Section 3.

Fourier series and Periodic Functions

A function f (x) is said to be periodic if
f (x) = f (x + p) for any x, and xed period p.
In this section we shall consider how such functions can be approximated
using trigonometric functions. In order to simplify the analysis we shall, for
the moment, restrict our attention to functions f () which are 2 periodic,
such that f (x) = f (x + 2). With this in mind we shall attempt to
approximate f (x) by the 2 periodic trigonometric sum
f (x) f
N
(x) =
a
0
2
+
N

n=1
a
n
cos(nx) + b
n
sin(nx). (1)
(School of Mathematical Sciences) HG2M13 September 29, 2008 1 / 31
Before proceeding to calculate the coecients of this polynomial it is
helpful to evaluate the following integrals:
I
mn
=
_

cos(mx) cos(nx)dx, J
mn
=
_

sin(mx) sin(nx)dx,
K
mn
=
_

sin(mx) cos(nx)dx,
where m and n are positive integers. Using standard trigonometric
identities we can write
I
mn
=
_

1
2
[cos((m n)x) + cos((m + n)x)] dx,
J
mn
=
_

1
2
[cos((m n)x) cos((m + n)x)] dx,
K
mn
=
_

1
2
[sin((m n)x) + sin((m + n)x)] dx.
(School of Mathematical Sciences) HG2M13 September 29, 2008 2 / 31
It follows (for m 1 and n 0) that
I
mn
= 0 for m = n, I
mm
=
_

1
2
dx = ,
J
mn
= 0 for m = n, I
mm
=
_

1
2
dx = , K
mn
= 0.
(School of Mathematical Sciences) HG2M13 September 29, 2008 3 / 31
To summarise
_

cos(mx) cos(nx)dx =
mn
,
_

sin(mx) sin(nx)dx =
mn
,
_

sin(mx) cos(nx)dx = 0, (2)

_

1 cos(mx)dx = 2
0m
_

1 sin(mx)dx = 0.
where the Kronecker delta function
mn
is equal to 1 if m = n and is equal
to zero otherwise.
(School of Mathematical Sciences) HG2M13 September 29, 2008 4 / 31
Fourier Series
By taking N , we hope to obtain an exact series representation of
f (x). In other words can we write any 2 periodic piecewise continuous
function f as an innite series of the following form:
f (x) =
a
0
2
+

n=1
a
n
cos(nx) + b
n
sin(nx). (3)
If this is the case, coecients a
n
and b
n
can be found as follows:
Multiplying the above by cos(mx), integrating between and
gives
(School of Mathematical Sciences) HG2M13 September 29, 2008 5 / 31
_

f (x) cos(mx)dx =
a
0
2
_

cos(mx)dx
+

n=1
_
a
n
__

cos(mx) cos(nx)dx
_
+b
n
__

sin(nx) cos(mx)dx
__
= a
0

0m
+

n=1
a
n

mn
= a
m
.
Here we have made use of the results in (2). The coecients a
m
are thus
a
m
=
1

f (x) cos(mx)dx, for all m. (4)

(School of Mathematical Sciences) HG2M13 September 29, 2008 6 / 31
A similar result can be derived for the coecients b
m
by multiplying (3) by
sin(mx) and integrating between and .
This gives
b
m
=
1

f (x) sin(mx)dx, for all m. (5)

Example
Obtain the Fourier series for the 2 -periodic sawtooth wave
f (x) =
_
_
_
x +

2
for < x < 0,

2
x for x 0.
Using (4) and (5) to calculate the coecients in the Fourier series gives
(School of Mathematical Sciences) HG2M13 September 29, 2008 7 / 31
a
0
= 0,
a
m
=
1

__

2
cos(mx)dx +
_
0

x cos(mx)dx
+
_

0
x cos(mx)dx
_
for m 1,
b
m
=
1

__

2
sin(mx)dx +
_
0

x sin(mx)dx
_

0
x sin(mx)dx
_
= 0.
Note that the rst integrals in the expressions for both a
m
and b
m
are zero
and that the second and third integral in the expression for a
m
take the
same value whilst those in the expression for b
m
cancel. Thus
(School of Mathematical Sciences) HG2M13 September 29, 2008 8 / 31
a
m
=
2

_

0
x cos(mx)dx =
2

_
x sin(mx)
m
+
cos(mx)
m
2
_

0
a
m
=
2
m
2

(1 (1)
m
) ,
and the Fourier series for f (x) is
f (x) =

m=1
2
m
2

(1 (1)
m
) cos(mx).
Plots of the rst 10 terms in the Fourier series are shown in gure (1).
(School of Mathematical Sciences) HG2M13 September 29, 2008 9 / 31
1.5
1
0.5
0.5
1
1.5
4 2 2 4
x
f (x)
x
Figure 1: The rst 10 terms in the Fourier series of f (x).
(School of Mathematical Sciences) HG2M13 September 29, 2008 10 / 31
Remark
Note that f (x) is an even function and furthermore that its Fourier series
contains no terms involving sin(mx). It is in general true that the Fourier
series for an even function f (x) takes the form
f (x) =
a
0
2
+

n=1
a
n
cos(nx),
while that for an odd function g(x) takes the form
g(x) =

n=1
b
n
sin(nx),
(convince yourself that these statements are true). Remembering these
facts can save time when evaluating the coecients in a Fourier series.
(School of Mathematical Sciences) HG2M13 September 29, 2008 11 / 31
3.1 Fourier series for 2-periodic functions
Consider now a function f (x) which has period 2. It is straightforward to
transform the problem of nding a Fourier series for f (x) on the interval
< x < into one on the interval , < x < x.
It follows that its Fourier series is given by
f (x) =
a
0
2
+

n=1
a
n
cos
_
n

x
_
+ b
n
sin
_
n

x
_
, (6)
where
a
m
=
1

f (x) cos
_
m

x
_
dx,
b
m
=
1

f (x) sin
_
m

x
_
dx
_

_
(7)
Example
Obtain the Fourier series for the periodic function
g(x) = x for 1 < x < 1.
(School of Mathematical Sciences) HG2M13 September 29, 2008 12 / 31
Note that this is an odd function in x and hence a
m
= 0 for all m. The
values of b
m
are determined using (7)
b
m
=
_
1
1
x sin(mx)dx =
_

x cos(mx)
m
+
sin(mx)
(m)
2
_
1
1
b
m
=
2
m
(1)
m
.
Substituting this expression for b
m
back into (6) gives the Fourier series as
g(x) =

n=1
2
1
n
(1)
n
sin(nx).
The rst 5 and rst 40 terms of this series are plotted in gure 2.
(School of Mathematical Sciences) HG2M13 September 29, 2008 13 / 31
1
0.5
0.5
1
1.5 1 0.5 0.5 1 1.5
x
x
g(x)
Figure 2: The rst 5 terms and the rst 40 terms in the Fourier series of
g(x).
(School of Mathematical Sciences) HG2M13 September 29, 2008 14 / 31
Remark
The Fourier series converges much more slowly for the function g(x) (as
we take more terms in the series) than it did for f (x). This is because the
periodic function g(x) has a discontinuity in contrast to f (x) which is
continuous. In particular note that the Fourier series has diculty
approximating g close to the discontinuity where it has many nescale
oscillations; this is termed Gibbs phenomenon. We consider these aspects
in more detail after a summary and some examples.
(School of Mathematical Sciences) HG2M13 September 29, 2008 15 / 31
3.2 Summary of formulae and examples
If f (x) is a periodic function with fundamental period 2 then

the Fourier series for f (x) is written

1
2
a
0
+

n=1
(a
n
cos nx + b
n
sin nx)
where a
0
, a
n
, b
n
for n = 1, 2, . . . are the Fourier coecients.

the Fourier coecients are determined by the Euler formulae

a
0
=
1

f (x)dx
a
n
=
1

f (x) cos nxdx, n = 1, 2, . . .

b
n
=
1

f (x) sin nxdx n = 1, 2, . . . .

(School of Mathematical Sciences) HG2M13 September 29, 2008 16 / 31
Example
Find the Fourier coecients of f (x) dened
f (x) = x for < x and f (x + 2) = f (x)
The graph of the function in the region 3 < x < 3 is
0
f (x)

x
3 2 2 3

(School of Mathematical Sciences) HG2M13 September 29, 2008 17 / 31

Note:
_

x cos nxdx =
_
x sin nx
n
_

sin nx
n
dx
= 0 +
_
cos nx
n
2
_

= 0
_

x sin nxdx =
_

x cos nx
n
_

+
_

cos nx
n
dx
=
2 cos n
n
+
_
sin nx
n
2
_

=
2(1)
n
n
+ 0
=
2(1)
n+1
n
since cos n = (1)
n
and cos n = (1)
n+1
.
(School of Mathematical Sciences) HG2M13 September 29, 2008 18 / 31
The Fourier coecients are
a
0
=
1

xdx
a
n
=
1

x cos nxdx n = 1, 2, . . .
b
n
=
1

x sin nxdx n = 1, 2, . . . .
Evaluating the integrals we obtain the results
a
0
= 0, a
n
= 0, b
n
=
2(1)
n+1
n
for n = 1, 2, . . . .
Hence the required Fourier series is
2
_
sin x
1
2
sin 2x +
1
3
sin 3x
1
4
sin 4x . . .
_
ie f (x) = 2

n=1
(1)
n+1
sin nx
n
.
(School of Mathematical Sciences) HG2M13 September 29, 2008 19 / 31
Example
Find the Fourier coecients of f (x) dened
f (x) = x
2
for < x , and f (x + 2) = f (x).
The graph of the function in the region 3 < x < 3 is
f (x)
0 x
(School of Mathematical Sciences) HG2M13 September 29, 2008 20 / 31
The Fourier coecients are
a
0
=
1

x
2
dx
a
n
=
1

x
2
cos nxdx n = 1, 2, . . . .
b
n
=
1

x
2
sin nxdx n = 1, 2, . . . .
Evaluating the integrals we obtain the results
a
0
=
2
2
3
, a
n
=
4(1)
n
n
2
, b
n
= 0 for n = 1.2. . . . .
Hence the required Fourier series is

2
3
+ 4
_
cos x +
1
4
cos 2x
1
9
cos 3x +
1
16
cos 4x + . . .
_
ie

2
3
+ 4

n=1
(1)
n
cos nx
n
2
.
(School of Mathematical Sciences) HG2M13 September 29, 2008 21 / 31
If f (x) is a periodic function with period 2 i.e. f (x + 2) = f (x).
The Fourier series for f (x) is
1
2
a
0
+

n=1
_
a
n
cos
_
nx

_
+ b
n
sin
_
nx

__
where
a
0
=
1

f (x)dx
a
n
=
1

f (x) cos
_
nx

_
dx n = 1, 2, . . .
b
n
=
1

f (x) sin
_
nx

_
dx n = 1, 2, . . .
(School of Mathematical Sciences) HG2M13 September 29, 2008 22 / 31
Example
If g(x) = x
2
for x <
g(x + 2) = g(x)
the Fourier series for g(x) is
1
2
a
0
+

n=1
a
n
cos
_
nx

_
where a
0
=
1

x
2
dx =
2
2
3
and a
n
=
1

x
2
cos
_
nx

_
dx =
4
2
(1)
n
n
2

2
Note:
(1) b = 0 because g(x) is an even function
(2) set = then series reduces to that in a previous example.
(School of Mathematical Sciences) HG2M13 September 29, 2008 23 / 31
3.3 Properties of Fourier Series
A Fourier series is an innite series used to represent function values and
therefore we must ask the questions

Is it convergent?

If so, to what does it converge?

Can we apply dierentiation to represent f

1
(x)?

Can we apply integration to represent

_
f (x)dx?
(School of Mathematical Sciences) HG2M13 September 29, 2008 24 / 31
Continuity

A function f (x) is continuous at x = a if

(1) the limit as x a from values of x > a, written x a
+
, is f (a) and
(2) the limit as x a from values of x < a, written x a

, is also f (a)
Examples
(1) a continuous function (2) a discontinuous function
f(x)
x
a
f(a)
x
f(x)
2
1
a
(School of Mathematical Sciences) HG2M13 September 29, 2008 25 / 31

A function is continuous in
the interval a < x < b if it
is continuous at every point
of the interval a < x < b
Example
f(x)
x
a b

A function is said to be
piecewise continuous in an
interval a < x < b if it has
at most a nite number of
nite discontinuities in the
interval a < x < b
Example
b
a
x
f(x)
(School of Mathematical Sciences) HG2M13 September 29, 2008 26 / 31
Dierentiability

A function f (x) is dierentiable at x = a if

lim
xa
_
f (x) f (a)
x a
_
exists.
If it exists the derivative is labelled f

(a).

This limit may not exist at x = a, that is f (x) may not be

dierentiable at x = a if
(1) f (x) is discontinuous
at x = a
or (2) the curve y = f (x)
does not have a tangent
at x = a.
f(x)
x
a
x
f(x)
a
(School of Mathematical Sciences) HG2M13 September 29, 2008 27 / 31
Example
f (x) =
_
0 for < x 0
x for 0 < x
f (x + 2) = f (x)
The Fourier coecients are
a
0
=
1
2
, a
n
=
((1)
n
1)
5n
2

, b
n
=
(1)
n+1
n
n = 1, 2, . . .
The Fourier series is therefore
1
4
+

n=1
_
((1)
n
1)
n
2

cos nx +
(1)
n+1
n
sin nx
_
Dene S
1
=
1
4

2

cos x + sin x
S
2
= S
1

1
2
sin 2x
S
3
= S
2

2
3
2

cos 3x +
1
3
sin 3x etc.
(School of Mathematical Sciences) HG2M13 September 29, 2008 28 / 31
S
1
S
5
S
9
(School of Mathematical Sciences) HG2M13 September 29, 2008 29 / 31
We see from this example that as n

At points where f (x) is continuous, say x = a, the sum of the Fourier

series is f (a)

At points where f (x) is discontinuous, say x = b, the sum of the

Fourier series is
1
2
_
f (b
+
) + f (b

ie mid-way between the two values.

This result is Fouriers theorem.
(School of Mathematical Sciences) HG2M13 September 29, 2008 30 / 31
A Fourier series can be used for integrating functions.
Theorem
If f (x) is piecewise continuous in < x < and periodic with period
then the Fourier series can be integrated term by term.
The Fourier series for a smooth function f (x) may be dierentiated term
by term.
Theorem
The Fourier series for f

(x) can be obtained by dierentiating the Fourier

series for f (x)
ONLY
if f (x) is continuous for all x.
(School of Mathematical Sciences) HG2M13 September 29, 2008 31 / 31