)]
2
in particular we have
n  nπx o∞
sin
L n=1

is mutually orthogonal on 0 ≤ x ≤ L this means that


Z L 
nπx   mπx   L if n = m
sin sin dx = (1)
0 L L
0 if n ̸= m


2
1. Now we proceed to solve the P DE

PDE

∂ 2u 2


 ∂x2
+ ∂∂yu2 = 0
u(x, H) = f1(x)



 u(x, 0) = f2(x)
 u(L, y) = u(0, y) = 0

for that, we separate the problem into two sub-problems

Sub-problem 1.

u1(x, H) = f1(x)

u1(0, y) = 0 u1(L, y) = 0
y

u1(x, 0) = 0
x

3
 Sub-problem 2.

u2(x, H) = 0

u2(0, y) = 0 u2(L, y) = 0
y

u2(x, 0) = f2(x)
x
If u1(x, y) is a solution of problem 1. and u2(x, y)is a solution of
problem 2. then u(x, y) = u1(x, y) + u2(x, y) is a solution of original
problem, because:
∂ 2u ∂ 2u ∂ 2u1 ∂ 2u1 ∂ 2u2 ∂ 2u2
+ = + + + =0
∂x2 ∂y 2 ∂x2 ∂y 2 ∂x2 ∂y 2
and u also verifies the boundary conditions since:

u(x, H) = u1(x, H) + u2(x, H) = f1(x) + 0 = f1(x)

u(x, 0) = u1(x, 0) + u2(x, 0) = 0 + f2(x) = f2(x)
u(0, y) = u1(0, y) + u2(0, y) = 0 + 0 = 0
u(L, y) = u1(L, y) + u2(L, y) = 0 + 0 = 0

2.
Let us now to solve each of this sub-problems.

Sub-problem 1.

4
 
∂ 2 u1 2


 ∂x2
+ ∂∂yu21 = 0

u1(x, H) = f1(x)


 u1(x, 0) = 0
u1(L, y) = u1(0, y) = 0


We set a solution of the form u1(x, y) = F (x)G(y), then we obtain

∂ 2u1 ′′
= F .G (2)
∂x2
∂ 2u1
2
= F.G′′ (3)
∂y
Where the ′ indicates in the equation (1) derivative with respect to
x and in the equation (2) derivative with respect to y. Since
∂ 2u1 ∂ 2u1
+ =0
∂x2 ∂y 2
We must have that
∂ 2u1 ∂ 2u1 ′′ ′′
0= + = F .G + F.G
∂x2 ∂y 2
It is convenient rewrite this equation as
F ′′ G′′
=−
F G
because it becomes clear that we are comparing a function of x with
a function of y, and the only way that this equality might be true is

5
that both of these functions are the same constant,hence
F ′′ G′′
=− = k, k∈R
F G
The boundary conditions are

u1(0, y) = F (0)G(y) = 0
u1(L, y) = F (L).G(y) = 0

Or (F (0) − F (L)).G(y) = 0, which in order to be true excluding the

trivial solution G(y) = 0, become F (0) = F (L) = 0. In other words
the original PDE problem with boundary conditions

Original problem

∂ 2u 2


 ∂x2
+ ∂∂yu2 = 0
u(x, H) = f1(x)



 u(x, 0) = f2(x)
 u(L, y) = u(0, y) = 0

becomes the following system of coupled equations:

Associated ODE.

′′


 F (x) = k.F (x)
F (0) = F (L) = 0



 G′′(y) = −kG(y)


6
We first solve the system for F (x), distinguishing the cases

k > 0, k = 0, k < 0

For k > 0 the general solution of the ODE

 ′′
F (x) = k.F (x)
F (0) = F (L) = 0
is √ √
kx − kx
F (x) = Ae + Be
which is , however, not compatible with the boundary conditions, in
the sense that the only solution of this form satisfying the B.C is
the trivial solution A = B = 0.√ In fact√0 = F (0) = A + B hence
B = −A and then F (x) = A(e kx√ − e− kx√), but then imposing the
other condition
√ 0 = F (L) =√A(e kL − e− kL) if and √ only if either
A = 0 or e2 kL = 1. But 2 kL ̸= 0 and therefore e2 kL ̸= 1. In
consequence A = B = 0
For k = 0 the general solution is F (x) = Ax + B, which is also
not compatible with the boundary conditions unless A = B = 0. It
remains the case k < 0, in which it is convenient to write in the form
k = −λ2 with λ ∈ R λ >. We see from the table 1 that the general
solutions of the equation F ′′(x) + λ2F (x) = 0 are

F (x) = A cos(λx) + B sin(λx)

7
. And we have F (0) = 0 if and only if A = 0, and F (L) = 0 if and
only if B sin(λL) = 0. So if we want non trivial solutions we must
have B ̸= 0 and we need have
nπ
λ=
L
for some integer n ≥ 1

Conclusion.

We have a nontrivial solution for each integer n ≥ 1

( 2 2
kn = − nLπ2
Fn(x) = sin( nπx
L )

The corresponding equation for G is

′′ n2π 2
G (y) = 2 G(y)
L
which according with the table 1 have the general solution
nπy
− nπy
G(y) = Ane L + Bne L (4)

Now since G(0) = 0 we must have for each n, Bn = −An an the

solution must be
nπy
− nπy
G(y) = An(e L −e L ) (5)

8
 ew −e−w
And finally using that sinh(w) = 2 we can write the general
solution as
nπ y
G(y) = Cn(sinh ) (6)
L
with Cn = 2An. The product solution in this case is
nπx nπy
(u1)n(x, y) = sin( ) × Cn(sinh( )) (7)
L L
Using the superposition principle we obtain a solution
∞ ∞
X X nπx nπy
u1(x, y) = (u1)n(x, y) = Cn sin( ) × (sinh( )) (8)
n=1 n=1
L L

Finally, let’s apply the nonhomogeneous boundary condition to get

the coefficients for this solution.
4T1
u1(x, H) = f1(x) = − 2 x(x − L)
L
We have
∞
X nπH nπx
u1(x, H) = f1(x) = Cn sinh( ) sin( ) (9)
n=1
L L

Hence, using the orthogonality relationships for sine functions we get

that
Z L
nπ y nπ H
f1(x) sin( )dy = LCn sinh( )
0 L L

9
Then
  Z L
nπ H 1  n πx 
C n sinh = f1 (x) sin dx n = 1, 2, 3, . . .
L L 0 L
Z L
1  n πx 
Cn = f1 (x) sin dx n = 1, 2, 3, . . .
L sinh n πLH 0


L

Now we use the formula of f1 and we will obtain the exact expression
for Cn
Z L
1 4T1  n πx 
Cn = nπH

− 2 x(x − L) sin dx
L sinh L 0 L L

Or
Z L
4T1 2
 n πx 
Cn = − 3 (x − Lx) sin dx
L sinh n πLH


0 L

Now we integrate by parts, the expression

Z L  n πx 
2
(x − Lx) sin dx
0 L
and we obtain

Z L  n πx  Z L
2 L  nπ x x=L
(x − Lx) sin dx = − (x2 − Lx) cos + (2x − L
0 L nπ | {z L x=0} 0
=0

10
Hence
Z L  n πx  Z L
2 L  n πx 
(x − Lx) sin dx = (2x − L) cos dx
0 L nπ 0 L
Integrating by parts, again

L L
L2 
Z  n πx  Z
2 nπ x x=L n
(x − Lx) sin dx = 2 2 (2x − L) sin( ) − 2 sin
0 L nπ | {z L x=0} 0
=0

we get
Z L 2
 Z L
2
 n πx  L  n πx   2L3 
(x − Lx) sin dx = 2 2 − 2 sin dx = 3 3 cos
0 L n π 0 L nπ
And finally we obtain
Z L
2
 n πx  2L3(−1)n 2L3 2L3 n
(x − Lx) sin dx = 3π3
− 3π3
= 3π3
((−1) − 1)
0 L n n n
Therefore
Z L
4T1 2
 n πx  8T1 1
Cn = − 3 (x − Lx) sin dx = − ((−
L sinh n πLH sinh n πLH n3π 3



0 L
In consequence
 16T1 1

 n3 π 3 sinh nπ H
if n is odd
L
Cn = (10)

0 if n is even


11
 We conclude, finally using (8) that the expression for u1(x, y) is
∞
16T1 X 1 (2n + 1)π x (2n + 1)π y
u1(x, y) = 3 sin( ) × sinh
π n=0 (2n + 1)3 L L
(11)

3.
We proceed as before by separating variables and finding that our
solutions must be of the form
Fn(x)Gn(y)
Where 
nπx
 Fn(x) = sin( L )

and where
 G′′(y) = n2π2 G(y)

L2
But where G must now satisfy the homogeneous boundary condition
G(H) = 0
The most convenient form for the solution is
nπ H − y
G(y) = sinh
L
This means that our general solution, using the superposition prin-
ciple, has the form
∞
X nπx nπ(H − y)
u2(x, y) = Cn sin( ) × (sinh( )) (12)
n=1
L L

12
Setting y = 0 we have
∞
X nπx nπ(H)
u2(x, 0) = f2(x) = Cn sin( ) × (sinh( ))
n=1
L L
P∞
Hence f2(x) = nπx
n=1 Cn sin( L ) × (sinh( nπ(H)
L )) This implies that
we must have
 16T2 1

 n3 π 3 sinh nπ H
if n is odd
L
Cn = (13)

0 if n is even


Because, basically, the integrals involved to calculate the coefficients

for f2 are the same that for f1 And we conclude that
∞
16T2 X 1 (2n + 1)π x (2n + 1)π (H − y)
u2(x, y) = 3 sin( ) × sinh
π n=0 (2n + 1)3 L L
(14)

13