Binomial expansion, power series,

limits,
approximations, Fourier series
Notice: this material must not be used as a substitute for attending
the lectures

Binomial expansion

We know that
(a + b)1 = a + b
(a + b)2 = a2 + 2ab + b2
(a + b)3 = a3 + 3a2 b + 3ab2 + b3
The question is (at this stage): what about (a + b)n where n is any positive integer?

1.1

Pascals triangle
1
1

1
1

5
6

1
1

10

10

15

20

15

To expand (a + b)n we look for the row starting with 1 and n.

1.2

Example

Lets expand (a + b)3 . The row in Pascals triangle starting with 1 and 3 is
1

Therefore the expansion of (a + b)3 is

(a + b)3 = a3 + 3a2 b + 3ab2 + b3

1.3

Example

Lets expand (a + b)6 .

The row starting with 1 and 6 in Pascals triangle is the row
1

15

20

15

This means that the expansion of (a + b)6 is

(a + b)6 = a6 + 6a5 b + 15a4 b2 + 20a3 b3 + 15a2 b4 + 6ab5 + b6

1.4

Factorial notation

The factorial n! of a positive integer n is defined by

n! = n(n 1)(n 2) (3)(2)(1)
so for example
5! = 5 4 3 2 1 = 120
and
8! = 8 7 6 5 4 3 2 1 = 40320
We work with the convention that
1! = 1

and

0! = 1

Expressions involving factorials can often be simplified as shown in the example below:
8!
87654321
876
=
=
= 56
5! 3!
(5 4 3 2 1)(3 2 1)
6

1.5

Binomial theorem

Pascals triangle can be difficult to use if the exponent is very high. In such cases the
following binomial theorem is usually better. This states that if n is a positive
integer then
n(n 1) n2 2 n(n 1)(n 2) n3 3
a b +
a b + + bn
2!
3!

(a + b)n = an + nan1 b +

An important particular case is when a = 1 and b = x giving

(1 + x)n = 1 + nx +

n(n 1) 2 n(n 1)(n 2) 3

x +
x + + xn
2!
3!

(1.1)

which, like the previous result, holds for positive integers n.

In the binomial theorem, the general term has the form anm bm with coefficient
n(n 1)(n 2) (n (m 1))
m!
which equals
n(n 1)(n 2) (n (m 1))(n m)!
m! (n m)!
or
n!
m! (n m)!

often denoted

n
m

In terms of the notation introduced above, the binomial theorem can be written as
n

(a+b) =

n
0

!
n

a +

n
1

!
n1

b+

n
2

!
n2 2

b + +

n
n

!
n

b =

n
X
i=0

n
i

ani bi

1.6

Example


4

Expand 2 + x3 .
Solution. Using the binomial theorem:


x
2+
3

1.7

4

x
(4)(3) 2 x 2 (4)(3)(2)
x
(4)(3)(2)(1) x 4
= 24 + (4)(23 )( ) +
(2 )( ) +
(2)( )3 +
( )
3
2!
3
3!
3
4!
3
8 2
8 3
1 4
32
= 16 + x + x + x + x .
3
3
27
81

Example


15

Expand 1 + x3
up to and including the term in x3 .
Solution. By the binomial theorem:


1.8

x
1+
3

15

x
(15)(14) x 2 (15)(14)(13) x 3
= 1 + 15( ) +
( ) +
( ) +
3
2!
3
3!
3
35 2 455 3
= 1 + 5x + x +
x +
3
27

Example

Expand (1 x)3 (2 + x)6 up to and including the term in x2 .

Solution.
!

(1 x)3 (2 + x)6 = (1 x)3

(6)(5) 4 2
(2 )x +
26 + (6)(25 )x +
2!

(3)(2)
2
=
(x)2 + (x)3
1 + 3(x) +
(64 + 192x + 240x + )
| {z }
2!
redundant

= (1 3x + 3x x )(64 + 192x + 240x2 + )

= 64 + (192 (64)(3))x + (3(64) 3(192) + 240)x2
= 64 144x2 +

1.9

Powers that are NOT positive integers

The binomial expansion as discussed up to now is for the case when the exponent is
a positive integer only.
For the case when the number n is not a positive integer the binomial theorem
becomes, for 1 < x < 1,
(1 + x)n = 1 + nx +

n(n 1) 2 n(n 1)(n 2) 3

x +
x +
2!
3!

(1.2)

This might look the same as the binomial expansion given by expression (1.1), but
let us make the following important distinctions between (1.1) and (1.2):

 the expansion for positive integer powers (expansion (1.1)) terminates, i.e. it
has only a finite number of terms. However, for powers that are not positive
integers the series (1.2) is an infinite series that goes on forever.
it can be mathematically proven that the series (1.2) is valid only for 1 < x <
1.
expression (1.2) cannot be applied to something of the form (a + x)n . Such an
expression must first be rewritten as follows:
x
(a + x) = a 1 +
a
 

n

=a

x n
1+
|
{za }


apply binomial to this

1.10

Example

Expand 1 + 2x and state what values of x the series is valid.

Solution.

1 + 2x = (1 + 2x)1/2
( 1 )( 12 )
( 1 )( 21 )( 32 )
( 1 )( 12 )( 23 )( 52 )
1
= 1 + (2x) + 2
(2x)2 + 2
(2x)3 + 2
(2x)4 +
2
2!
3!
4!
1
1
5
= 1 + x x2 + x3 x4 +
2
2
8
This series is valid when 1 < 2x < 1. i.e. when 12 < x < 12 .

1.11

Example


Expand 1
Solution.


x
1
2

5

x
2

5

. For what values of x is the expansion valid?

x
(5)(6)
x
= 1 + (5)
+

2
2!
2
5
15 2 35 3
= 1 + x + x + x +
2
4
8


2

x
(5)(6)(7)
+

3!
2


3

This is valid when 1 < x2 < 1, i.e. when 2 < x < 2.

1.12

Example
1

Expand (3 + x) 2 .
Solution. Remember that when the power is not a positive integer your expression
has to be of the form (1 + something)power . Deal with this as follows:
12

(3 + x)

x
3 1+
3

 

 1
2

12

=3

x
1+
|
{z3

 1

expand this

( 12 )( 32 ) x 2
1 x
= 3
1 + ( )( ) +
( ) +
2 3
2!
3
!
2
1
x x
= 1 +
+
6 24
3

21

This is valid when 1 < x/3 < 1, i.e. when 3 < x < 3.

1.13

Example
1/2

for the cases (i) |x| > 1 and (ii) 0 < x < 1.
Find expansions for 1 + x1
Solution. the following calculation produces an expansion which will be valid
when 1/|x| < 1, i.e. |x| > 1:


1
1+
x

( 1 )( 21 ) 1 2 ( 12 )( 12 )( 23 ) 1
1 1
= 1+
+ 2
+
2 x
2!
x
3!
x
1
1
1
= 1+

+
+
2x 8x2 16x3

1/2

 

 3

 

valid for |x| > 1.

The above expansion is no good if |x| < 1. For this case the following trick
produces a valid expansion:


1
1+
x

1/2

x+1
x

1/2

1
x1/2

(1 + x)1/2
|

{z

expand this
1
( 2 )( 12 ) 2

( 1 )( 12 )( 23 ) 3
1
= 1/2 1 + x +
x + 2
x +
x
2
2!
3!


1
1
1 2
1 3
= 1/2 1 + x x + x +
x
2
8
16
1 1/2 1 3/2
1
1
= 1/2 + x x + x5/2 +
x
2
8
16
1

Note that this is actually defined only for 0 < x < 1.

1.14

Example
2

(1+x)
2
Expand (1x/2)
3 up to and including the term in x .
Solution.

(1 + x)2
x
2
=
(1
+
x)
1

(1 x/2)3
2


3

x
(3)(4)
x
= (1 + 2x + x ) 1 + (3)
+

2
2!
2
!
2
3x 3x
= (1 + 2x + x2 ) 1 +
+
+
2
2


2

3
3
3 2
+2 x+ 1+2
+
x +
2
2
2
7x 11x2
= 1+
+

2
2


 

= 1+

2
2.1

Taylor and Maclaurin series

Taylor series

The idea is to expand a function f (x) about a point a in the form of a sum of powers
of (x a), i.e. to form a series of the form
f (x) = a0 + a1 (x a) + a2 (x a)2 + a3 (x a)3 + =

an (x a)n

(2.3)

n=0

we want to know the coefficients an , n = 0, 1, 2, . . . in the above expansion.

If we differentiate expression (2.3) again and again, we get the following expressions for the first, second, third, etc derivatives of f (x):
f 0 (x) = a1 + 2a2 (x a) + 3a3 (x a)2 + 4a4 (x a)3 +
f 00 (x) = 2a2 + (3)(2)a3 (x a) + (4)(3)a4 (x a)2 +
f 000 (x) = (3)(2)a3 + (4)(3)(2)a4 (x a) +
..
..
.
.
Putting x = a in these expressions gives
f 0 (a) = a1

a1 = f 0 (a)
1
a2 = f 00 (a)
2

f 00 (a) = 2a2

f 000 (a) = (3)(2)a3

a3 =

1
f 000 (a)
(2)(3)

Spotting the pattern, we see that the general formula for the coefficient an will be
an =

1 (n)
f (a)
n!

where f (n) (a) means the nth derivative of f (x), evaluated at the value x = a.
This gives us what we call the Taylor expansion of a function f (x) valid for
values of x near to a:
f (x) = f (a) + (x a)f 0 (a) +

(x a)3 000
(x a)2 00
f (a) +
f (a) +
2!
3!
n

(2.4)

The series carries on to infinity, and has general term (xa)

f (n) (a).
n!
Taylors expansion, and the related Maclaurin expansion discussed below, are
used in approximations. In practice usually only the first few terms in the series are
kept and the rest are discarded. The idea is that the resulting truncated expansion
should provide a good approximation to the function f (x) for values of x close to the
particular value a. The more terms we keep, the better the approximation.

2.2

Maclaurin series

There is also the Maclaurin expansion, which is just the Taylor expansion in the
particular case when a = 0, i.e.
f (x) = f (0) + xf 0 (0) +
or, in summation notation
f (x) =

x2 00
x3
f (0) + f 000 (0) +
2!
3!

X
n=0

f (n) (0)

(2.5)

xn
n!

Not all functions have Taylor or Maclaurin expansions but most do.

2.3

Example

Let us find the Maclaurin series of ex .

Solution. Let f (x) = ex .
Then f (0) = 1.
Also f 0 (x) = ex so f 0 (0) = 1.
f 00 (x) = ex so f 00 (0) = 1. Clearly in this particular example f (n) (0) = 1 for all
n = 1, 2, 3, . . .. Putting these values for f (0), f 0 (0), f 00 (0), etc, into (2.5) gives us the
Maclaurin series for the particular function f (x) = ex , namely
ex = 1 + x +
or, in summation notation, ex =

2.4

x2 x3
+
+
2!
3!

xn
n=0 n!

Example

Deduce the Maclaurin series of e5x from that for ex .

Solution. Just replace every x by 5x in expression (2.6) above to get
(5x)2 (5x)3
+
+
2!
3!
25x2 125x3
= 1 + 5x +
+
+
2
6

e5x = 1 + 5x +

2.5

Example

Find the Maclaurin series of cos x.

Solution. Let f (x) = cos x.
Then f (0) = 1.
Also f 0 (x) = sin x so f 0 (0) = 0.
f 00 (x) = cos x so f 00 (0) = 1.
f 000 (x) = sin x so f 000 (0) = 0.

(2.6)

f 0000 (x) = cos x so f 0000 (0) = 1.

f 00000 (x) = sin x so f 00000 (0) = 0.
We see the pattern emerging. The values f (0), f 0 (0), f 00 (0), f 000 (0), etc, cycle through the values 1, 0, 1, 0, 1, 0, 1, 0, . . .. Putting these values into the general Maclaurin expansion (2.5) gives the Maclaurin expansion for the function cos x,
namely
x2 x4
cos x = 1
+

2!
4!
or, in summation notation,

X
(1)n x2n
cos x =
(2n)!
n=0
Similarly, it can be shown that the Maclaurin expansion of sin x is
sin x = x

2.6

x3 x5
+

3!
5!

Example

Find the Taylor series of the function f (x) = 1/x about x = 2.

Solution. We are asked for a Taylor series here, not the Maclaurin one. The
relevant formula is therefore (2.4) in the case when a = 2. So we need to work out
the values f (2), f 0 (2), f 00 (2), etc. We do this next:
f (2) = 12 .
f 0 (x) = x12 so f 0 (2) = 14 .
f 00 (x) = x23 so f 00 (2) = 14 .
f 000 (x) = x64 so f 000 (2) = 83 ,
and so on. The Taylor series about the value x = 2 is
f (x) = f (2) + (x 2)f 0 (2) +

(x 2)2 00
(x 2)3 000
f (2) +
f (2) +
2!
3!

which becomes, since f (x) = 1/x,

1
1 1
1
1
= (x 2) + (x 2)2 (x 2)3 +
x
2 4
8
16
What this means, is that the first few terms of the above series expansion will constitute a good approximation to 1/x for values of x close to 2.
Note that the function f (x) = 1/x does not have a Taylor series expansion about
the point x = 0. This is because this function goes to infinity as x 0, so we could
hardly expect the function to have an approximation for small values of x as a series
of powers of x. Had we attempted to find f (0), f 0 (0), f 00 (0), etc, they would all turn
out to be infinity.

2.7

Example

Find the first three non-zero terms of the Maclaurin series of e2x sin x.
Solution. One way to do this would be to write down the Maclaurin series for
e2x (which can be inferred from the one for ex by replacing every x by 2x) and
the series for sin x and then multiplying the series together and expanding out. The
approach below is a direct approach not requiring such advance knowledge of the two
separate Maclaurin expansions.
Let f (x) = e2x sin x. Then f (0) = 0.
f 0 (x) = e2x cos x 2e2x sin x so f 0 (0) = 1. Differentiating again
f 00 (x) = e2x ( sin x) 2e2x cos x 2(e2x cos x 2e2x sin x)
= 3e2x sin x 4e2x cos x
and
f 000 (x) = 3(e2x cos x 2e2x sin x) 4(e2x sin x 2e2x cos x)
From these expressions we get f 00 (0) = 4 and f 000 (0) = 11. Putting these values into
the general Maclaurin series (2.5) gives the following expression for our particular
function f (x) = e2x sin x:
e2x sin x = x 2x2 +

11x3
+
6

which will constitute a good approximation to e2x sin x provided x is reasonably

small.

2.8

Example

Find the binomial expansion of (1 x2 )1/2 and deduce from it a power series expansion for sin1 x.
Solution. First we find the expansion of (1 + x)1/2 .
1/2

(1 + x)

( 12 )( 23 ) 2 ( 21 )( 23 )( 25 ) 3
1
x +
x +
= 1 + ( )x +
2
2!
3!
1
5
3
= 1 x + x2 x3 +
2
8
16

In the above, we now replace every x by x2 to deduce that

3
5
1
(1 x2 )1/2 = 1 (x2 ) + (x2 )2 (x2 )3 +
2
8
16
1 2 3 4
5 6
= 1 + x + x + x +
2
8
16
Now
sin

x =

Z x
0

dt
1 t2

10

Z x

(1 t2 )1/2 dt

Z x

3
5
1
1 + t2 + t4 + t6 + dt
2
8
16
0
3 5
5 7
1 3
x +
= x+ x + x +
6
40
112
=

Applications to working out limits

The notation
lim f (x)

xa

means the value (if any) that f (x) approaches, when x approaches a. The word lim
means limit.

3.1

Important issues to do with limits

Two trivial examples of working out limits would be

lim (x2 3) = 1,

lim cos x = 1

x2

x0

In the above examples we can just put the value in. But in many situations we cannot
do this because we end up with the mathematically meaningless expression 00 which
could be anything.
For example, lets work out
x2 4
lim
x2 x 2
In this example we cannot put x = 2 into the expression otherwise we get 00 which
could be anything. But we can simplify the expression by factorising and cancelling
factors to get
(x 2)(x + 2)
x2 4
= lim
= lim (x + 2) = 4
x2
x2
x2 x 2
x2
lim

Similarly, lets work out

x2 + x 2
x1
x2 x
lim

Again we cannot just put x = 1 into this expression or we would get 00 . But we can
factorise and simplify as follows:
x2 + x 2
(x 1)(x + 2)
x+2
=
lim
=
lim
= 3.
x1
x1
x1
x2 x
x(x 1)
x
lim

It is not always possible to work out limits simply by looking for factors and simplifying as in the above examples. We now want to add binomial expansion and
Taylor/Maclaurin series to our list of methods for working out limits.

11

3.2

Example

Lets work out

(1 + x/2)5/7 1
x0
x
Again, we cannot put x = 0 into this expression as it stands. But we can use binomial
expansion, as follows;
lim


5/7

(1 + x/2)
x

1 + ( 75 )( x2 ) +

=
=

( 75 )( 27 ) x 2
(2)
2!

+ 1

x
5
x
14

5
x2
196

x
5
5
=

x +
14 196
We can let x 0 in the above expression to deduce that
5
(1 + x/2)5/7 1
=
x0
x
14
lim

3.3

Example

Lets work out

sin x
x0 x
Solution. We mentioned earlier that
lim

sin x = x

and

sin 2x
x0
x
lim

x3 x5 x7
+

+
3!
5!
7!

Hence

sin x
x2 x4 x6
=1
+

+
x
3!
5!
7!
We can let x 0 in this to deduce that
sin x
lim
=1
x0 x
From the Maclaurin expansion for sin x given above, we can deduce the expansion for
sin 2x to be
(2x)3 (2x)5
+

sin 2x = 2x
3!
5!
4x3 32x5
= 2x
+

3
120
Hence
sin 2x
4x2
=2
+
x
3
Letting x 0 we deduce that
sin 2x
lim
=2
x0
x
It is in fact a general result that limx0 sinxkx = k for any constant k.

12

3.4

Example

Find

sin2 x x2 cos x
x0
x4
lim

Solution. Recall that

sin x = x

x3 x5 x7
+

+
3!
5!
7!

cos x = 1

x2 x4 x6
+

+
2!
4!
6!

and
Squaring the formula for sin x gives

x5
x3
x5
x3
+
x
+

sin x = x
6
120
6
120
x4
x4
= x2
+ (something) x6
+ (something) x6
6
6
4
x
= x2
+ (something) x6
3
Hence, using also the expansion for cos x given above, we have
!

sin2 x x2 cos x
=
x4

x2

x4
3

+ (something) x6 + x2 1

x2
2

x4
24

x4
1 4
x + (something) x6 + even higher powers of x
= 6
x4
1
=
+ (something) x2 +
6
Let x 0 in the above to get
1
sin2 x x2 cos x
=
4
x0
x
6
lim

3.5

Example

Find
lim x(e1/x 1)

Solution. To deal with x going to infinity, we shall let y = 1/x and let y 0. This
gives
1 y
1/x
(e 1)
lim
x(e

1)
=
lim
x
y0 y
(
)
!
1
(y)2
= lim
1 + (y) +
+ 1
y0 y
2!


y
= lim 1 + +
y0
2!
= 1
where we have used the Maclaurin expansion for the exponential, given by (2.6).

13

LHopitals rule

Another way of working out a limit when in a

if f (a) = 0 and g(a) = 0 then

0
0

situation is the following result:

f (x)
f 0 (x)
lim
= lim 0
xa g(x)
xa g (x)

The above result is called LHopitals rule.

It is absolutely crucial to check the condition f (a) = 0 and g(a) = 0 before using
the rule, because it does not work otherwise.

4.1

Example
3x sin x
x0
x
3 cos x
= lim
x0
1
3

cos
0
=
1
lim

would be

0
0

no longer

0
0

if we put x = 0 in, so use LHopital

=2

4.2

Example
1 cos x
x0 x + x2
sin x
= lim
x0 1 + 2x
= 01
lim

would be

0
0

no longer

0
0

if we put x = 0 in, so use LHopital

=0

4.3

Example
lim

x2

x2
1
1
= lim
=
2
x2
x 4
2x
4

Sometimes we have to apply LHopitals rule more than once to get an answer, as the
next example illustrates:

14

4.4

Example
x sin x
x0
x3
1 cos x
= lim
x0
3x2
sin x
= lim
x0 6x
cos x
= lim
x0
6
1
=6

0
0

lim

4.5

still

0
0

so use LHopital again

still

0
0

so use LHopital again

no longer

0
0

Example
ln cos x
x0 ln cos 3x

0
0

lim

= lim 

sin x
cos
x

so use LHopital

sin 3x
3cos
3x
tan x
= lim
x0 3 tan 3x
sec2 x
= lim
x0 9 sec2 3x
= 19
x0

so use LHopital

now simplify this

still

0
0

so use LHopital again

no longer

0
0

Fourier Series

A Fourier Series is an expansion of a periodic function as an infinite sum of sines

and cosines.
Simple examples of periodic functions (other than sin and cos) are the square wave
and sawtooth functions. An example of a square wave function (of period 4 in this
particular case) is the periodic function
(

f (t) =

1, 2 < t < 0,
1, 0 < t < 2,

with f (t + 4) = f (t).
An example of a sawtooth function of period 2 would be the periodic function
of period 2 such that f (t) = t for t (, ). Since this function has period 2 we
might suppose that it has an expansion in terms of the functions cos t, cos 2t, cos 3t, . . .
and the functions sin t, sin 2t, sin 3t, . . . since these functions also have period 2. Such
an expansion does indeed exist and in fact any periodic function of period 2 has an
expansion in terms of these trigonometric functions.
If the period is T rather than 2 this is no particular problem. All we have to do
is modify the period of the cos and sin functions we work with, i.e. we instead seek

15

an expansion in terms of the functions cos 2nt

and sin 2nt
for n = 1, 2, 3 . . ., rather
T
T
than cos nt and sin nt. Letting f (t) be a T0 periodic function, this expansion, called
the Fourier series of f (t), turns out to be
f (t) =

1
a
2 0


X
n=1

2nt
2nt
an cos
+ bn sin
T0
T0

(5.7)

where
2 Z T0
2nt
f (t) cos
dt,
T0 0
T0
2 Z T0
2nt
=
dt,
f (t) sin
T0 0
T0

an =

n = 0, 1, 2, 3, . . .

(5.8)

bn

n = 1, 2, 3, . . .

(5.9)

A number of important points need to be made:

When working out the integrals in (5.8,5.9) you can in fact use any interval of
length T0 . As a consequence, the alternative formulae:
2 Z T0 /2
2nt
f (t) cos
dt,
T0 T0 /2
T0
2 Z T0 /2
2nt
=
dt,
f (t) sin
T0 T0 /2
T0

an =

n = 0, 1, 2, 3, . . .

(5.10)

bn

n = 1, 2, 3, . . .

(5.11)

will work just as well.

to work out a0 in (5.7) you use the an formula (either (5.8) or (5.10)) with
n = 0. You will sometimes find that the n = 0 case needs to be dealt with
separately from the other an coefficients due to division by zero problems.
the quantity T0 is the period of the wave so the frequency would be 1/T0 ,
usually measured in cycles per second. It is, however, more usual to define the
frequency to be the quantity 0 defined by
0 =

2
T0

rather than

1
T0

often we want to work out the Fourier series of a periodic function that contains
points of discontinuity (the abovementioned square wave and sawtooth functions
being examples). It is known that, at a point of discontinuity (at x = a, say)
the Fourier series of the function converges to
1
(f (a+)
2

+ f (a))

rather than to f (a). This applies regardless of how f (t) is defined (if it is defined
at all) at the point a itself. In the above formula the notation f (a+) means
the value just to the right of the discontinuity and f (a) means the value to
the left. More formally, f (a+) is the limit of f (a + h) as h tends to zero from
above, and f (a) is the limit of f (a h) as h tends to zero from above.

16

5.1

Example

Let

f (t) =

1 < t < 0
1 0<t<

with f (t + 2) = f (t). Find the Fourier series of f (t).

Solution. In this case the period T0 is given by T0 = 2. Let us find an first. The
following formulae will be useful
cos n = (1)n

and

sin n = 0,

n = 0, 1, 2, 3, . . .

We have
an =
=
=
=
an =

2 Z T0 /2
2nt
dt
f (t) cos
T0 T0 /2
T0
1Z
f (t) cos nt dt

1Z0
1Z
(1) cos nt dt +
cos nt dt

0




1 sin nt
1 sin nt 0
+

n 0

Because of the n in the denominator of the above calculations we need to find a0

separately, but it turns out also to be zero. We would warn you in advance, however,
that in plenty of other situations a separate calculation for a0 is absolutely essential
for a correct Fourier series.
Now lets find bn . we have
bn =
=
=
=
=

1Z
f (t) sin nt dt

Z 0

Z
1
(1) sin nt dt +
sin nt dt
"
0



 #
cos nt
1 cos nt 0
+

n
n
0
1
(1 cos(n) + ( cos n + 1))
n
1
(2 2(1)n )
n

and so

2
(1 (1)n )
n
With all the an , n = 0, 1, 2, . . ., equal to zero, and also recalling that the period
T0 = 2, the Fourier series becomes
bn =

f (t) =

bn sin nt

n=1

= b1 sin t + b2 sin 2t + b3 sin 3t +

17

i.e.
f (t) =

4
4
4
sin t +
sin 3t +
sin 5t +

3
5

Even and odd functions

If a function is an even function or an odd function then certain simplifications are

possible in the calculations required for computing the Fourier series. But note that
plenty of functions are neither even nor odd, eg f (t) = t2 + t.

6.1

Even functions

f (t) is said to be an even function if f (t) = f (t). This means the graph is
symmetrical about the y-axis.
Examples of even functions are f (t) = constant, f (t) = t2 , f (t) = t4 , f (t) = t6 , . . .
(all even powers of t); also f (t) = cos t and f (t) = cosh t.
An even function has the important property that
Z a

f (t) dt = 2

6.2

Z a

f (t) dt

Odd functions

f (t) is said to be an odd function if f (t) = f (t). This means the graph has 180o
rotational symmetry about the origin.
Examples of odd functions are f (t) = t, f (t) = t3 , f (t) = t5 , . . . (all odd powers
of t); also f (t) = sin t and f (t) = sinh t.
An odd function has the important property that
Z a

f (t) dt = 0

6.3

Useful rules of even and odd functions

even even
even odd
odd even
odd odd

6.4

=
=
=
=

even
odd
odd
even

Fourier series of an even function

Suppose that f (t) is an even function and we want its Fourier series. Since sin t is
an odd function we might anticipate that the Fourier series of an even function will
contain no sine terms. We shall show that this is indeed the case. With f (t) being

18

even the bn Fourier coefficient is given by

2 Z T0 /2
2nt
f (t) sin
dt
=
|
{z
}
T0 T0 /2
T0

bn

even

{z

odd

2 Z T0 /2
(something odd) dt
T0 T0 /2
= 0
=

So if f (t) is even we can declare from the outset that the bn terms are all zero, and we
only need to work out an , n = 0, 1, 2, . . ., a separate calculation often being needed
for a0 . With f (t) being even we get an alternative formula for an as follows:
an

2nt
2 Z T0 /2
f (t) cos
dt
=
{z
}
|
T0 T0 /2
T0
even

{z

even

2nt
4 Z T0 /2
f (t) cos
dt
=
T0 0
T0
This can save us time and effort.

6.5

Fourier series of an odd function

If f (t) is odd then the an coefficients (including a0 ) are zero because

an =

2 Z T0 /2
2nt
dt
f (t) cos
T0 T0 /2 | {z } | {zT0 }
odd

even

2 Z T0 /2
=
(something odd) dt
T0 T0 /2
= 0
Thus the Fourier series of an odd function contains only sine terms. Moreover, calculation of the bn coefficients of these sine terms can be simplified by exploiting the
oddness property.

6.6

Example

Find the Fourier series of the sawtooth function given by

f (t) = t

when

2<t<2

with f (t + 4) = f (t) for all t (i.e. the function has period 4).
Solution. This function is odd. Its graph has 180o rotational symmetry about
the origin. Since it is odd, we can immediately say that an = 0 for all n (including
n = 0) and we only need to calculate bn .

19

Also note that since the period is 4, we have in this case T0 = 4. We now find bn :
2 Z T0 /2
2nt
dt
f (t) sin
T0 T0 /2
T0
2Z 2
2nt
=
dt
f (t) sin
4 2
4
1Z 2
nt
=
t sin
dt
odd odd = even
|{z}
2 2
| {z 2 }

bn =

odd

odd

Z 2

nt
=
dt
t sin
2
0
#2 Z
"
2 cos nt
t cos nt
2
2
+
dt
=
n/2
n/2
0
0
4
2 sin nt
2
=
cos n +
n
n n/2
"

#2
0

so that

4
(1)n
n
Recalling that T0 = 4, the Fourier series is
bn =


X

4
nt
f (t) =
(1)n sin
n
2
n=1

n
4 X (1)
nt
=
sin
n=1 n
2


or, in expanded form,

t 1
3t 1
4
1
f (t) =
sin
+ 2 sin t sin
+ sin 2t +

2
3
2
4


6.7

Example

Find the Fourier series of the function such that

f (t) = t2 + t

for

<t<

with f (t + 2) = f (t) for all t.

Solution. This function is neither even nor odd. It is 2-periodic so T0 = 2.
Using this value for T0 the formula for an becomes
an

1Z
=
f (t) cos nt dt

1Z 2
=
(t + t) cos nt dt
"
#


Z
1
sin nt
sin nt
2
=
(t + t)

(2t + 1)
dt

n
n

20

if n 6= 0

1 Z
(2t + 1) sin nt dt
n "
#


Z
1
cos nt
2 cos nt
=
(2t + 1)
+
dt
n
n
n

"

 #
(2 + 1)(1)n (2() + 1)(1)n 2 sin nt
1

+
+
=
n
n
n
n
n
"
#
n
1
4(1)
4
=

= 2 (1)n
n
n
n
=

We have shown that

4
(1)n if n 6= 0
2
n
A separate calculation has to be done for a0 , since we obviously cannot put n = 0
into the above formula for an . Putting n = 0 into the original an integral gives
an =

a0

"
#
1Z 2
1 t3 t2
=
(t + t) dt =
+

3
2

a0 =

2 2
3

Next we find bn . We have

bn =
=
=
=
=

(t2 + t) sin nt dt

"

Z
cos nt
cos nt
+
(t + t)
(2t + 1)
dt
n
n

#
"
(1)n
(1)n 1 Z
2
2
( + )
+ ( )
+
(2t + 1) cos nt dt
n
n
n
"
"
##

Z
(1)n 1
sin nt
2 sin nt
2
+
(2t + 1)

dt
n
n
n
n

"
#
(1)n
2 Z
2
2
2 sin nt dt
n
n
2

so that

2
bn = (1)n
n
Putting the formulae for a0 , an and bn into the general Fourier series expansion, which
when T0 = 2 is
f (t) = 12 a0 +

(an cos nt + bn sin nt)

n=1

gives

2 X
4
2
f (t) =
+
(1)n cos nt (1)n sin nt
2
3
n
n=1 n

21

Complex form of a Fourier series

Recall that
j=

Recall also, Eulers formula

ejt = cos t + j sin t

(7.12)

and the other useful version of it:

ejt = cos t j sin t

(7.13)

It is easy to see where (7.12) comes from. We simply expand ejt using Maclaurin
series:
(jt)2 (jt)3
ejt = 1 + jt +
+
+
2!
3!
!
!
t3 t5
t2 t4
= 1 + + j t +
2! 4!
3! 5!
= cos t + j sin t
Formula (7.13) follows from (7.12) when we replace t by t.
Now, if we add equations (7.12) and (7.13) the sin t terms disappear and we get
the following result:
ejt + ejt
(7.14)
cos t =
2
Similarly, subtracting (7.13) from (7.12) gives
ejt ejt
sin t =
2j

(7.15)

Since cos t and sin t are periodic of period 2, so is ejt . Furthermore, so are the
functions enjt for any integer n (positive or negative). This leads us to suppose
that any 2 periodic function f (t) can be represented as a sum of the functions enjt
involving all integers n (positive and negative).
With suitable adjustments to the period it can be shown that a similar statement
can be made for a function f (t) of period T0 . In fact
f (t) =

cn e2jnt/T0

(7.16)

n=

where

1 Z T0 /2
f (t)e2jnt/T0 dt
for n = 0, 1, 2, 3, . . .
(7.17)
cn =
T0 T0 /2
Expression (7.16) with coefficients cn given by (7.17) is called the complex form
of the Fourier series. Note that the sum is over all integer values of n including
negative ones.
The convergence properties of the complex form are the same as for the form we
discussed earlier (expression (5.7)), i.e. the infinite sum in (7.16) converges to f (t)
unless there us a jump discontinuity in which case the sum converges to the midpoint
of the jump (regardless of the actual value of f (t) at such a point).

22

7.1

Example

Find the complex Fourier series of the function f (t) such that

f (t) =

0 < t < /2
1 /2 < t < /2

0 /2 < t <

with f (t + 2) = f (t) for all t.

Solution. The period is 2 so T0 = 2. With this value of T0 expressions (7.16)
and (7.17) reduce to

f (t) =

cn ejnt

n=

where
1 Z
f (t)ejnt dt
2
"
#/2
1 ejnt
1 Z /2 jnt
e
dt =
=
2 /2
2 jn /2

cn =

i
1 h jn/2
e
ejn/2
2jn
"
#
1 ejn/2 ejn/2
=
n
2j

jx

jx

Recalling the formula sin x = e e

, the above formula for cn can be put into the
2j
form
1
n
sin
for n = 1, 2, 3, . . .
cn =
n
2
The above calculation does not work if n = 0, because it has n in the denominator,
so we do a separate calculation for n = 0. The first line of the above calculation for
cn , with n = 0, gives
1 Z
1 Z /2
1
c0 =
f (t) dt =
dt =
2
2 /2
2
So the complex Fourier series of the function is

f (t) =

cn ejnt

n=

with the above expressions for cn and c0 .

It is possible to convert the complex form into a real form, as follows. We can
write it in the form
f (t) =

1
2

1
X

cn ejnt +

n=

23

n=1

cn ejnt

in which the 21 at the front is the n = 0 term of the sum. Making the substitution
n = m in the first sum of the above expression gives us
f (t) =

1
2

cm ejmt +

m=1

cn ejnt

n=1

and we can now simply replace m by n in the first sum, since they are dummy variables
playing a similar role to the variable in a definite integral. This observation gives
f (t) =

1
2

1
2

1
2

jnt

cn e

n=1

X

cn ejnt

insert expression for cn

n=1

1
n jnt
1
n jnt
+
sin
e
+
sin
e
2
n
2
n=1 n


X
1
n  jnt
+
sin
e
+ ejnt
2 |
{z
}
n=1 n

=2 cos nt

and so
f (t) =

1
2

n
2
sin
cos nt
2
n=1 n

We have converted the complex form into a real form. In the above sum the terms
with n = 2, 4, 6, . . . are all zero. Replacing n by 2n 1 has the effect of removing
these zero terms to give
f (t) =

1
2

1
2

2
(2n 1)
sin
cos(2n 1)t
2
n=1 (2n 1)

2(1)n+1
cos(2n 1)t
n=1 (2n 1)

which is more computationally efficient.

Amplitude and phase spectrum

Recall that the Fourier series of a periodic function of period T0 (and frequency
0 = 2/T0 ) is


X
2nt
2nt
1
an cos
f (t) = 2 a0 +
+ bn sin
T0
T0
n=1
with certain formulae, namely (5.8) and (5.9), for the coefficients an and bn .
We can write the Fourier series in terms of 0 as
f (t) =
=

1
a
2 0
1
a
2 0

+
+

X
n=1

(an cos n0 t + bn sin n0 t)

An cos(n0 t + n )

n=1

24

where An is called the amplitude and n the phase. To find formulae for the numbers
An and n , n = 1, 2, 3, . . ., we use elementary trigonometry. Now
An cos(n0 t + n ) = An cos n0 t cos n An sin n0 t sin n
Comparing the right hand side of the above expression with
an cos n0 t + bn sin n0 t
we see that we need
An cos n = an

and

An sin n = bn

These expressions imply that

An =

a2n + b2n

and

bn
an
The numbers n , n = 1, 2, 3, . . . are called the phase spectrum and the numbers
An , n = 1, 2, 3, . . . are called the amplitude spectrum. These two sets of numbers together form the spectrum and (with the frequency 0 ) constitute one way of
describing a periodic function.
tan n =

25