Lecture 4: Fourier Series

page 1/44
Outline

1. Fourier series for 2π-periodic functions

2. Fourier series of odd and even functions

3. Convergence of Fourier series

4. Fourier series of T -periodic functions

page 1/44
1. Fourier series for 2π-periodic functions

Definition (Periodic functions)

If a function satisfies f (t + T ) = f (t), then it is called a T -periodic
function.

J. Fourier (a French mathematician and a physicist) found that

any 2π-periodic function can be viewed as a composition of
sine and cosine functions:
∞
a0 X

f (t) = + an cos(nt) + bn sin(nt) ,
2
n=1

called Fourier series.

The numbers an and bn are called Fourier coefficients of f .

page 2/44
1. Fourier series for 2π-periodic functions
Orthogonality of Sine and Cosine functions:

Z π Z π
cos(nt)dt = sin(nt)dt = 0, ∀ n = 1, 2, . . .
−π −π

Z π
cos(nt) sin(mt)dt = 0, ∀ n, m = 0, 1, 2, . . .
−π

(
Z π 0 if n 6= m,
cos(nt) cos(mt)dt =
−π π if n = m ≥ 1,

(
Z π 0 if n 6= m,
sin(nt) sin(mt)dt =
−π π if n = m ≥ 1,

page 3/44
1. Fourier series for 2π-periodic functions
Trigonometric identities:

2 cos θ cos ϕ = cos(θ − ϕ) + cos(θ + ϕ)

2 sin θ sin ϕ = cos(θ − ϕ) − cos(θ + ϕ)
2 sin θ cos ϕ = sin(θ + ϕ) + sin(θ − ϕ)
2 cos θ sin ϕ = sin(θ + ϕ) − sin(θ − ϕ)
   
θ±ϕ θ∓ϕ
sin θ ± sin ϕ = 2 sin cos
2 2
   
θ+ϕ θ−ϕ
cos θ + cos ϕ = 2 cos cos
2 2
   
θ+ϕ θ−ϕ
cos θ − cos ϕ = −2 sin sin
2 2

page 4/44
1. Fourier series for 2π-periodic functions
Proof of orthogonality: If n 6= m then
Z π
cos(nt) sin(mt)dt
−π
Z π  
1
= sin((n + m)t) − sin((n − m)t) dt
−π 2
  π
1 1 1
= − cos((n + m)t) + cos((n − m)t)
2 n+m n−m −π
1  
=− cos((n + m)π) − cos(−(n + m)π)
2(n + m)
1  
+ cos((n − m)π) − cos(−(n − m)π)
2(n − m)
=0

If n = m then blue part disappear.

page 5/44
1. Fourier series for 2π-periodic functions
Proof of orthogonality: If n 6= m then
Z π
cos(nt) cos(mt)dt
−π
Z π  
1
= cos((n + m)t) + cos((n − m)t) dt
−π 2
  π
1 1 1
= sin((n + m)t) + sin((n − m)t)
2 n+m n−m −π
1  
= sin((n + m)π) − sin(−(n + m)π)
2(n + m)
1  
+ sin((n − m)π) − sin(−(n − m)π)
2(n − m)
=0
Rπ Rπ 1
 
If n = m then −π cos(nt) cos(nt)dt = −π 2 cos(2nt) + 1 dt = π
page 6/44
1. Fourier series for 2π-periodic functions

Exercise: Verify the following identity:

(
Z π 0 if n 6= m,
sin(nt) sin(mt)dt =
−π π if n = m ≥ 1,

page 7/44
1. Fourier series for 2π-periodic functions
∞
a0 X

If f (t) = + an cos(nt) + bn sin(nt) ,
2
n=1
then integrating the equation yields
Z π Z π
a0
f (t)dt = dt
−π −π 2
X∞ Z π
+ an cos(nt)dt
n=1 −π
X∞ Z π
+ bn sin(nt)dt
n=1 −π

= a0 π.
This gives us a formula:
Z π
1
a0 = f (t)dt.
π −π

page 8/44
1. Fourier series for 2π-periodic functions
∞
a0 X

If f (t) = + an cos(nt) + bn sin(nt) ,
2
n=1
then integrating the equation against cos(mt) yields
Z π Z π
a0
f (t) cos(mt)dt = cos(mt)dt
−π −π 2
X∞ Z π
+ an cos(nt) cos(mt)dt
n=1 −π
X∞ Z π
+ bn sin(nt) cos(mt)dt
n=1 −π

= am π, m ≥ 0.
This gives us a formula:
Z π
1
am = f (t) cos(mt)dt.
π −π
page 9/44
1. Fourier series for 2π-periodic functions
∞
a0 X

Similarly, if f (t) = + an cos(nt) + bn sin(nt) ,
2
n=1
then integrating the equation against sin(mt) yields
Z π Z π
a0
f (t) cos(mt)dt = sin(mt)dt
−π −π 2
X∞ Z π
+ an cos(nt) sin(mt)dt
n=1 −π
X∞ Z π
+ bn sin(nt) sin(mt)dt
n=1 −π

= bm π, m ≥ 1.
This gives us a formula:
Z π
1
bm = f (t) sin(mt)dt.
π −π
page 10/44
1. Fourier series for 2π-periodic functions

To summarize, if
∞
a0 X

f (t) = + an cos(nt) + bn sin(nt)
2
n=1

then
Z π
1
a0 = f (t)dt
π −π
Z π
1
an = f (t) cos(nt)dt n ≥ 1,
π −π
Z π
1
bn = f (t) sin(nt)dt n ≥ 1.
π −π

page 11/44
1. Fourier series for 2π-periodic functions

Example
(
− 1 if t ∈ [−π, 0),
Find the Fourier series of the function f (t) =
1 if t ∈ [0, π),

Solution:
Z π Z 0 Z π
1 1 1
an = f (t) cos(nt)dt = (−1) · cos(nt)dt + 1 · cos(nt)dt = 0
π −π π −π π 0

1 π 1 0 1 π
Z Z Z
bn = f (t) sin(nt)dt = (−1) · sin(nt)dt + 1 · sin(nt)dt
π −π π −π π 0
2(1 − (−1)n )
=
nπ
Therefore,
∞ ∞
a0 X
X 2(1 − (−1)n )
f (t) = + an cos(nt) + bn sin(nt) = sin(nt).
2 n=1 n=1
nπ
page 12/44
1. Fourier series for 2π-periodic functions

Example
(
1 if t ∈ [−π, 0),
Find the Fourier series of the function f (t) =
3 if t ∈ [0, π),

Solution: f (t) = 2 + g(t) with

(
−1 if t ∈ [−π, 0),
g(t) =
1 if t ∈ [0, π),

Since
∞
X 2(1 − (−1)n )
g(t) = sin(nt)
n=1
nπ

it follows that
∞
X 2(1 − (−1)n )
f (t) = 2 + sin(nt)
n=1
nπ

page 13/44
1. Fourier series for 2π-periodic functions
Example
Find the Fourier series of the function f (t) = |t| for t ∈ [−π, π]
Solution:
1 π 1 0 1 π
Z Z Z
a0 = f (t)dt = (−t)dt + tdt = π,
π −π π −π π 0

1 π 1 0 1 π
Z Z Z
an = |t| cos(nt)dt = (−t) · cos(nt)dt + t · cos(nt)dt
π −π π −π π 0
Z π
2(1 − (−1)n )
 
=− computation of t · cos(nt)dt is in next page
n2 π 0
Z π Z 0
1 π
Z
1 1
bn = f (t) sin(nt)dt = (−t) · sin(nt)dt + t · sin(nt)dt = 0
π −π π −π π 0
Therefore,
∞ ∞
π X 2(1 − (−1)n ) π X 4
|t| = − 2
cos(nt) = − cos((2n − 1)t)
2 n=1 n π 2 n=1 (2n − 1)2 π
page 14/44
1. Fourier series for 2π-periodic functions

Z π Z π
t
t cos(nt)dt = d sin(nt)
0 0 n
π Z π
t sin(nt)
= sin(nt) − dt
n 0 0 n
cos(nt) π
=0+
n2 0
cos(nπ) − cos(0)
=
n2
n
(−1) − 1
=
n2

page 15/44
1. Fourier series for 2π-periodic functions

Example
Find the Fourier series of the function f (t) = t2 for t ∈ [−π, π]

Solution:
π
2π 2
Z
1
a0 = t2 dt = ,
π −π 3
π
(−1)n 4
Z
1
an = t2 cos(nt)dt = (shown on next page)
π −π n2
Z π
1
bn = t2 sin(nt)dt = 0
π −π

Therefore,
∞ ∞
a0 X
π2 X (−1)n 4
t2 = + an cos(nt) + bn sin(nt) = + cos(nt)
2 n=1
3 n=1
n2

page 16/44
1. Fourier series for 2π-periodic functions

π π π Z π
t2 t2
Z Z
sin(nt) 2
t2 cos(nt)dt = d sin(nt) = sin(nt) − dt
−π −π n n 0 −π n
Z π
2t sin(nt)
=− dt
−π n
Z π
2t
= 2
d cos(nt)
−π n
π Z π
2t cos(nt) 2 cos(nt)
= − dt
n2 −π −π n2
π π
2t cos(nt) 2 sin(nt)
= −
n2 −π n3 −π
2π cos(nπ) − (−π) cos(−nπ))
=
n2
n
4π(−1)
=
n2

page 17/44
2. Fourier series for odd and even functions
If f (t) is an odd function, i.e., f (−t) = −f (t), then

f (t) cos(nt) is an odd function

f (t) sin(nt) is an even function

Consequently,
1 π
Z
an = f (t) cos(nt)dt = 0, n = 0, 1, 2, . . . ,
π −π
1 π 2 π
Z Z
bn = f (t) sin(nt)dt = f (t) sin(nt)dt, n = 1, 2, . . .
π −π π 0

Conclusion: odd function f (t) is a combination of sine functions:

∞
X
f (t) = bn sin(nt).
n=1

page 18/44
2. Fourier series for odd and even functions
If f (t) is an even function, i.e., f (−t) = f (t), then

f (t) cos(nt) is an even function

f (t) sin(nt) is an odd function

Consequently,
1 π 2 π
Z Z
an = f (t) cos(nt)dt = f (t) cos(nt)dt, n = 0, 1, 2, . . . ,
π −π π 0
1 π
Z
bn = f (t) sin(nt)dt = 0, n = 1, 2, . . .
π −π

Conclusion: even function f (t) is a combination of cosine functions:

∞
a0 X
f (t) = + an cos(nt).
2
n=1

page 19/44
2. Fourier series for 2π-periodic functions

Example
Find the Fourier series of the function f (t) = |t| for t ∈ [−π, π]

Solution: |t| is an even function, because | − t| = |t|. Therefore,

1 π 2 π
Z Z
a0 = |t|dt = tdt = π,
π −π π 0
1 π 2 π 2(1 − (−1)n )
Z Z
an = |t| cos(nt)dt = t cos(nt)dt =
π −π π 0 n2 π
1 π
Z
bn = |t| sin(nt)dt = 0
π −π

Therefore,
∞
π X 2(1 − (−1)n )
|t| = − cos(nt)
2 n=1 n2 π

page 20/44
2. Fourier series for 2π-periodic functions

Definition (Odd extension from [0, π] to [−π, π])

If f (t) is initially defined for t ∈ [0, π] satisfying f (0) = f (π) = 0, then it
can be extended to t ∈ [−π, π] to be an odd function, i.e.,

f (t) = −f (−t) for t ∈ [−π, 0].

After extending a function f (t) to be an odd function on [−π, π], the

function can be represented by a series of sine functions.
Consequently, any function f (t) defined for t ∈ [0, π] can be represented by
a series of sine functions.

page 21/44
2. Fourier series for 2π-periodic functions

Definition (Even extension from [0, π] to [−π, π])

If f (t) is initially defined for t ∈ [0, π], then it can be extended to
t ∈ [−π, π] to be an even function, i.e.,

f (t) = f (−t) for t ∈ [−π, 0].

After extending a function f (t) to be an even function on [−π, π], the

function can be represented by a series of cosine functions.
Consequently, any function f (t) defined for t ∈ [0, π] can be represented by
a series of cosine functions.

page 22/44
2. Fourier series for 2π-periodic functions
Example
Express the function f (t) = 1 as sine series on [0, π].

Solution: The function f (t) = 1 does not satisfy f (0) = f (π) = 0. We need to
first modify the definition at t = 0, π and consider the modified function
(
0 for t = 0, π
fe(t) =
1 for t ∈ (0, π)
This function has odd extension to [−π, π]. Therefore, fe(t) has Fourier sine series
with
2 π e 2 π 2(1 − (−1)n )
Z Z
bn = f (t) sin(nt)dt = sin(nt)dt =
π 0 π 0 nπ
∞
X 2(1 − (−1)n )
Therefore, fe(t) = sin(nt) for t ∈ [0, π].
n=1
nπ
∞
X 2(1 − (−1)n )
1= sin(nt) for t ∈ (0, π).
n=1
nπ
page 23/44
2. Fourier series for 2π-periodic functions

Example
Express the function f (t) = t as sine series on [0, π].

Solution: The function f (t) = t satisfies f (0) = 0 but f (π) 6= 0.

We first modify the definition at t = π and consider the modified function
(
0 for t = π
fe(t) =
t for t ∈ [0, π)
This function has odd extension to [−π, π], with Fourier coefficient
2 π e 2 π 2(−1)n−1
Z Z
bn = f (t) sin(nt)dt = t sin(nt)dt =
π 0 π 0 n
Therefore,
∞
X 2(−1)n−1
t= sin(nt) for t ∈ [0, π).
n
n=1
page 24/44
2. Fourier series for 2π-periodic functions

Example
 π
0 if t ∈ [0, ),
Express the function f (t) = 2 as sine series on [0, π].
1 if t ∈ [ π , π],
2
Solution: The function satisfies f (0) = 0 but f (π) 6= 0. We first modify
the function by redefining f (π) = 0.
Then the function has odd extension to [−π, π]. Therefore, f (t) has
Fourier sine series with
2 π 2 π 2(cos( πn n
2 ) − (−1) )
Z Z
bn = f (t) sin(nt)dt = sin(nt)dt =
π 0 π π nπ
2

 π
∞
2(cos( πn ) − (−1)n ) 0 if t ∈ [0, ),
2 .
X
2
Therefore, sin(nt) =
nπ 1 if t ∈ ( π , π),
n=1
2
page 25/44
2. Fourier series for 2π-periodic functions

Example
Express the function f (t) = 1 as cosine series on [0, π].
∞
a0 X
Solution: We want to have 1 = + an cos(nt)
2
n=1

In fact, a0 = 2 and an = 0 for n ≥ 1 satisfy the equality:

1 = 1.

This is the cosine series of 1.

page 26/44
2. Fourier series for 2π-periodic functions

Example
Express the function f (t) = t as cosine series on [0, π].

Solution:
Z π
2
a0 = tdt = π,
π 0
π
2(1 − (−1)n )
Z
2
an = t cos(nt)dt = −
π 0 n2 π

Therefore,
∞
π X 2(1 − (−1)n )
t= − cos(nt)
2 n2 π
n=1

page 27/44
2. Fourier series for 2π-periodic functions
Example
(
0 if t ∈ [0, π/2),
Express the function f (t) = as cosine series on [0, π].
1 if t ∈ [π/2, π],

Solution:
Z π Z π
2 2
a0 = f (t)dt = 1dt = 1,
π 0 π π
2
Z π Z π
2 2 2  nπ 
an = f (t) cos(nt)dt = cos(nt)dt = − sin
π 0 π π
2
nπ 2
Therefore,
∞
1 X 2  nπ 
f (t) = − sin cos(nt)
2 n=1 nπ 2
∞
1 X 2(−1)n
= + cos((2n − 1)t)
2 n=1 (2n − 1)π

page 28/44
2. Fourier series for 2π-periodic functions
Example
(
0 if t ∈ [0, π/2),
Express the function f (t) = as cosine series on [0, π].
t if t ∈ [π/2, π],

Solution:
Z π Z π
2 2 3π
a0 = f (t)dt = tdt = ,
π 0 π π
2
4
Z π Z π
2 2
an = f (t) cos(nt)dt = t cos(nt)dt
π 0 π π
2
   
nπ nπ
−nπ sin 2 + 2(−1)n − 2 cos 2
=
n2 π
Therefore,
   
nπ nπ
3 X −nπ sin
∞
2 + 2(−1)n − 2 cos 2
f (t) = + cos(nt)
8 n=1 n2 π
page 29/44
2. Fourier series for 2π-periodic functions

Example
(
0 if t ∈ [0, π/2),
Express the function f (t) = as cosine series on [0, π].
t−1 if t ∈ [π/2, π],

Solution: In the previous two examples, we have considered the two functions
( (
0 if t ∈ [0, π/2), 0 if t ∈ [0, π/2),
g(t) = and h(t) =
1 if t ∈ [π/2, π], t if t ∈ [π/2, π],
∞
with 1 X 2 nπ

g(t) = − sin cos(nt)
2 n=1 nπ 2
 
∞ −nπ sin nπ + 2(−1)n − 2 cos nπ


3 X 2 2
h(t) = + cos(nt).
8 n=1 n2 π
∞
1 X n(2 − π) sin nπ + 2(−1)n − 2 cos nπ



2 2
f (t) = h(t) − g(t) = − + cos(nt)
8 n=1 n2 π

page 30/44
3. Convergence of Fourier series

∞
a0 X

If f (t) has Fourier series + an cos(nt) + bn sin(nt) ,
2 n=1
then


f (t) if f is continuous at t,

∞

 f (t− ) + f (t+ )

a0 X if f is discontinuous at t,
+ an cos(nt) + bn sin(nt) = 2
2 n=1

f (−π) + f (π)


if t = ±π,


2
where f (t− ) and f (t+ ) denote the left-side and right-side value of f at t.

page 31/44
3. Convergence of Fourier series
Example
∞
π X 4
|t| = − cos((2n − 1)t).
2 n=1 (2n − 1)2 π

f (t) = |t| is continuous at every point of t ∈ [−π, π], and f (π) = f (−π).
∞
Therefore, π X 4
|t| = − cos((2n − 1)t) ∀ t ∈ [−π, π]
2 n=1 (2n − 1)2 π

Substituting t = 0 into the last expression yields

∞
π X 4
0= − .
2 n=1 (2n − 1)2 π

This gives us an identity:

1 1 1 1 π2
1+ 2
+ 2 + 2 + 2 + ··· =
3 5 7 9 8
page 32/44
3. Convergence of Fourier series
Example
∞
π 2 X (−1)n 4
t2 = + cos(nt)
3 n=1
n2

f (t) = |t| is continuous at every point of t ∈ [−π, π], and f (π) = f (−π).
Therefore, the series is convergent to t2 for all t ∈ [−π, π].
∞
π 2 X (−1)n−1 4
Substituting t = 0 into the series yields 0 = − ,
3 n=1
n2
which implies ∞
X (−1)n−1 π2
2
=
n=1
n 12
This gives us an identity:
1 1 1 1 1 π2
1− 2
+ 2 − 2 + 2 − 2 + ··· =
2 3 4 5 6 12
page 33/44
3. Convergence of Fourier series

Example
In the previous two examples, we obtained two identities:

1 1 1 1 π2
1+ + + + + · · · =
32 52 72 92 8
1 1 1 1 1 1 π2
1− 2
+ 2 − 2 + 2 − 2 + 2 − ··· =
2 3 4 5 6 7 12
Subtracting the second identity from the first identity, we obtain

1 1 1 1 π2 π2 π2
+ + + + · · · = − =
22 42 62 82 8 12 24
1 1 1 1 1 π2
=⇒ 2
+ 2 2
+ 2 2
+ 2 2
+ 2 2
+ ··· =
2 2 ×2 2 ×3 2 ×4 2 ×5 24
1 1 1 1 π2
This implies 1 + 2
+ 2 + 2 + 2 + ··· =
2 3 4 5 6
page 34/44
4. Fourier series for T -periodic functions
If a function f (t) is T -periodic, i.e. f (t + T ) = f (t), then through the change of
variable
T
t= s,
2π
T  T  T 
the function f s becomes 2π-periodic in s, i.e., f (s + 2π) = f s
2π 2π 2π
T 
Therefore, f s has Fourier series expansion in s:
2π
  ∞
T a0 X

f s = + an cos(ns) + bn sin(ns)
2π 2 n=1

Correspondingly,
  ∞
T a0 X

f (t) = f s = + an cos(ns) + bn sin(ns)
2π 2 n=1
∞   2π 
a0 X  2π 
= + an cos nt + bn sin nt
2 n=1
T T

page 35/44
4. Fourier series for T -periodic functions

π
1 2π
Z 
 Z  
1 T T
an = f s cos(ns)ds = f s cos(ns)ds
π −π 2π π 0 2π
2 T
Z  2π 
= f (t) cos nt dt
T 0 T

T
here we have used t= s
2π
π
1 2π
Z   Z  
1 T T
bn = f s sin(ns)ds = f s sin(ns)ds
π −π 2π π 0 2π
Z T
2  2π 
= f (t) sin nt dt
T 0 T

page 36/44
4. Fourier series for T -periodic functions

Example
Let f (t) = e−t for t ∈ [0, 1] and periodically extended to all t ∈ R. Find the
Fourier series expansion of f (t):

Solution:
Z 1
a0 = 2 e−t dt = 2(1 − 1/e),
0
1
2(1 − 1/e)
Z
e−t cos 2πnt dt =


an = 2 ,
0 4n2 π 2 + 1
1
4nπ(1 − 1/e)
Z
e−t sin 2πnt dt =


bn = 2 .
0 4n2 π 2 + 1
Consequently, we have
∞  
X 2(1 − 1/e)
4nπ(1 − 1/e)

f (t) = 1 − 1/e + cos 2πnt + sin 2πnt .
n=1
4n2 π 2 + 1 4n2 π 2 + 1

page 37/44
Outcome of this lecture

1 Calculate the Fourier series expansion of a 2π-periodic function

2 Calculate the Fourier sine series of a 2π-periodic odd function

3 Calculate the Fourier cosine series of a 2π-periodic even function

4 Calculate the Fourier series expansion of a T -periodic function

5 Calculate the value of some special series by using Fourier series

page 38/44
Exercises

Calculate the Fourier series of the following 2π-periodic functions on

[−π, π].

1 f (t) = t − 1.
∞
X (−1)n−1 2
Solution: −1 + sin(nt)
n
n=1

(
0 if t ∈ [−π, 0),
2 f (t) =
t if t ∈ [0, π),

∞
(−1)n − 1 (−1)n−1
 
π X
Solution: + cos(nt) + sin(nt)
4 n2 π n
n=1

page 39/44
Exercises
Calculate the Fourier series of the following 2π-periodic functions on
[−π, π].
(
t/2 if t ∈ [−π, 0),
3 f (t) =
t if t ∈ [0, π),
∞
(−1)n − 1 (−1)n−1 3
 
π X
Solution: + cos(nt) + sin(nt)
8 2n2 π 2n
n=1

(
0 if t ∈ [−π, 0),
4 f (t) = t
e if t ∈ [0, π),

Solution:
∞ 
eπ − 1 X (−1)n eπ − 1 n − (−1)n neπ

+ cos(nt) + sin(nt)
2π (n2 + 1)π (n2 + 1)π
n=1
page 40/44
Exercises

Calculate the Fourier cosine series expansion of the following functions on

[0, π]

5 f (t) = t − 1.
∞
π − 2 X 2((−1)n − 1)
Solution: + cos((2n − 1)t)
2 n2 π
n=1

6 f (t) = t2 .
∞
π 2 X 4(−1)n
Solution: + cos(nt)
3 n2
n=1

page 41/44
Exercises

Calculate the Fourier cosine series expansion of the following functions on

[0, π]

7 f (t) = sin(t).
∞
2 X 4
Solution: − 2
cos(2nt).
π (4n − 1)π
n=1

8 f (t) = et .
∞
eπ − 1 X 2(eπ (−1)n − 1)
Solution: + cos(nt).
π (n2 + 1)π
n=1

page 42/44
Exercises

By using the Fourier cosine series of the function f (t) = sin(t), calculate
the value of the following series:
∞
X (−1)n−1
9
n2 − 1/4
n=1

∞
X (−1)n−1
Solution: =π−2
n2 − 1/4
n=1

page 43/44
Exercises

By using the Fourier cosine series of the function f (t) = et , calculate the
value of the following series:
∞
X 1 1 1 1 1
10
2
= + + + + ···
(2n − 1) + 1 2 10 26 50
n=1

∞
X 1 eπ − 1 − π
Solution: =
(2n − 1)2 + 1 2(eπ + 1)
n=1

page 44/44