Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

MATH 300 Fall 2004

Advanced Boundary Value Problems I


Solutions to Sample Final Exam
Friday December 3, 2004
Department of Mathematical and Statistical Sciences
University of Alberta

Question 1. Given the function


f (x) = cos a x,

0x<a

find the Fourier sine series for f.


Solution:
Writing f (x) = cos a x

n=1

bn sin n
a x, the coefficients bn in the Fourier sine series are computed as follows:

Z

1 a
cos
sin (n+1)
=
x + sin (n1)
x dx
a
a
a
0
0
a
a


1
1
1
1
(n+1) 
(n1) 

cos a x +

cos a x
=

n+1

n1
0
0

2
bn =
a

n
a x sin a x dx


 1 + (1)n
1
1
1 
1
(1)n + 1 +
(1)n + 1 =
+
.
(n + 1)
(n 1)

n+1 n1

Therefore,
bn =

4n
(n2 1)

If n = 1,
b1 =

2
a

if n is odd, n 3.

a
0

sin a x cos a x dx =

The Fourier sine series for f is therefore


cos a x
for 0 x < a.

if n is even

a

1
sin2 a x = 0.
a
0

n
8 P
sin 2n
a x.
n=1 4n2 1

Question 2. Let
f (x) =

cos x

|x| < ,

|x| > .

(a) Find the Fourier integral of f.


(b) For which values of x does the integral converge to f (x)?
(c) Evaluate the integral
Z

sin cos x
d
1 2

for < x < .


Solution:
(a) The function
f (x) =

cos x

|x| <

|x| >

is even, piecewise smooth, and is continuous at every x (, ) except at x = , therefore from


Dirichlets theorem the Fourier integral representation of f converges to f (x) for all x 6= , and
f (x)

A() cos x d,
0

where

1
=

1 sin( 1)
1 sin( + 1)
+

+1

2 sin
.
1 2

A() =

f (x) cos x dx =
0

cos x cos x dx
0

cos( + 1)x + cos( 1)x dx



sin( 1)x
sin( + 1)x
+
+ 1 0
1 0

The Fourier integral representation of f is therefore


2
f (x)

sin cos x
d.
1 2

(b) From Dirichlets theorem, the integral converges to f (x) for all x 6= , and converges to 12 for x = .
(c) Therefore, we have
Z

sin cos x
d =

1 2

cos x

for

|x| < ,

for

|x| > ,

for

x = .

Question 3. Let Fc denote the Fourier cosine transform and Fs denote the Fourier sine transform. Assume
that f (x) and xf (x) are both integrable.
(a) Show that
d
Fs (f (x)).
d

Fc (xf (x)) =
(b) Show that

Fs (xf (x)) =

d
Fc (f (x)).
d

Solution:
(a) From the definition of the Fourier sine transform, we have
d
d
Fs (f (x)) =
d
d

"r

f (t) sin t dt ,
0

and differentiating under the integral sign,


d
Fs (f (x)) =
d

f (t)
0

d
(sin t) dt
d

tf (t) cos t dt
0

= Fc (xf (x)),
and therefore

d
Fs (f (x)) = Fc (xf (x))
d

as required.
(b) From the definition of the Fourier cosine transform, we have
d
d
Fc (f (x)) =
d
d

"r

f (t) cos t dt ,
0

and differentiating under the integral sign,


r

d
Fc (f (x)) =
d

f (t)
0

d
(cos t) dt
d

tf (t) sin t dt
0

= Fs (xf (x)),
and therefore

as required.

d
Fc (f (x)) = Fs (xf (x))
d

Question 4. Chebyshevs differential equation reads


(1 x2 )y 00 xy 0 + y = 0,
y(1) = 1,

1 < x < 1

|y 0 (1)| <

(a) Divide by 1 x2 and bring the differential equation into Sturm-Liouville form. Decide if the resulting
Sturm-Liouville problem is regular or singular.
(b) For n 0, the Chebyshev polynomials are defined as follows:
Tn (x) = cos(n arc cos x),

1 x 1.

Show that Tn (x) is an eigenfunction of this Sturm-Liouville problem and for each n 0 find the
corresponding eigenvalue.
Hint: If v = arc cos x, then cos v = x, and v 0 =
(c) Show that
Z

1
1

1
1
=
.
sin v
(1 x2 )1/2

Tm (x)Tn (x)
dx = 0
(1 x2 )1/2

for m 6= n, so that these eigenfunctions are orthogonal on the interval [1, 1] with respect to the weight
1
.
function w(x) =
(1 x2 )1/2

Solution:

(a) We can rewrite the differential equation as




(1 x2 )1/2 y 0

0

y
= 0,
(1 x2 )1/2

which is the self-adjoint form of the Sturm-Liouville problem, with


p(x) = (1 x2 )1/2 ,

q(x) = 0,

r(x) =

1
.
(1 x2 )1/2

This is clearly a singular Sturm-Liouville problem since p(x) vanishes at the endpoints x = 1, and
since r(x) is not defined on the closed interval [1, 1] let alone continuous there. It also fails to be
regular because of the boundary conditions, one of which is a boundedness condition.
(b) If y = Tn (x), then
y = cos nk
where k = k(x) = arc cos x, so that x = cos k and using the chain rule, we have


1
n sin nk
,
y 0 = n sin nk k 0 = n sin nk
=
sin k
sin k
and
y 00 =

n2 cos nk + n sin nk cot k


n2 y
xy 0
=
+
,
1 x2
1 x2
sin2 k

and y = Tn (x) satisfies the differential equation (1 x2 )y 00 xy 0 + n2 y = 0, 1 < x < 1, for each
n 0. Therefore, Tn (x) is an eigenfunction of this Sturm-Liouville problem with eigenvalue n2 for
n = 0, 1, 2 . . . .

(c) In the integral


Z

1
1

Tm (x)Tn (x)
dx
(1 x2 )1/2

make the substitution x = cos t, so that


dx = sin t dt = (1 cos2 t)1/2 dt = (1 x2 )1/2 dt
that is,
dt =

1
dx.
(1 x2 )1/2

Therefore,
Z

1
1

Tm (x)Tn (x)
dx =
(1 x2 )1/2

cos mt cos nt dt = 0
0

if m 6= n, and the Chebyshev polynomials are orthogonal on the interval [1, 1] with respect to the
1
weight function w(x) =
.
(1 x2 )1/2
Question 5. Solve the following initial value problem for the damped wave equation
u
2u
2u
+
2
+
u
=
t2
t
x2
u(x, 0) =

1
,
1 + x2

u
(x, 0) = 1.
t
Hint: Do not use separation, instead consider w(x, t) = et u(x, t).
Solution: Note that u(x, t) = et w(x, t), so that
2u
2w
= et 2
2
x
x
and

and

u
w
= et w + et
t
t
2
2u
t
t w
t w
=
e
w

2e
+
e
.
t2
t
t2

Therefore,
2
u
2u
t w
+
2
+
u
=
e
,
t2
t
t2

while

2
2u
t w
=
e
x2
x2

and if u is a solution to the original partial differential equation, then w is a solution to the equation
et


2w 2w

= 0,
t2
x2

and since et 6= 0, then w satisfies the initial value problem


2w
2w
=
,
2
t
x2
w(x, 0) =

< x < ,

t > 0,

1
,
1 + x2

w
(x, 0) = 1.
t
From DAlemberts equation to the wave equation, we have (since c = 1)


Z
1
1 x+t
1
1
1 ds,
+
+
w(x, t) =
2 1 + (x + t)2
1 + (x t)2
2 xt
so that
u(x, t) =
for < x < ,

t 0.



1
et
1
+ tet ,
+
2 1 + (x + t)2
1 + (x t)2

You might also like