Homework 7 Solutions

#1 (Section 10.4): The following functions are defined on an interval of length L. Sketch the even and
odd extensions of each function over the interval [L, L].

(a)
f (x)

f (x) = 1 x2 , 0x1

0 1
x

Even extension of f(x) Odd extension of f(x)

1
1

f (x)
o
f (x)
e

-1 0 1 -1 0 1
x x

(b)
f (x)

x, 0 x < 2
f (x) = 1

1, 2x3

0
0 1 2 3
x

Even extension of f(x) Odd extension of f(x)

2 2

f (x) 1
e
f (x)
o

1 0

-1

0 -2
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
x x
#2 (Section 10.4): The following functions are defined on an interval of length L. Find the Fourier Sine
series and the Fourier Cosine series for each and sketch the functions to which the
series converge over the interval [3L, 3L].

1
(a)
f (x)

f (x) = 1, 0x1

0 1
x

Fourier Cosine series

Even extension of f(x)

fe (x) L=1

bn = 0, n = 1, 2, . . .

-1 0 1
x

The Fourier coefficients are calculated as follows.

L 1
2 2
Z Z
a0 = f (x) dx = 1 dx = 2
L 0 1 0

a0 = 2

L 1
2 2
Z  nx  Z
an = f (x) cos dx = 1 cos(nx) dx
L 0 L 1 0
1
2  2
= sin(nx) = [sin(n) sin 0] = 0
n 0 n

an = 0 n = 1, 2, . . .
 Then,

a0 X  nx 
f (x) = FC (x) = + an cos for 0 < x < L
2 n=1
L

2 X
= + 0 cos(nx)
2 n=1

f (x) = FC (x) = 1 0<x<1

Fourier Cosine series representation of f(x)

FC (x)

-3 -2 -1 0 1 2 3
x
 Fourier Sine series
Odd extension of f(x)
1

f (x)
o
L=1
0
an = 0, n = 0, 1, 2, . . .

-1
-1 0 1
x

The Fourier coefficients are calculated as follows.

2 L  nx  2 1
Z Z
bn = f (x) sin dx = 1 sin(nx) dx
L 0 L 1 0
1
2 2 2
[1 (1)n ]

= cos(nx) = [cos(n) cos 0] =
n 0 n n

2 1 (1)n
bn = n = 1, 2, . . .
n
Then,

X  nx 
f (x) = FS (x) = bn sin for 0 < x < L
n=1
L

2X
1 (1)n
f (x) = FS (x) = sin(nx) 0<x<1
n=1 n

Fourier Sine series representation of f(x)

1

FS (x)

-1
-3 -2 -1 0 1 2 3
x
(b)
1

1, 0x< f (x)

f (x) =
0, x 2

0 pi 2*pi
x

Fourier Cosine series

Even extension of f(x)

fe (x) L = 2

bn = 0, n = 1, 2, . . .

-2*pi -pi 0 pi 2*pi

x

The Fourier coefficients are calculated as follows.

L 2 Z 2 
2 2 1 1
Z Z Z
a0 = f (x) dx = f (x) dx = 1 dx + 0 dx = ( + 0) = 1
L 0 2 0 0

a0 = 1

L Z 2
2 2
Z  nx   nx 
an = f (x) cos dx = f (x) cos dx
L 0 L 2 0 2
Z Z 2  nx  
1  nx 
= 1 cos dx + 0 cos dx
0 2 2

1

2  nx  
2 h  n  i 2  n 
= sin  + 0 = sin sin 0 = sin
n 2 0 n 2 n 2

2 n
 
an = sin n = 1, 2, . . .
n 2
 Then,

a0 X  nx 
f (x) = FC (x) = + an cos for 0 < x < L
2 n=1
L

1 2X 1 n nx
   
f (x) = FC (x) = + sin cos 0 < x < 2
2 n=1 n 2 2

Fourier Cosine series representation of f(x)

FC (x)

-6*pi -4*pi -2*pi 0 2*pi 4*pi 6*pi

x
 Fourier Sine series
Odd extension of f(x)
1

f (x)
o

L = 2
0
an = 0, n = 0, 1, 2, . . .

-1
-2*pi -pi 0 pi 2*pi
x

The Fourier coefficients are calculated as follows.

Z 2
2 L  nx  2  nx 
Z
bn = f (x) sin dx = f (x) sin dx
L 0 L 2 0 2
Z Z 2  nx  
1  nx 
= 1 sin dx + 0 sin dx
0 2 2

1

2  nx  
2 h  n  i 2 h  n i
= cos  + 0 = cos cos 0 = 1 cos
n 2 0 n 2 n 2

2 n
  
bn = 1 cos n = 1, 2, . . .
n 2
Then,

X  nx 
f (x) = FS (x) = bn sin for 0 < x < L
n=1
L

2X 1 n nx
    
f (x) = FS (x) = 1 cos sin 0<x<1
n=1 n 2 2

Fourier Sine series representation of f(x)

1

F (x)
S

-1
-6*pi -4*pi -2*pi 0 2*pi 4*pi 6*pi
x
#3 Section 10.5: (page 610) # 7, 9 (7th edition: pages 579580 # 7,9)

Problems 7 and 9 both involve solving a heat conduction problem of the form

ut = 2 uxx , 0 < x < L, t>0 (1)

u(0, t) = 0, u(L, t) = 0, t>0 (2)

u(x, 0) = f (x), 0 x L, (3)
where , L, and f (x) are given. It is efficient to first solve the general case, and then substitute the
specific quantities given in problems 7 and 9.

Separation of Variable Method:

Let u(x, t) = X(x) T (t).

Plugging this expression for u into the PDE (1) gives

X(x) T 0 (t) = 2 X 00 (x)T (t).

Divide by 2 X(x) T (t) to get

1 T 0 (t) X 00 (x)
= .
2 T (t) X(x)
Since the left hand side depends on t alone and the right hand side depends on x alone, they can be
equal only when both are equal to the same constant, call it .

1 T 0 (t) X 00 (x)
= = .
2 T (t) X(x)

This expression separates into two equations: one for T (t) and one for X(x),

T 0 + 2 T = 0, X 00 + X = 0. (4)

We can use (2) to determine boundary conditions for X(x). At x = 0,

u(0, t) = X(0) T (t) = 0,

Since this equality holds for all t, it must be that

X(0) = 0. (5)

At x = L,
u(L, t) = X(L) T (t) = 0,
and by the same argument we have
X(L) = 0. (6)
Equations (4)(6) suggest the following eigenvalue problem for X(x).

X 00 + X = 0

X(0) = 0, X(L) = 0.

We solved this eigenvalue problem in Section 10.1 (see lecture notes) and found the following eigenvalues
and eigenfunctions.
 2
n = n L
n = 1, 2, . . .
nx
 
Xn (x) = sin L
For each n , equation (4) can be used to determine the corresponding solution Tn (t),

Tn0 + 2 n Tn = 0.

The is the familiar first order exponential ODE with a decay rate of 2 n . It is both separable and
linear so it can be solved by either the separable variable technique or the integrating factor method.
In either case, we get
n 2
Tn (t) = Cn e n t = Cn e( L ) t .
2

We have found infinitely many solutions of the PDE which are given by

un (x, t) = Xn (x) Tn (t) = e( n

L )2 t sin nx

, n = 1, 2. . . . .
L

It is easy to check that each solution satisfies the PDE (1) and the boundary conditions (2). The last
step is to satisfy initial condition (3). Note that any linear combination of the un s also satisfies the
PDE and the boundary conditions (principle of superposition). We can attempt to satisfy the initial
conditions by considering the linear combination of all the un s given by

2
Bn e( ) t sin nx
n
X  
u(x, t) = L
L . (7)
n=1

Applying the initial condition (3) yields

nx
X  
u(x, 0) = Bn sin = f (x).
L
n=1

By comparing this equation to the Fourier Sine series of f (x) on the interval (0, L), it can be seen that
the Bn s are the Fourier Sine series coefficient, and therefore

2ZL
f (x) sin nx
 
Bn = L dx . (8)
L 0
Then the solution given by equations (7) and (8) satisfies the PDE, the boundary conditions, and the
initial conditions.
Now equations (7) and (8) can be used to solve problems 7 and 9.

#7 In this problem, equations (1)(3) are given with

L = 1, = 10, f (x) = sin(2x) sin(5x).

Then by equation (7), the solution is

2
Bn e(10n) t sin(nx),
X
u(x, t) =
n=1

where the Bn s are the coefficients of the Fourier Sine series of f(x) on the interval (0,1). Though
the Bn s can be calculated using equation (8), the initial condition has a special form and it is
more convenient to apply the initial condition directly to get

X
u(x, 0) = Bn sin(nx) = sin(2x) sin(5x)
n=1

The Bn s can be determined by equating the coefficients of like terms (sin(nx), n = 1, 2, ...),
which yields
B2 = 1, B5 = 1, and Bn = 0 otherwise.
Therefore all terms of the series are zero except for the second (n = 2) and fifth (n = 5) terms,
so the solution is
2 2
u(x, t) = e(20) t sin(2x) e(50) t sin(5x) .

#9 In this problem, equations (1)(3) are given with

L = 40, = 1, f (x) = 50.

Then by equation (7), the solution is

n2 2 nx
 
Bn e 1600 t sin
X
u(x, t) = 40 , (9)
n=1

where the Bn s are the coefficients of the Fourier Sine series of f(x) on the interval (0,40). The
Bn s are determined from equation (8).
L 40
2 2
Z Z
nx nx
   
Bn = f (x) sin L dx = 50 sin 40 dx
L 0 40 0

5 40 40 100 100
nx 

= cos 40  = [cos(n) cos 0] = [1 (1)n ]
2 n 0 n n

100
[1 (1)n ] , n = 1, 2, . . .
Bn = (10)
n
The solution is given by equations (9) and (10) and can be written as

100 X 1 (1)n n2 2 t
e sin nx
 
u(x, t) = 1600
40 .
n=1 n
#4 Section 10.6: Consider the heat flow problem

ut = uxx , 0 < x < 2, t>0 (11)

u(0, t) = 4, u(2, t) = 0, t>0 (12)

u(x, 0) = 2x, 0 x 2. (13)

(a) Find the steady state (time independent) solution, call it v(x), by setting ut = 0 in the PDE and
solving the resulting boundary value problem.

With ut = 0 and u(x, t) = v(x), equations (11) and (12) reduce to

vxx = 0, 0 < x < 2, (14)

v(0) = 4, v(2) = 0. (15)

The general solution of equation (14) is v(x) = C1 +C2 x. The boundary conditions (15) determine
the constants to be C1 = 4 and C2 = 2, and therefore

v(x) = 2 (2 x) .

(b) Let w(x, t) = u(x, t) v(x), where v(x) is the steady state solution found in part (a). Find the
resulting PDE, boundary conditions, and initial condition for w(x, t).

The temperature u(x, t) can be written as u(x, t) = w(x, t) + v(x) and substituted into equation
(11) to give
wt + vt = wxx + vxx ,
and since vt = vxx = 0, we get
wt = wxx .
Equation (12) determines the boundary conditions for w.

w(0, t) = u(0, t) v(0) = 4 4 = 0

w(0, t) = 0

w(2, t) = u(2, t) v(2) = 2 2 = 0

w(2, t) = 0

Equation (13) determines the initial condition for w.

w(x, 0) = u(x, 0) v(x) = 2x 2 (2 x) = 4

w(x, 0) = 4
(c) Let w(x, t) = X(x) T (t) and find the separated equations for X(x) and T (t). Solve the eigenvalue
problem for X(x), and then determine T (t).

Note that w satisfies equations (1)(3) with

L = 2, = 1, f (x) = 4.

The separated equations were already determined and solved above. By equation (4), the sepa-
rated equations are
T 0 + T = 0, X 00 + X = 0,
and the boundary conditions for X(x) are

X(0) = 0, X(L) = 0.

The eigenvalues and eigenfunctions are given by

2
n

n = 2
n = 1, 2, . . . ,
nx
 
Xn (x) = sin 2
and the corresponding solutions for T (t) were found to be
n2 2
Tn (t) = Bn e 4 t
, n = 1, , 2 . . . .

(d) Complete the solution by solving the intial-boundary value problem for w(x, t) and writing the
solution for u(x, t).

It follows from equations (7) and (8) that

n2 2
sin nx
 
t
X
w(x, t) = Bn e 4
2
n=1
and
L 2
2 2
Z Z
nx nx
   
Bn = f (x) sin L dx = 4 sin 2 dx
L 0 2 0

2 2 8 8
nx 

= 4 cos 2  = [cos(n) cos 0] = [1 (1)n ]
n 0 n n

8
Bn = [1 (1)n ] , n = 1, 2, . . . .
n
Therefore,
8X
1 (1)n n2 2 t  nx 
w(x, t) = e 4 sin 2 ,
n=1 n

and since u(x, t) = w(x, t) + v(x), the solution is

8X
1 (1)n n2 2 t  nx 
u(x, t) = 2 (2 x) e 4 sin 2 .
n=1 n
#5 Section 10.6: Suppose that a metal rod of unit length is completely insulated and that the temp-
erature u(x, t) of the rod satisfies

ut = 3 uxx, 0 < x < 1, t>0

ux (0, t) = ux (1, t) = 0, u(x, 0) = f (x)

where f (x) is the initial temperature of the rod.

(a) Let u(x, t) = X(x) T (t). Find the separated equations, solve the eigenvalue problem for X(x)
(dont forget to check = 0), and determine T (t).

The equations are separated as before.

T 0 + 3 T = 0 X 00 + X = 0

The boundary condition for X(x) are found as follows.

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

X 0 (0) = 0

ux (1, t) = X 0 (1) T (t) = 0

X 0 (1) = 0

The eigenvalue problem for X(x) was solved on homework 6 (problem #2(c) with L = 1). The
eigenvalues and eigenfunctions were found to be

n = (n)2
n = 0, 1, 2, . . . ,
Xn (x) = cos(nx)

The corresponding solutions for T (t) are found by solving

Tn0 + 3n Tn = 0,

which gives
2 2 t
Tn (t) = Cn e3n t = Cn e3n , n = 0, 1, 2, . . ..
(b) Find the general solution and show that u(x, t) approaches a constant as t . How is the
constant related to the initial temperature f (x)?

We have infinitely many solutions given by

2 2
un (x, t) = Xn (x) Tn (t) = e3 n t cos(nx), n = 0, 1, 2. . . . .

The initial conditions can be satisfied by considering the following linear combination of the un s.

A0 2 2
An e3n t cos(nx)
X
u(x, t) = +
2 n=1

Applying the initial condition yields

A0 X
u(x, t) = + An cos(nx) = f (x)
2 n=1

By comparing this equation to the Fourier Cosine series of f (x) on the interval (0, 1), it can be
seen that the An s are the Fourier Cosine series coefficient, and therefore
Z 1
An = 2 f (x) cos(nx) dx , n = 0, 1, 2, . . .
0

A0
As t , all terms of the series approach zero except for the constant term 2 , so that

A0
u(x, t) as t .
2

Recall from calculus that the integral

L
1
Z
A0
= f (x) dx
2 L 0

gives the average value of f (x) over the interval (0,L). It makes sense that the temperature in
the rod approaches this value as t . Since the rod is completely insulated on all sides, it
neither gains nor loses heat, so the total heat in the rod is always equal to the initial amount. As
t , the heat diffuses through the rod until it is uniformly distributed and thermal equilibrium
(constant temperature) is attained. Since the rod is insulated, this constant temperature must be
the average of the initial temperature distribution f (x).