TMA4170 Fourier

Analysis
Spring 2017
Norwegian University of Science and
Technology Exercise set 2 solution
Department of Mathematical
Sciences

1 B&N: 1.1 First off, observe that f (x) = x 2 is even, so the Fourier series will be a cosine
series. We find the coefficients:
Zπ
1 π2
a0 = x 2 dx = ,
2π 3
−π
Zπ
1 4 4
an = x 2 cos(nx)dx = 2
cos(nπ) = 2 (−1)n ,
2π n n
−π

so

π2 X∞ 4
f (x) = + (−1)n cos(nx).
3 n=1 n 2

The sums for N = 1, 2, 5, 7 are shown in the figures below. Notice that on the interval
[−2π, 2π], the Fourier series fails, as is to be expected since it assumes the function is
2π-periodic.

10
N=1
N=2
N=5
8 N=7
f(x)

-2
-4 -3 -2 -1 0 1 2 3 4

Figure 1: Convergence on [−π, π]

January 25, 2017 Page 1 of 5

 Exercise set 2 solution

40
N=1
35 N=2
N=5
N=7
30 f(x)

25

20

15

10

-5
-8 -6 -4 -2 0 2 4 6 8

Figure 2: Convergence on [−2π, 2π]

P∞ i nx P∞ i mx
2 B&N: 1.18 With f (x) = n=−∞ a n e and g (x) = m=−∞ b m e , we can observe that
Zπ
1
( f ∗ g )(x) = f (t )g (x − t )dt
2π
−π
Zπ ∞ ∞
1
a n e i nt b m e i m(x−t ) dt
X X
=
2π n=−∞ m=−∞
−π
π
1 X ∞ ∞ Z
a n b m e i mx e i (n−m)t dt
X
=
2π n=−∞ m=−∞
−π
1 X ∞ ∞
a n b m e i mx 2πδnm
X
=
2π n=−∞ m=−∞
∞
a n b n e i nx ,
X
=
n=−∞

which is what we wanted to show.

3 B&N: 1.20 If f is continuous on 0 ≤ x ≤ a, then the even extension is, at least, continuous
on −a ≤ x < 0. We have to check for continuity at x = 0. to that end, we check that the
upper and lower limits coincide:

lim f e (x) = lim− f (−x) = lim+ f (x) = lim+ f e (x).

x→0− x→0 x→0 x→0

For the odd extension, continuity also holds on 0 ≤ x ≤ a, but not necessarily at x = 0 since

lim f o (x) = lim− − f (−x) = − lim− f (−x) = − lim+ f (x) = − lim+ f o (x).
x→0− x→0 x→0 x→0 x→0

The only way this holds is if f o (0) = 0, i.e. f (0) = 0.

January 25, 2017 Page 2 of 5

 Exercise set 2 solution

4 B&N: 1.22 From Theorem 1.30, we know that the Fourier series of a continuous, piecewise
smooth function converges uniformly. The function f (x) = x 2 is continuous and smooth
on [−π, π], and f (−π) = f (π), i.e. f is 2π-continuous. Therefore, the Fourier series from
exercise 1 converges uniformly. By evaluating f at x = π, we see that

π2 X∞ 4
f (π) = π2 = + (−1)n cos(nπ)
3 n=1 n 2
π2 X∞ 4
= +
3 n=1 n 2
π2 X∞ 1
⇒ = .
6 n=1 n 2

5 B&N: 1.25 Observe that since F is 2π-periodic and by a change of variables, we have

−π+c
Z −π+c
Z π+c
Z
F (x)dx = F (x + 2π)dx = F (y)dy.
−π −π π

From this, we can see that

−π+c
Z Z−π Zπ π+c
Z
F (x)dx = F (x)dx + F (x)dx + F (x)dx
−π+c −π+c −π π
−π+c
Z Zπ π+c
Z
=− F (x)dx + F (x)dx + F (x)dx
−π −π π
π+c
Z Zπ π+c
Z
=− F (x)dx + F (x)dx + F (x)dx
π −π π
Zπ
= F (x)dx.
−π

6 B&N: 1.40
a) This is pretty straightforward. If u(x, y) = (x + i y)n , then
( (
0, n<2 0, n<2
u xx (x, y) = u y y (x, y) =
n(n − 1)(x + i y)n−2 , n ≥ 2 −n(n − 1)(x + i y)n−2 , n ≥ 2

In any case, u xx + u y y = 0. The same goes for u(x, y) = (x − i y)n .

A n z n +A −n z n , and observe that by linearity

PN
b) Next, with z = x+i y, we define S N = n=0
of the Laplace operator,

N N
∆S N = ∆ A n z n + A −n z n = A n ∆z n + A −n ∆z n = 0,
X X
n=0 n=0

January 25, 2017 Page 3 of 5

 Exercise set 2 solution

since all z n and z n are solutions of the Laplace equation from part a). If we change
to polar coordinates and write z = r e i φ , we have
∞ ∞
u(r, φ) = A n z n + A −n z n = A n r n e i nφ + A −n r n e −i nφ
X X
n=0 n=0
∞ 0
A n r n e i nφ + A n r −n e i nφ
X X
=
n=0 n=−∞
∞
A n r |n| e i nφ ,
X
=
n=−∞

where we have taken a shortcut and written 2A 0 = A 0 .

c) With u in the above form, we see that the boundary condition reduces to
∞ ∞
u(1, φ) = A n 1|n| e i nφ = A n e i nφ = f (φ),
X X
n=−∞ n=−∞

from which we see that the A n are the Fourier coefficients of f .

d) The complex Fourier coefficients of f are given as

Zπ
1
An = f (φ)e −i nφ dφ.
2π
−π

From this, we can see that since f is real valued, i.e. f = f , and φ is real,

Zπ Zπ Zπ
1 i nφ 1 1
A −n = f (φ)e dφ = f (φ)e −i nφ dφ = f (φ)e −i nφ dφ = A n .
2π 2π 2π
−π −π −π

Also, in the case n = 0, we see that A 0 is real valued. Now, we can see that
∞
u(r, φ) = A n r |n| e i nφ
X
n=−∞
∞
A n r n e i nφ + A −n r n e −i nφ − A 0
X
=
n=0
∞
(A n e i nφ + A n e i nφ )r n − A 0
X
=
n=0
½ ∞ ¾
i nφ n
X
= Re 2A n e r − A0
n=0
Zπ Zπ
   
X∞ 1 1 
= Re 2 f (t )e −i nt dt  e i nφ r n − f (t )dt
n=0 2π 2π 
−π −π
 π 
µ ∞ ¶ 
1 Z X n i n(φ−t )
= Re f (t ) 2 r e − 1 dt .
2π  n=0

−π

e) We know that the formula is valid on the boundary r = 1; there, due to the A n
being the Fourier coefficients, u(1, φ) = f (φ). On the interior, i.e. r < 1, we have
|r n e i n(φ−t ) | = |r n | < 1, such that using the formula for a geometric sum is valid:
∞ ∞ 1
r n e i n(φ−t ) = (r e i (φ−t ) )n =
X X
.
n=0 n=0 1 − r e i (φ−t )

January 25, 2017 Page 4 of 5

 Exercise set 2 solution

Hence, we have
 π 
µ ¶ 
1 Z 2
u(r, φ) = Re f (t ) − 1 dt
2π  1 − r e i (φ−t ) 
−π
Zπ ½ ¾
1 2
= f (t )Re − 1 dt
2π 1 − r e i (φ−t )
−π
Zπ
1
= f (t )P (r, φ − t )dt ,
2π
−π

where
½ ¾
2
P (r, u) = Re −1 .
1 − r eiu
f ) We start the last point by observing that

2 (1 + r e i u )(1 − r e −i u )
−1 =
1 − r eiu (1 − r e i u )(1 − r e −i u )
1 + r e i u − r e −i u − r 2
=
1 − r e i u − r e −i u + r 2
1 − r 2 + 2i r sin(u)
= .
1 − 2r cos(u) + r 2

Taking the real part of this (note that the denominator is real), we have

1−r2
P (r, u) = .
1 − 2r cos(u) + r 2

Using this within the formula from e), we find that

Zπ Zπ
1 1 1−r2
u(r, φ) = f (t )P (r, φ − t )dt = f (t ) dt .
2π 2π 1 − 2r cos(φ − t ) + r 2
−π −π

January 25, 2017 Page 5 of 5