ANSWERS

PHYS2004: Wave Phenomena (PHYS2008: Fourier Methods)

Fourier Series Problem Sheet

n.b. The questions marked with a ? are more challenging. They are not conceptually difficult, but
the maths is quite long-winded, so you’ll need to do the calculations carefully to get the right an-
swer.

1. Are the following functions periodic? If so what is the period?

a) f (θ) = cos θ
cos θ = cos (θ + 2π) for all θ. Therefore this is periodic with period 2π.
b) f (t) = cos (ω0 t)
This is periodic with period T = 2π/ω0 (since cos (ω0 t) = cos (ω0 t + 2π)).
c) f (t) = exp (iω0 t)
This is also periodic with period T = 2π/ω0 (since exp (iω0 t) = cos (iω0 t) + i sin (iω0 t) and both
terms have period T = 2π/ω0 ).
d) f (x) = x2
This is not periodic as there is no L for which f (x) = f (x + L) for all x.
e) f (t) = sin (3ω0 t)
This is periodic with period T = 2π/(3ω0 ).
f) f (t) = cos (ω0 t) + 14 sin (3ω0 t)
The 1st term has period T1 = 2π/ω0 . The 2nd term has period T2 = 2π/(3ω0 ). T1 = 3T2 , therefore
the full function has period T = T1 = 2π/ω0 . i.e. the 2nd term goes through 3 cycles in the same
time, t = T1 = 2π/ω0 , that the 1st term goes through 1 cycle. See Fig. 1 below.
g) f (t) = cos (ω0 t) + sin (πω0 t)
1
The 1st term has period T1 = 2π/ω0 . The 2nd term has period T2 = 2/(ω0 ). T2 /T1 = π
is
irrational, therefore the full function is not periodic. See Fig. 1 below.

1
 Figure 1: q1: f (t) from part f) (red solid line) and part g) (dotted blue line) .

2. Write down (without doing any calculations) the coefficients of the trigonometric Fourier
series of the function in 1f)

The trigonometric Fourier series is

∞
a0 X
f (t) = + [ar cos (rω0 t) + br sin (rω0 t)] ,
2 r=1

therefore a1 = 1, b3 = 1/4 and all other coefficients are zero.

3. Write down (without doing any calculations) the coefficients of the complex Fourier series
of the function in 1f)

The complex Fourier series is

∞
X
f (t) = cr exp (iω0 rt) ,
r=−∞

Using
exp (iω0 t) + exp (−iω0 t)
cos (ω0 t) = ,
2
exp (iω0 t) − exp (−iω0 t) −i exp (iω0 t) + i exp (−iω0 t)
sin (ω0 t) = = ,
2i 2
we can rewrite f (t) as
1 1 i i
f (t) = exp (iω0 t) + exp (−iω0 t) − exp (i3ω0 t) + exp (−i3ω0 t) ,
2 2 8 8
and therefore c1 = c−1 = 1/2, c3 = −i/8, c−3 = i/8 and all other coefficients are zero.

2
4. Are the following functions even, odd or neither?

a) f (x) = sin (x2 )

f (−x) = sin ((−x)2 ) = sin (x2 ) = f (x), therefore this function is even.

b) f (x) = cot(x)
f (−x) = cot (−x) = cos (−x)/ sin (−x) = cos x/(− sin x) = − cot (x) = −f (x), therefore this
function is odd.

c) f (x) = x + x4
f (−x) = −x+(−x)4 = −x+x4 6= f (x) or f (−x), therefore this function is neither odd nor even.

d) f (x) = x2 + cos x
f (−x) = (−x)2 + cos (−x) = x2 + cos x, therefore this function is even.

5. A function is defined as f (x) = 1 − x in the range 0 < x < 1. Sketch the analytic continu-
ations of the function outside this range which are:
i) periodic with period L = 1,
ii) even with period L = 2,
iii) odd with period L = 2.

Figure 2: 5: analytic continuations of f (x) = 1 − x which are (from left to right) i) periodic with
period L = 1, ii) even with period L = 2, iii) odd with period L = 2.

3
6. Two voltage signals are observed to have the form:
1
V1 (t) = 3 cos (t/3) + cos t + cos (2t) ,
2
1
V2 (t) = 3 cos (t/π) + cos t + cos (2t) .
2
In each case:
a) Is the signal periodic?

Signal V1 (t) is periodic with period T = 6π (this is the smallest value of t such that t/3, t and 2t
are all integer multiples of 2π, so that V1 (t) = V1 (t + T ) for all t).
Signal V2 (t) is not periodic. There is no value of t such that t/π, t and 2t are all integer multiples
of 2π.
See Fig. 3 for plots of V1 (t) and V2 (t).

Figure 3: q6a: V1 (t) (red solid line) and V2 (t) (dotted blue line) .

If so:
b) What is the period?

For signal 1 the fundamental frequency is

2π 2π 1
ω0 = = = ,
T 6π 3
therefore it can be rewritten as
1
V1 (t) = 3 cos (ω0 t) + cos (3ω0 t) + cos (6ω0 t) ,
2
and hence br = 0 for all r, a1 = 3, a3 = 1, a6 = 1/2 and all other ar are zero.

4
 Figure 4: q6c: sketch of Fourier components of trigonometric Fourier series.

c) Sketch the coefficients of the trigonometric Fourier series.

7. Use trig identities to re-write each of the expressions below in the form of a linear com-
bination of exp (iω0 t), exp (−iω0 t) and their harmonics exp (irω0 t), exp (−irω0 t) (i.e. in the
form Σr cr exp (irω0 t)):
a) cos (ω0 t)

1
cos (ω0 t) = [exp (iω0 t) + exp (−iω0 t)] .
2

b) cos (2ω0 t)

1
cos (2ω0 t) = [exp (i2ω0 t) + exp (−i2ω0 t)] .
2

c) 2 cos (ω0 t) sin (ω0 t)

1
2 cos (ω0 t) sin (ω0 t) = sin (2ω0 t) = [exp (i2ω0 t) − exp (−i2ω0 t)] ,
2i
i
= [exp (−i2ω0 t) − exp (i2ω0 t)] .
2

d) cos2 (ω0 t)

1 1 1
cos2 (ω0 t) = [1 + cos (2ω0 t)] = + [exp (i2ω0 t) + exp (−2iω0 t)] .
2 2 4

5
8. A train of rectangular pulses of height h centered at t = 0 has a pulse width equal to 1/4 of
the pulse period T . Show that the 4th, 8th, 12th etc. harmonics are missing.
This can be written mathematically as
(
h if |t| < T /8 ,
f (t) =
0 otherwise .

This is an even function, therefore br = 0 and

4 T /2
Z  
2πrt
ar = f (t) cos dt ,
T 0 T
4h T /8
Z  
2πrt
= cos dt ,
T 0 T
  T /8
4h T 2πrt
= sin ,
T 2πr T 0
2h  πr 
= sin ,
πr 4
which is zero if r is a multiple of 4.

n.b. In general a0 has to be calculated separately, but this question doesn’t ask us to calculate the
full Fourier series (just to show that the harmonics which are multiples of 4 are missing) therefore
in this case we don’t need to calculate it.

9. The function f (x) is periodic with period L and given by f (x) = exp (αx) for 0 ≤ x ≤ L,
where α is a real positive constant.
a) Sketch this function.

Figure 5: 8a: sketch of f (x).

b) Calculate the complex Fourier coefficients cr .

6
 1 L
Z  
2πirx
cr = f (x) exp − dx ,
L 0 L
1 L 1 L
Z   Z   
2πirx 2πir
= exp (αx) exp − dx = exp α − x dx ,
L 0 L L 0 L
   L     
1 1 2πir 1 2πir
= exp α − x = exp α − L −1 ,
L α − 2πir
L
L 0 Lα − 2πir L
1 1
= [exp (αL) exp (−2πir) − 1] = [exp (αL) − 1] ,
(αL − 2πir) (αL − 2πir)

since exp (−2πir) = 1 as r is an integer.

c) Is this function odd, even or neither?

This function is nether odd nor even.

d) What does this tell us about the trigonometric Fourier series coefficients, ar and br ?
This means both ar and br are non-zero. The question didn’t ask you to calculate them, but if you
did you’d find:
 
1 1
ar = cr + c−r = (exp (αL) − 1) + ,
αL − 2πir αL − 2πir
   
αL + 2πir + αL − 2πir 2αL
= (exp (αL) − 1) = (exp (αL) − 1) 2 2 .
α2 L2 + 4π 2 r2 α L + 4π 2 r2
 
1 1
br = i(cr − c−r ) = i (exp (αL) − 1) − ,
αL − 2πir αL − 2πir
   
αL + 2πir − αL + 2πir 4πr
= i (exp (αL) − 1) = (1 − exp (αL)) 2 2 .
α2 L2 + 4π 2 r2 α L + 4π 2 r2

10? . Show that the Fourier series of the odd analytic continuation of the function
(
2h
x if 0 < x ≤ l/2 ,
f (x) = 2hl
l
(l − x) if l/2 < x ≤ l ,

is given by      
8h πx 1 3πx
f (x) = 2 sin − sin + ... .
π l 9 l
(we’ll use this result in Fourier 12).

7
The odd analytic continuation will have L = 2l, ar = 0 and

2 L/2 4 L/2 4 2l/2

Z   Z   Z  
2πrx 2πrx 2πrx
br = f (x) sin = f (x) sin = f (x) sin ,
L −L/2 L L 0 L 2l 0 L
" Z #
2 2h l/2 2h l
 πrx  Z  πrx 
= x sin dx + (l − x) sin dx ,
l l 0 l l l/2 l
"Z #
l/2 Z l Z l
4h  πrx   πrx   πrx 
= 2 x sin dx − x sin dx + l sin dx .
l 0 l l/2 l l/2 l

The last term gives us

Z l  πrx   2 h   2
l πrx il l
l sin dx = − cos = [cos (rπ/2) − cos (rπ)] ,
l/2 l πr l l/2 πr

and integrating by parts

Z  πrx    πrx  Z l
lx  πrx 
x sin dx = − cos + cos dx ,
l rπ l rπ l
" #
lx  πrx   l 2  πrx 
= − cos + sin ,
rπ l rπ l

so that
l/2
l2 l2
Z  πrx 
x sin dx = − cos (πr/2) + sin (πr/2) ,
0 l 2rπ rπ
and
l
l2 l2 l2
Z  πrx 
x sin dx = − cos (πr) + cos (πr/2) − sin (πr/2) .
l/2 l rπ 2rπ rπ
Putting these results together we get
l2 l2 l2 l2 l2

4h
bn = 2 − cos (πr/2) + sin (πr/2) + cos (πr) − cos (πr/2) + sin (πr/2)
l 2rπ rπ rπ 2rπ rπ
 2  2 
l l
+ cos (rπ/2) − cos (rπ) ,
rπ rπ
8h
= sin (πr/2) ,
π 2 r2
and hence    
8h 8h 1
b1 = , b2 = 0 , b3 = − , b4 = 0 , ... .
π2 π2 9

11? . A sawtooth function with period T is defined as

f (t) = |t| for − T /2 ≤ t ≤ T /2 .

a) Calculate the coefficients cr of its exponential Fourier series. Hint: you’ll need to consider
the positive and negative t integrals separately.

8
 Z T /2   Z T /2
1 2πirt
cr = f (t) exp − dt = f (t) exp (−irω0 t) dt ,
T −T /2 T −T /2

with ω0 = 2π/T , and we have f (t) = |t|. Note that since exp (−irω0 t) is neither even nor odd
the integrand isn’t even and (unlike with the trigonometric series) we have to integrate over the full
range.
1 0 1 T /2
Z Z
cr = (−t) exp (−irω0 t) dt + t exp (−irω0 t) dt ,
T −T /2 T 0
" Z #
0 Z T /2
1
= − t exp (−irω0 t) dt + t exp (−irω0 t) dt .
T −T /2 0

Integrating by parts we get

Z  Z 
t 1
t exp (−irω0 t) dt = − exp (−irω0 t) + exp (−irω0 t) dt ,
irω0 irω0
 
t 1
= − exp (−irω0 t) + exp (−irω0 t) ,
irω0 (rω0 )2
 
1 t
= − exp (−irω0 t) .
(rω0 )2 irω0
Therefore
     
1 1 1 T irω0 T
cr = − −0 − + exp ,
T (rω0 )2 (rω0 )2 2irω0 2
     
1 1 T irω0 T 1
+ − exp − − −0 .
T (rω0 )2 2irω0 2 (rω0 )2
Remembering that ω0 T = 2π and so irω0 T /2 = irπ and using
exp (±irπ) = cos (rπ) = (−1)r ,
we get
   
1 1 1 T r
cr = − − + (−1) ,
T (rω0 )2 (rω0 )2 2irω0
  
1 1 T r 1
+ − (−1) − ,
T (rω0 )2 2irω0 (rω0 )2
 
1 2 2 r
= − + (−1) ,
T (rω0 )2 (rω0 )2
2 1
= 2
[−1 + (−1)r ] = 2
[−1 + (−1)r ] , ,
T (rω ) πω0 r
( 0
2
− πω0 r2 ir r is odd ,
=
0 otherwise .
This expression breaks down for r = 0, therefore we need to calculate this co-efficient separately:
 T /2
1 T /2 2 T /2 2 t2
Z Z
T
c0 = f (t) dt = t dt = = .
T −T /2 T 0 T 2 0 4

9
b) The trigonometric Fourier series of f (t) from part a) has coefficients
T
a0 = .
2
(
− πω40 r2 ir r is odd ,
ar =
0 otherwise .
br = 0 .

Verify that the general relationships

a0 = 2c0 ,
ar = cr + c−r ,
br = i(cr − c−r ) ,

are valid in this case.

Inserting these values into the general relationships between the coefficients of the trignometric
and complex forms of the Fourier series we get
T
a0 = ,
2  
2 1 r 1 r
ar = cr + c−r = (−1 + (−1) ) + (−1 + (−1) ) ,
T ω02 r2 (−r)2
4 2
= 2 2
[−1 + (−1)r ] = 2
[−1 + (−1)r ] ,
Tω r πω0 r
( 0
4
− πω0 r2 ir r is odd ,
=
0 otherwise ,
 
2 1 r 1 r
br = i(cr − c−r ) = (−1 + (−1) ) − (−1 + (−1) ) ,
T ω02 r2 (−r)2
= 0.

as required.
c) Calculate the average over a period of |f (t)|2 directly (by integrating). Then use Parseval’s
theorem to show that ∞
X 1 π4
= .
j=1
(2j − 1)4 96

Calculating the average of |f (t)|2 directly we get

T /2 T /2  3 T /2  3 
T2
Z Z
1 2 2 2 2 t 2 T
|t| dt = t dt = = = .
T −T /2 T 0 T 3 0 T 3×8 12

Parseval’s theorem tells us that

Z T /2 ∞
1 X
|f (t)|2 dt = |cr |2 ,
T −T /2 r=−∞

10
and we have
∞ −1 ∞  2  2 X −1  2 X∞
X
2 2
X
2
X
2 T 2 1 2 1
|cr | = |c0 | + |cr | + |cr | = + 4
+ 4
,
r=−∞ r=−∞ r=1
4 πω 0 j=−∞
(2j − 1) πω 0 j=1
(2j − 1)
 2  2 X∞  2  2 X ∞
T 2 1 T T 1
= +2 4
= + 2 2 4
.
4 πω0 j=1
(2j − 1) 4 π j=1
(2j − 1)

Therefore ∞
T2 T2 T2
 X
1
= +2 ,
12 16 π4 j=1
(2j − 1)4

and hence ∞
π4
 
X 1 1 1 1
4
= − π4 = .
j=1
(2j − 1) 2 12 16 96

11