Math 135-2, Homework 7

Solutions

Problem 40.1
Find the eigenvalues λm and Eigenfunctions yn (x) for the equation y ′′ + λy = 0 in each of the
following cases:
(a) y(0) = 0, y(π/2) = 0
(f ) y(a) = 0, y(b) = 0

(a) We are looking for a solution of the form yn (x) = sin(ωn (x − φ)), where λ2 = ωn2 . From yn (0) = 0, we
have φ = 0. From yn (π/2) = 0, we have ωn π/2 = nπ, so that ωn = 2n. Finally, yn (x) = sin(2nx), with
λn = 4n2 .
(b) Begin again with the form yn (x) = sin(ωn (x − φ)). From yn (a) = 0, we have φ = a. From yn (b) = 0, we
nπ nπ nπ 2
have ωn (b − a) = nπ, so that ωn = b−a . Finally, yn (x) = sin( b−a (x − a)), with λn = ( b−a ) .

Problem 40.2
If y = F (x) is an arbitrary function, then y = F (x + at) represents a wave of fixed shape that
moves to the left along the x-axis with velocity a (Fig. 49). Similarly, if y = G(x) is another
arbitrary function, then y = G(x − at) is a wave moving to the right, and the most general
one-dimensional wave with velocity a is

y(x, t) = F (x + at) + G(x − at) (1)

(a) Show that (2) satisfies the wave equation (8).

(b) It is easy to see that the constant a in equation (8) has the dimensions of velocity. Also,
it is intuitively clear that if a stretched string is disturbed, then waves will move in both
directions away from the source of the disturbance. These considerations suggest introducing
the new variables α = x + at and β = x − at. Show that with these independent variables,
equation (8) becomes

∂2y
= 0, (2)
∂α∂β

and from this derive (2) by integration. Formula (2) is called d’Alembert’s solution of the wave
equation. It was also obtained by Euler, independently of d’Alembert but slightly later.

1
(a) Equation (8) is
∂2y ∂2y
a2 2
= 2
∂x ∂t
∂y
= F ′ (x + at) + G′ (x − at)
∂x
∂2y
= F ′′ (x + at) + G′′ (x − at)
∂x2
∂y
= aF ′ (x + at) − aG′ (x − at)
∂t
∂2y
= a2 F ′′ (x + at) + a2 G′′ (x − at)
∂t2
a2 (F ′′ (x + at) + G′′ (x − at)) = (a2 F ′′ (x + at) + a2 G′′ (x − at))
(b)
ŷ(α, β) = ŷ(x + at, x − at) = y(x, t)
∂y ∂ ŷ ∂ ŷ
(x, t) = (x + at, x − at) + (x + at, x − at)
∂x ∂α ∂β
∂2y ∂ 2 ŷ ∂ 2 ŷ ∂ 2 ŷ
(x, t) = (x + at, x − at) + 2 (x + at, x − at) + (x + at, x − at)
∂x2 ∂α2 ∂α∂β ∂β 2
∂y ∂ ŷ ∂ ŷ
(x, t) = a (x + at, x − at) − a (x + at, x − at)
∂t ∂α ∂β
∂2y 2
2 ∂ ŷ
2
2 ∂ ŷ
2
2 ∂ ŷ
(x, t) = a (x + at, x − at) − 2a (x + at, x − at) + a (x + at, x − at)
∂t2 ∂α2 ∂α∂β ∂β 2
∂2y ∂2y
0 = 2 − a2 2
∂t ∂x
2
∂ ŷ
= −4a2 (x + at, x − at)
∂α∂β
∂ 2 ŷ
(x + at, x − at) = 0
∂α∂β
∂ 2 ŷ
Z Z
(α, β) dβ = 0 dβ + F̂ (α)
∂α∂β
∂ ŷ
(α, β) = F̂ (α)
∂α
∂ ŷ
Z Z
(α, β) dα = F̂ (α) dα + G(β)
∂α
ŷ(α, β) = F (α) + G(β)
y(x, t) = ŷ(α, β) = F (α) + G(β) = F (x + at) + G(x − at)

Problem 40.3
Consider an infinite string stretched taut on the x-axis from −∞ to ∞. Let the string be
drawn aside into a curve y = f (x) and released, and assume that its subsequent motion is
described by the wake equation (8).
(a) Use (2) to show that the string’s displacement is given by d’Alembert’s formula,
1
y(x, t) = [f (x + at) + f (x − at)]. (3)
2

2
Hint: remember the initial conditions (11) and (12).
(b) Assume further that the string remains motionless at the points x = 0 and x = π (such
points are called nodes), so that y(0, t) = y(π, t) = 0, and use (3) to show that f (x) is an odd
function that is periodic with period 2π [that is, f (−x) = −f (x) and f (x + 2π) = f (x)].
(c) Show that since f (x) is odd and periodic with period 2π, it necessarily vanishes at 0 and π.
(d) Show that Bernoulli’s solution (17) can be written in the form of (3). Hint: 2 sin nx cos nat =
sin[n(x + at)] + sin[n(x − at)].

(a)

f (x) = y(x, 0) = F (x) + G(x)

∂y
0= (x, 0) = aF ′ (x) − aG′ (x)
∂t
F (x) = G(x) + c
f (x) = 2G(x) + c
1
G(x) = (f (x) − c)
2
1
F (x) = (f (x) + c)
2
1
y(x, t) = [f (x + at) + f (x − at)]
2
(b)
1
0 = y(0, t) = [f (at) + f (−at)]
2
f (−at) = −f (at)
f (−u) = −f (u) u = at
1
0 = y(π, t) = [f (π + at) + f (π − at)]
2
f (π + at) = −f (π − at)
= f (at − π)
f (v + 2π) = f (v) v = at − π

(c)

f (0) = −f (−0)
f (0) = 0
f (π) = −f (−π)
= −f (−π + 2π)
f (π) = 0

3
(d)
∞
X
y(x, t) = bn sin nx cos nat
n=1
∞
X 1
= bn (sin[n(x + at)] + sin[n(x − at)])
n=1
2
∞ ∞
1X 1X
= bn sin[n(x + at)] + bn sin[n(x − at)]
2 n=1 2 n=1
1
= [f (x + at) + f (x − at)]
2
∞
X
f (x) = bn sin nx
n=1

Problem 41.4
In the preceding problem, find w(x, t) if the ends of the rod are kept at 0◦ C, w0 = 0◦ C, and
the initial temperature distribution is f (x).

∂2w ∂w
a2 = + c(w − w0 ) w(0, t) = w(π, t) = 0
∂x2 ∂t
2
∂ w ∂w
a2 2 = + cw
∂x ∂t
w(x, t) = X(x)T (t)
a2 X ′′ (x)T (t) = X(x)T ′ (t) + cX(x)T (t)
X ′′ (x) T ′ (t)
a2 = +c=k
X(x) T (t)
a2 X ′′ (x) − kX(x) = 0
X(x) = sin nx
k = −a2 n2
′
T (t)
+c=k
T (t)
T ′ (t) = −(a2 n2 + c)T (t)
2 2
T (t) = e−(a n +c)t
∞
X 2 2
w(x, t) = an e−(a n +c)t sin nx
n=1
X∞
f (x) = w(x, 0) = an sin nx
n=1
π
2
Z
an = f (x) sin nx dx
π 0

4
Problem 41.7
The two-dimensional heat equation is
 2
∂ w ∂ 2w ∂w

a2 + =
∂x2 ∂y 2 ∂t
Use the method of separation of variables to find a steady-state solution of this equation in
the infinite strip of the xy-plane bounded by the lines x = 0 and x = π, and y = 0 if the
following conditions are satisfied:

w(0, y) = 0 w(π, y) = 0 w(x, 0) = f (x) lim w(x, y) = 0

y→∞

First, lets find solutions to the problem, ignoring the condition w(x, 0) = f (x).

∂w
=0 (steady state)
∂t
∂2w ∂2w
+ =0
∂x2 ∂y 2
w(x, y) = X(x)Y (y)
∂2w
= X ′′ (x)Y (y)
∂x2
∂2w
= X(x)Y ′′ (y)
∂y 2
X ′′ (x)Y (y) + X(x)Y ′′ (y) = 0
X ′′ (x) Y ′′ (y)
+ =0
X(x) Y (y)
X ′′ (x)
=c
X(x)
Y ′′ (y)
= −c
Y (y)

Lets first consider the equation in x.

X ′′ (x) = cX(x) X(0) = X(π) = 0

X(x) = sin nx
c = −n2

Note that c ≥ 0 leads to X(x) = 0. Next, lets consider y.

Y ′′ (y) = −n2 Y (y) lim Y (y) = 0

y→∞

Y (y) = e−ny

We can now assemble solutions of this form to form a more general solution, imposing the requirement

5
w(x, 0) = f (x).
∞
X
w(x, y) = an sin nxe−ny
n=1
X∞
f (x) = w(x, 0) = an sin nx
n=1
π
2
Z
an = f (x) sin nx dx
π 0

Problem A
The three-dimensional heat equation is
 2
∂ w ∂2w ∂2w ∂w

a2 + + = .
∂x 2 ∂y 2 ∂z 2 ∂t

This describes a block of material, in this case [0, π] × [0, π] × [0, π]. The initial condi-
tions are w(x, y, z, 0) = f (x)g(y)h(z). Boundary conditions are w(0, y, z, t) = w(π, y, z, t) =
w(x, 0, z, t) = w(x, π, z, t) = w(x, y, 0, t) = w(x, y, π, t) = w1 . Solve for w(x, y, z, t).

Note that I can write w(x, y, z, t) = w1 + ŵ(x, y, z, t), so that ŵ(0, y, z, t) = ŵ(π, y, z, t) = ŵ(x, 0, z, t) =
ŵ(x, π, z, t) = ŵ(x, y, 0, t) = ŵ(x, y, π, t) = 0.
Following separation of variables,

ŵ(x, y, z, t) = X(x)Y (y)Z(z)T (t)

2
∂ ŵ ∂ ŵ ∂ 2 ŵ
2
 
∂ ŵ
a2 2
+ 2
+ 2
=
∂x ∂y ∂z ∂t
a2 (X ′′ (x)Y (y)Z(z)T (t) + X(x)Y ′′ (y)Z(z)T (t) + X(x)Y (y)Z ′′ (z)T (t)) = X(x)Y (y)Z(z)T ′ (t)
 ′′ 
X (x) Y ′′ (y) Z ′′ (z) T ′ (t)
a2 + + =
X(x) Y (y) Z(z) T (t)
X ′′ (x) = kx X(x)
Y ′′ (y) = ky Y (y)
Z ′′ (z) = kz Z(z)
T ′ (t) = a2 (kx + ky + kz )T (t)

6
If kx ≥ 0, the only solution is X(x) = 0. Thus, kx < 0. Similar conclusions follow for Y (y) and Z(z).

X(x) = sin mx kx = −m2

Y (y) = sin ny ky = −n2
Z(z) = sin pz kz = −p2
T ′ (t) = −a2 (m2 + n2 + p2 )T (t)
2
(m2 +n2 +p2 )t
T = e−a
∞ X ∞
∞ X
2
(m2 +n2 +p2 )t
X
w(x, y, z, t) = w1 + am,n,p e−a sin mx sin ny sin pz
m=1 n=1 p=1
X∞ X ∞ X ∞
f (x)g(y)h(z) = w(x, y, z, 0) = w1 + am,n,p sin mx sin ny sin pz
m=1 n=1 p=1
X ∞
∞ X
∞ X
f (x)g(y)h(z) − w1 = am,n,p sin mx sin ny sin pz
m=1 n=1 p=1

Observe that f (x)g(y)h(z) − w1 is the difference of separable things. This suggests letting am,n,p =
Mm Nn Pp − Am Bn Cp .

X ∞
∞ X
∞ X
f (x)g(y)h(z) = Mm Nn Pp sin mx sin ny sin pz
m=1 n=1 p=1
∞
! ∞
! ∞
!
X X X
= Mm sin mx Nn sin ny Pp sin pz
m=1 n=1 p=1
∞
X
Mm sin mx = f (x)
m=1
X∞
Nn sin ny = g(y)
n=1
∞
X
Pp sin pz = h(z)
p=1
2 π
Z
Mm = f (x) sin mx dx
π 0
Z π
2
Nn = g(y) sin ny dy
π 0
Z π
2
Pp = h(z) sin pz dz
π 0

7
 X ∞
∞ X
∞ X
1= Am Bn Cp sin mx sin ny sin pz
m=1 n=1 p=1
∞
! ∞
! ∞
!
X X X
= Am sin mx Bn sin ny Cp sin pz
m=1 n=1 p=1
∞
X
Am sin mx = 1
m=1
X∞
Bn sin ny = 1
n=1
∞
X
Cp sin pz = 1
p=1
π
2
Z
Am = Bm = Cm = sin mx dx
π 0
A2m = B2m = C2m =0
4
A2m+1 = B2m+1 = C2m+1 =
(2m + 1)π

∞ X ∞
∞ X
2
(m2 +n2 +p2 )t
X
w(x, y, z, t) = w1 + (Mm Nn Pp − Am Bn Cp )e−a sin mx sin ny sin pz
m=1 n=1 p=1

Problem B
Let φ1 , φ2 , φ3 , . . . and θ1 , θ2 , θ3 , . . . be two normalized orthogonal sequences, in the sense of
Z b (
1 m=n
hφm , φn ix = φm (x)φn (x) dx =
a 0 m 6= n
Z d (
1 m=n
hθm , θn iy = θm (y)θn (y) dy =
c 0 m 6= n

Let σi,j (x, y) = φi (x)θj (y) be a sequence of functions in two dimensions (it does not matter
the order in which they are enumerated).
(a) It is possible to construct an inner product hσi,j , σm,n ixy under which this sequence is
normalized and orthogonal. Construct this inner product and show that the sequence is
normalized and orthogonal with respect to it.
(b) It is desired to express f (x, y) as in terms of this basis. That is,
∞ X
X ∞
f (x, y) = ai,j σi,j (x, y).
i=1 j=1

How should ai,j be selected?

(c) Show how the solution from Problem A can be modified to handle the more general initial
condition w(x, y, z, 0) = r(x, y, z). You may assume the sequence σi,j is complete.

8
(a) Define
Z b Z d
hσi,j , σm,n ixy = σi,j (x, y)σm,n (x, y) dy dx
a c
Z b Z d
= φi (x)θj (y)φm (x)θn (y) dy dx
a c
! Z !
Z b d
= φi (x)φm (x) dx θj (y)θn (y) dy
a c

= hφi , φm ix hφj , φn iy
(
1 (i, j) = (m, n)
=
0 (i, j) 6= (m, n)

(b) This is an orthogonal and normalized sequence; fitting is done by choosing the Fourier coefficients.

ai,j = hf, σi,j ixy

(c) The coefficients should be chosen as the Fourier coefficients of the product basis.
Z πZ πZ π
am,n,p = r(x, y, z) sin mx sin ny sin pz dz dy dx
0 0 0