Debnath Nonlinear PDEs 3e: Chapter 1 - Exercise 5 Page 1 of 3

Exercise 5
Solve Example 1.6.2 with the boundary conditions

ux (0, t) = 0 = ux (`, t) for t > 0,

leaving the initial condition (1.6.38) unchanged.

Solution

The temperature distribution u(x, t) in a homogeneous rod of length l with diffusivity constant κ
satisfies the following initial boundary value problem:
ut = κuxx , 0 < x < `, t > 0
ux (0, t) = 0, t>0
ux (`, t) = 0, t>0
u(x, 0) = f (x), 0 ≤ x ≤ `.
The PDE and the boundary conditions are linear and homogeneous, which means that the
method of separation of variables can be applied. Assume a product solution of the form,
u(x, t) = X(x)T (t), and substitute it into the PDE and boundary conditions:

T 0 (t) X 00 (x)
X(x)T 0 (t) = κX 00 (x)T (t) → = =k (1.5.1)
κT (t) X(x)

ux (0, t) = 0 → X 0 (0)T (t) = 0 → X 0 (0) = 0

ux (`, t) = 0 → X 0 (`)T (t) = 0 → X 0 (`) = 0.

The left side of equation (1.5.1) is a function of t, and the right side is a function of x. Therefore,
both sides must be equal to a constant. Values of this constant and the corresponding functions
that satisfy the boundary conditions are known as eigenvalues and eigenfunctions, respectively.
We have to examine three special cases: the case where the eigenvalues are positive (k = µ2 ), the
case where the eigenvalue is zero (k = 0), and the case where the eigenvalues are negative
(k = −λ2 ). The solution to the PDE will be a linear combination of all product solutions.

Case I: Consider the Positive Eigenvalues (k = µ2 )

Solving the ordinary differential equation in (1.5.1) for X(x) gives

X 00 (x) = µ2 X(x), X 0 (0) = 0, X 0 (`) = 0.

X(x) = C1 cosh µx + C2 sinh µx

X 0 (x) = C1 µ sinh µx + C2 µ cosh µx
X 0 (0) = C2 µ = 0 → C2 = 0
0
X (`) = C1 µ sinh µ` = 0 → C1 = 0
X(x) = 0

Positive values of k lead to the trivial solution, X(x) = 0. Therefore, there are no positive
eigenvalues and no associated product solutions.

www.stemjock.com
Debnath Nonlinear PDEs 3e: Chapter 1 - Exercise 5 Page 2 of 3

Case II: Consider the Zero Eigenvalue (k = 0)

Solving the ordinary differential equation for X(x) in (1.5.1) gives

X 00 (x) = 0, X 0 (0) = 0, X 0 (`) = 0.
X(x) = C1 x + C2
X 0 (x) = C1
X 0 (0) = X 0 (`) = C1 → C1 = 0
X(x) = C2
k = 0 leads to a nontrivial solution for X(x), so zero is an eigenvalue. Solving the ordinary
differential equation for T (t), T 0 (t) = 0, gives T (t) = C3 . The product solution associated with
the zero eigenvalue is thus a constant.

Case III: Consider the Negative Eigenvalues (k = −λ2 )

Solving the ordinary differential equation for X(x) in (1.5.1) gives

X 00 (x) = −λ2 X(x), X 0 (0) = 0, X 0 (`) = 0.
X(x) = C1 cos λx + C2 sin λx
X 0 (x) = −C1 λ sin λx + C2 λ cos λx
X 0 (0) = C2 λ = 0 → C2 = 0
X 0 (`) = −C1 λ sin λ` = 0
sin λ` = 0 → λ` = nπ, n = 1, 2, . . .
nπ
X(x) = C1 cos λx λn = , n = 1, 2, . . .
`

2
The eigenvalues are k = −λ2n = − nπ ` , and the corresponding eigenfunctions are
Xn (x) = cos nπx` . Solving the ordinary differential equation for T (t), T 0 (t) = −κλ2 T (t), gives
−κλ 2t
T (t) = C3 e . The product solutions associated with the negative eigenvalues are thus
2
un (x, t) = Xn (x)Tn (t) = e−κλn t cos nπx
` for n = 1, 2, . . ..

According to the principle of superposition, the solution to the PDE is a linear combination of all
product solutions:
∞
nπ 2 nπx
An e−κ( ` ) t cos
X
u(x, t) = A0 + .
`
n=1

The coefficients, A0 and An , are determined from the initial condition, so set t = 0:
∞
X nπx
u(x, 0) = A0 + An cos = f (x). (1.5.2)
`
n=1
To determine A0 , simply integrate both sides of equation (1.5.2) with respect to x from 0 to `.
ˆ ` ˆ `X ∞ ˆ `
nπx
A0 dx + An cos dx = f (x) dx
0 0 n=1 ` 0
∞ ˆ ` ˆ `
X nπx
A0 ` + An cos dx = f (x) dx
0 ` 0
n=1 | {z }
=0

www.stemjock.com
Debnath Nonlinear PDEs 3e: Chapter 1 - Exercise 5 Page 3 of 3

ˆ `
1
A0 = f (x) dx
` 0

To determine An , multiply both sides of equation (1.5.2) by cos mπx

` and integrate both sides with
respect to x from 0 to `. (m is a positive integer.)
∞
mπx X nπx mπx mπx
A0 cos + An cos cos = f (x) cos
` ` ` `
n=1

ˆ ` ˆ `X
∞ ˆ `
mπx nπx mπx mπx
A0 cos dx + An cos cos dx = f (x) cos dx
0 ` 0 n=1 ` ` 0 `
ˆ ` ∞ ˆ ` ˆ `
mπx X nπx mπx mπx
A0 cos dx + An cos cos dx = f (x) cos dx
` ` ` `
|0 {z } n=1 | 0
{z } 0
=0 = `
δ
2 nm
ˆ `
` nπx
An = f (x) cos dx
2 0 `
ˆ `
2 nπx
An = f (x) cos dx
` 0 `
It is thanks to the orthogonality of the trigonometric functions that most terms in the infinite
series vanish upon integration. Only the n = m term remains, and this is denoted by the
Kronecker delta function, (
0 n 6= m
δnm = .
1 n=m

www.stemjock.com