Binomial expansion, power series,

limits,
approximations, Fourier series

Notice: this material must not be used as a substitute for attending

the lectures

1
1 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 Pascal’s triangle

1
1 1
1 2 1
1 3 3 1
1 4 6 4 1
1 5 10 10 5 1
1 6 15 20 15 6 1

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

1.2 Example
Let’s expand (a + b)3 . The row in Pascal’s triangle starting with 1 and 3 is

1 3 3 1

Therefore the expansion of (a + b)3 is

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

1.3 Example
Let’s expand (a + b)6 .
The row starting with 1 and 6 in Pascal’s triangle is the row

1 6 15 20 15 6 1

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

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

2
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! 8×7×6×5×4×3×2×1 8×7×6
= = = 56
5! 3! (5 × 4 × 3 × 2 × 1)(3 × 2 × 1) 6

1.5 Binomial theorem

Pascal’s 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) n−2 2 n(n − 1)(n − 2) n−3 3
(a + b)n = an + nan−1 b + a b + a b + · · · + bn
2! 3!
An important particular case is when a = 1 and b = x giving
n(n − 1) 2 n(n − 1)(n − 2) 3
(1 + x)n = 1 + nx + x + x + · · · + xn (1.1)
2! 3!
which, like the previous result, holds for positive integers n.
In the binomial theorem, the general term has the form an−m 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! n
often denoted
m! (n − m)! m
In terms of the notation introduced above, the binomial theorem can be written as
n
! ! ! ! !
n n n n n−1 n n−2 2 n n n
an−i bi
X
(a+b) = a + a b+ a b +· · ·+ b =
0 1 2 n i=0
i

3
1.6 Example
 4
Expand 2 + x3 .
Solution. Using the binomial theorem:
4
x x (4)(3) 2 x 2 (4)(3)(2) x (4)(3)(2)(1) x 4

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

1.7 Example
 15
Expand 1 + x3 up to and including the term in x3 .
Solution. By the binomial theorem:
15
x x (15)(14) x 2 (15)(14)(13) x 3

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

1.8 Example
Expand (1 − x)3 (2 + x)6 up to and including the term in x2 .
Solution.
!
(6)(5) 4 2
(1 − x)3 (2 + x)6 = (1 − x)3 26 + (6)(25 )x + (2 )x + · · ·
2!
 
(3)(2)
= 
1 + 3(−x) + (−x)2 + (−x)3  2
 (64 + 192x + 240x + · · ·)
2! | {z }
redundant
= (1 − 3x + 3x − x )(64 + 192x + 240x2 + · · ·)
2 3

= 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,

n(n − 1) 2 n(n − 1)(n − 2) 3

(1 + x)n = 1 + nx + x + x + ··· (1.2)
2! 3!
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):

4
 • 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:
n
x x n
   
n n
(a + x) = a 1 + =a 1+
a | {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 ( 1 )(− 12 ) ( 1 )(− 21 )(− 32 ) ( 1 )(− 12 )(− 23 )(− 52 )
= 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
 −5
x
Expand 1 − 2
. For what values of x is the expansion valid?
Solution.
−5 2 3
x x (−5)(−6) x (−5)(−6)(−7) x
    
1− = 1 + (−5) − + − + − + ···
2 2 2! 2 3! 2
5 15 2 35 3
= 1 + x + x + x + ···
2 4 8
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:
− 1 − 1
x x
  
2 2
− 12 − 12
(3 + x) = 3 1+ =3 1+
3 | {z3 }
expand this

5
 (− 12 )(− 32 ) x 2
!
− 21 1 x
= 3 1 + (− )( ) + ( ) + ···
2 3 2! 3
2
!
1 x x
= √ 1− + + ···
3 6 24

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

1.13 Example
 1/2
Find expansions for 1 + x1 for the cases (i) |x| > 1 and (ii) 0 < x < 1.
Solution. the following calculation produces an expansion which will be valid
when 1/|x| < 1, i.e. |x| > 1:

( 1 )(− 21 ) 1 2 ( 12 )(− 12 )(− 23 ) 1

1/2  3
1 1 1
    
1+ = 1+ + 2 + + ···
x 2 x 2! x 3! x
1 1 1
= 1+ − + + ···
2x 8x2 16x3
valid for |x| > 1.
The above expansion is no good if |x| < 1. For this case the following trick
produces a valid expansion:
1/2 1/2
1 x+1 1
 
1+ = = (1 + x)1/2
x x x1/2 | {z }
expand this

( 2 )(− 12 ) 2
1
( 1 )(− 12 )(− 23 ) 3
!
1 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 1/2 1 3/2 1
= 1/2 + x − x + x5/2 + · · ·
x 2 8 16
Note that this is actually defined only for 0 < x < 1.

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

Solution.
(1 + x)2
−3
x

2
= (1 + x) 1 −
(1 − x/2)3 2
2 !
x (−3)(−4) x
  
2
= (1 + 2x + x ) 1 + (−3) − + − + ···
2 2! 2
2
!
3x 3x
= (1 + 2x + x2 ) 1 + + + ···
2 2

6
 3 3 3 2
     
= 1+ +2 x+ 1+2 + x + ···
2 2 2
7x 11x2
= 1+ + ···
2 2

2 Taylor and Maclaurin series

2.1 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
X
(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 expres-
sions 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
f 00 (a) = 2a2 ⇒ a2 = f 00 (a)
2
1
f 000 (a) = (3)(2)a3 ⇒ a3 = f 000 (a)
(2)(3)
Spotting the pattern, we see that the general formula for the coefficient an will be
1 (n)
an = 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:
(x − a)2 00 (x − a)3 000
f (x) = f (a) + (x − a)f 0 (a) + f (a) + f (a) + · · · (2.4)
2! 3!
n
The series carries on to infinity, and has general term (x−a)
n!
f (n) (a).
Taylor’s 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.

7
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.

x2 00 x3
f (x) = f (0) + xf 0 (0) + f (0) + f 000 (0) + · · · (2.5)
2! 3!
or, in summation notation
∞
xn
f (n) (0)
X
f (x) =
n=0 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

x2 x3
ex = 1 + x + + + ··· (2.6)
2! 3!
∞
X xn
or, in summation notation, ex =
n=0 n!

2.4 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
e5x = 1 + 5x + + + ···
2! 3!
25x2 125x3
= 1 + 5x + + + ···
2 6

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.

8
 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, cy-
cle through the values 1, 0, −1, 0, 1, 0, −1, 0, . . .. Putting these values into the gen-
eral 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 =
n=0 (2n)!
Similarly, it can be shown that the Maclaurin expansion of sin x is

x3 x5
sin x = x − + − ···
3! 5!

2.6 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

(x − 2)2 00 (x − 2)3 000

f (x) = f (2) + (x − 2)f 0 (2) + 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 con-
stitute 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.

9
2.7 Example
Find the first three non-zero terms of the Maclaurin series of e−2x sin x.
Solution. One way to do this would be to write down the Maclaurin series for
e−2x (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) = e−2x sin x. Then f (0) = 0.
f 0 (x) = e−2x cos x − 2e−2x sin x so f 0 (0) = 1. Differentiating again

f 00 (x) = e−2x (− sin x) − 2e−2x cos x − 2(e−2x cos x − 2e−2x sin x)

= 3e−2x sin x − 4e−2x cos x

and
f 000 (x) = 3(e−2x cos x − 2e−2x sin x) − 4(−e−2x sin x − 2e−2x 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) = e−2x sin x:

11x3
e−2x sin x = x − 2x2 + + ···
6
which will constitute a good approximation to e−2x 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 expan-
sion for sin−1 x.
Solution. First we find the expansion of (1 + x)−1/2 .

−1/2 1 (− 12 )(− 23 ) 2 (− 21 )(− 23 )(− 25 ) 3

(1 + x) = 1 + (− )x + x + x + ···
2 2! 3!
1 3 5
= 1 − x + x2 − x3 + · · ·
2 8 16
In the above, we now replace every x by −x2 to deduce that
1 3 5
(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
Z x
−1 dt
sin x = √
0 1 − t2

10
 Z x
= (1 − t2 )−1/2 dt
0
Z x
1 3 5

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

3 Applications to working out limits

The notation
lim f (x)
x→a

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

x→2 x→0

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, let’s work out
x2 − 4
lim
x→2 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

x2 − 4 (x − 2)(x + 2)
lim = lim = lim (x + 2) = 4
x→2 x − 2 x→2 x−2 x→2

Similarly, let’s work out

x2 + x − 2
lim
x→1 x2 − x
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 = lim = 3.
x→1 x2 − x x→1 x(x − 1) x→1 x

It is not always possible to work out limits simply by looking for factors and sim-
plifying 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
Let’s work out
(1 + x/2)5/7 − 1
lim
x→0 x
Again, we cannot put x = 0 into this expression as it stands. But we can use binomial
expansion, as follows;
 
( 75 )(− 27 ) x 2
(1 + x/2)5/7
−1 1 + ( 75 )( x2 ) + 2!
(2) + ··· − 1
=
x x
5 5
14
x − 196
x2 + ···
=
x
5 5
= − x + ···
14 196
We can let x → 0 in the above expression to deduce that
(1 + x/2)5/7 − 1 5
lim =
x→0 x 14

3.3 Example
Let’s work out
sin x sin 2x
lim and lim
x→0 x x→0 x
Solution. We mentioned earlier that
x3 x5 x7
sin x = x − + − + ···
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
x→0 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
x→0 x
It is in fact a general result that limx→0 sinxkx = k for any constant k.

12
3.4 Example
Find
sin2 x − x2 cos x
lim
x→0 x4
Solution. Recall that
x3 x5 x7
sin x = x − + − + ···
3! 5! 7!
and
x2 x4 x6
cos x = 1 − + − + ···
2! 4! 6!
Squaring the formula for sin x gives
x3 x5 x3 x5
! !
2
sin x = 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
   
x4 x2 x4
sin2 x − x2 cos x x2 − 3
+ (something) x6 + · · · − x2 1 − 2
+ 24
+ ···
=
x4 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
sin2 x − x2 cos x 1
lim 4
=
x→0 x 6

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

(−y)2
( ) !
1
= lim 1 + (−y) + + ··· − 1
y→0 y 2!
y
 
= lim −1 + + · · ·
y→0 2!
= −1
where we have used the Maclaurin expansion for the exponential, given by (2.6).

13
4 L’Hopital’s rule
0
Another way of working out a limit when in a 0
situation is the following result:

f (x) f 0 (x)
if f (a) = 0 and g(a) = 0 then lim = lim 0
x→a g(x) x→a g (x)

The above result is called L’Hopital’s 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 0
lim would be 0
if we put x = 0 in, so use L’Hopital
x→0 x
3 − cos x 0
= lim no longer 0
x→0 1
= 3 − cos 0
1
=2

4.2 Example

1 − cos x 0
lim would be 0
if we put x = 0 in, so use L’Hopital
x→0 x + x2
sin x 0
= lim no longer 0
x→0 1 + 2x

= 01
=0

4.3 Example

x−2 1 1
lim 2
= lim =
x→2 x −4 x→2 2x 4
Sometimes we have to apply L’Hopital’s rule more than once to get an answer, as the
next example illustrates:

14
4.4 Example

x − sin x 0
lim 0
so use L’Hopital
x→0 x3
1 − cos x 0
= lim still 0
so use L’Hopital again
x→0 3x2
sin x 0
= lim still 0
so use L’Hopital again
x→0 6x
cos x 0
= lim no longer 0
x→0 6
=61

4.5 Example

ln cos x 0
lim 0
so use L’Hopital
x→0 ln cos 3x
 
sin x
− cos x
= lim   now simplify this
x→0 − 3cos
sin 3x
3x
tan x 0
= lim still 0
so use L’Hopital again
x→0 3 tan 3x
sec2 x 0
= lim no longer 0
x→0 9 sec2 3x

= 19

5 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
(
−1, −2 < t < 0,
f (t) =
1, 0 < t < 2,

with f (t + 4) = f (t).
An example of a sawtooth function of period 2π would be the periodic function
of period 2π such that f (t) = t for t ∈ (−π, π). Since this function has period 2π 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 2π. Such
an expansion does indeed exist and in fact any periodic function of period 2π has an
expansion in terms of these trigonometric functions.
If the period is T rather than 2π 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 2nπt
T
and sin 2nπt
T
for n = 1, 2, 3 . . ., rather
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
∞ 
2nπt 2nπt
X 
1
f (t) = a
2 0
+ an cos + bn sin (5.7)
n=1 T0 T0

where
2 Z T0 2nπt
an = f (t) cos dt, n = 0, 1, 2, 3, . . . (5.8)
T0 0 T0
2 Z T0 2nπt
bn = f (t) sin dt, n = 1, 2, 3, . . . (5.9)
T0 0 T0
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 2nπt
an = f (t) cos dt, n = 0, 1, 2, 3, . . . (5.10)
T0 −T0 /2 T0
2 Z T0 /2 2nπt
bn = f (t) sin dt, n = 1, 2, 3, . . . (5.11)
T0 −T0 /2 T0
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 ω0 defined by
2π 1
ω0 = rather than T0
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
2
(f (a+) + 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 (
−1 −π < t < 0
f (t) =
1 0<t<π
with f (t + 2π) = f (t). Find the Fourier series of f (t).
Solution. In this case the period T0 is given by T0 = 2π. Let us find an first. The
following formulae will be useful
cos nπ = (−1)n and sin nπ = 0, n = 0, ±1, ±2, ±3, . . .
We have
2 Z T0 /2 2nπt
an = f (t) cos dt
T0 −T0 /2 T0
1Zπ
= f (t) cos nt dt
π −π
1Z0 1Zπ
= (−1) cos nt dt + cos nt dt
π −π π 0
1 − sin nt 0 1 sin nt π
   
= +
π n −π π n 0
an = 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 let’s find bn . we have
1Zπ
bn = f (t) sin nt dt
π −π
Z 0 Z π
1

= (−1) sin nt dt + sin nt dt
π " −π 0
 #
1 cos nt 0 − cos nt π
  
= +
π n −π n 0
1
= (1 − cos(−nπ) + (− cos nπ + 1))
πn
1
= (2 − 2(−1)n )
πn
and so
2
bn = (1 − (−1)n )
πn
With all the an , n = 0, 1, 2, . . ., equal to zero, and also recalling that the period
T0 = 2π, the Fourier series becomes
∞
X
f (t) = bn sin nt
n=1
= b1 sin t + b2 sin 2t + b3 sin 3t + · · ·

17
i.e.
4 4 4
f (t) = sin t + sin 3t + sin 5t + · · ·
π 3π 5π

6 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 Z a
f (t) dt = 2 f (t) dt
−a 0

6.2 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
−a

6.3 Useful rules of even and odd functions

even × even = even

even × odd = odd
odd × even = odd
odd × odd = even

6.4 Fourier series of an even function

Suppose that f (t) is an even function and we want it’s 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 2nπt
bn = f (t) sin dt
T0 −T0 /2 | {z } T0
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:

2 Z T0 /2 2nπt
an = f (t) cos dt
T0 −T0 /2 | {z } T0
even | {z }
even
4 Z T0 /2 2nπt
= 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

2 Z T0 /2 2nπt
an = f (t) cos dt
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, cal-
culation 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 2nπt
bn = f (t) sin dt
T0 −T0 /2 T0
2Z 2 2nπt
= f (t) sin dt
4 −2 4
1Z 2 nπt
= t sin dt odd × odd = even
2 −2 |{z}
| {z 2 }
odd
odd
Z 2
nπt
= t sin dt
0 2
#2 Z
−t cos nπt 2 cos nπt
"
2 2
= + dt
nπ/2 0 0 nπ/2
#2
2 sin nπt
"
4 2
= − cos nπ +
nπ nπ nπ/2 0

so that
4
(−1)n
bn = −
nπ
Recalling that T0 = 4, the Fourier series is
∞ 
4 nπt

− (−1)n sin
X
f (t) =
n=1 nπ 2
∞ n
4 X (−1) nπt
= − sin
π n=1 n 2

or, in expanded form,

4 πt 1 1 3πt 1
 
f (t) = − − sin + 2 sin πt − sin + sin 2πt + · · ·
π 2 3 2 4

6.7 Example
Find the Fourier series of the function such that

f (t) = t2 + t for −π <t<π

with f (t + 2π) = f (t) for all t.

Solution. This function is neither even nor odd. It is 2π-periodic so T0 = 2π.
Using this value for T0 the formula for an becomes
1Zπ
an = f (t) cos nt dt
π −π
1Zπ 2
= (t + t) cos nt dt
π "−π
sin nt π Z π #
1 sin nt
 
2
= (t + t) − (2t + 1) dt if n 6= 0
π n −π −π n

20
 1 Zπ
= − (2t + 1) sin nt dt
πn "−π #
1 cos nt π Z π
2 cos nt
 
= − −(2t + 1) + dt
πn n −π −π n
(2π + 1)(−1)n (2(−π) + 1)(−1)n 2 sin nt π
"  #
1

= − − + +
πn n n n n −π
n
" #
1 4π(−1) 4
= − − = 2 (−1)n
πn n n

We have shown that

4
(−1)n if n 6= 0
an =
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
#π
1 t3 t2
"
1Zπ 2
a0 = (t + t) dt = +
π −π π 3 2 −π
2π 2
a0 =
3
Next we find bn . We have
Z π
1
bn = (t2 + t) sin nt dt
π −π
" #
1 cos nt π Z π
cos nt

2
= −(t + t) + (2t + 1) dt
π n −π −π n
(−1)n (−1)n 1 Z π
" #
1 2 2
= −(π + π) + (π − π) + (2t + 1) cos nt dt
π n n n −π
(−1)n 1 sin nt π
" " Z π ##
1 2 sin nt

= −2π + (2t + 1) − dt
π n n n −π −π n
(−1)n
" #
1 2 Zπ
= −2π − 2 2 sin nt dt
π n n −π

so that
2
bn = − (−1)n
n
Putting the formulae for a0 , an and bn into the general Fourier series expansion, which
when T0 = 2π is
∞
X
f (t) = 12 a0 + (an cos nt + bn sin nt)
n=1

gives
∞
π2 X 4 2
 
f (t) = + 2
(−1)n cos nt − (−1)n sin nt
3 n=1 n n

21
7 Complex form of a Fourier series
Recall that √
j= −1
Recall also, Euler’s formula
ejt = cos t + j sin t (7.12)
and the other useful version of it:
e−jt = 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!
t2 t4 t3 t5
! !
= 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 + e−jt
cos t = (7.14)
2
Similarly, subtracting (7.13) from (7.12) gives
ejt − e−jt
sin t = (7.15)
2j
Since cos t and sin t are periodic of period 2π, so is ejt . Furthermore, so are the
functions enjt for any integer n (positive or negative). This leads us to suppose
that any 2π 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
∞
cn e2jnπt/T0
X
f (t) = (7.16)
n=−∞

where
1 Z T0 /2
cn = f (t)e−2jnπt/T0 dt for n = 0, ±1, ±2, ±3, . . . (7.17)
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


 0 −π < t < −π/2
f (t) = 1 −π/2 < t < π/2


0 π/2 < t < π

with f (t + 2π) = f (t) for all t.

Solution. The period is 2π so T0 = 2π. With this value of T0 expressions (7.16)
and (7.17) reduce to
∞
cn ejnt
X
f (t) =
n=−∞

where
1 Zπ
cn = f (t)e−jnt dt
2π −π
#π/2
1 e−jnt
"
1 Z π/2 −jnt
= e dt =
2π −π/2 2π −jn −π/2
1 h −jnπ/2 i
= e − ejnπ/2
−2πjn
1 ejnπ/2 − e−jnπ/2
" #
=
πn 2j
jx −jx
Recalling the formula sin x = e −e 2j
, the above formula for cn can be put into the
form
1 nπ
cn = sin for n = ±1, ±2, ±3, . . .
πn 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 Zπ 1 Z π/2 1
c0 = f (t) dt = dt =
2π −π 2π −π/2 2
So the complex Fourier series of the function is
∞
cn ejnt
X
f (t) =
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
−1 ∞
cn ejnt + cn ejnt
X X
1
f (t) = 2
+
n=−∞ n=1

23
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
∞ ∞
c−m e−jmt + cn ejnt
X X
1
f (t) = 2
+
m=1 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
∞ ∞
−jnt
cn ejnt
X X
1
f (t) = 2
+ c−n e + insert expression for cn
n=1 n=1
∞ 
1 −nπ −jnt 1 nπ jnt
X 
1
= 2
+ sin e + sin e
n=1 −πn 2 πn 2
∞
1 nπ  −jnt 
+ ejnt
X
1
= 2
+ sin e
n=1 πn 2 | {z }
=2 cos nt

and so ∞
1
X 2 nπ
f (t) = 2
+ sin cos nt
n=1 πn 2
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
∞
1
X 2 (2n − 1)π
f (t) = 2
+ sin cos(2n − 1)t
n=1 π(2n − 1) 2
∞
1
X 2(−1)n+1
= 2
+ cos(2n − 1)t
n=1 π(2n − 1)

which is more computationally efficient.

8 Amplitude and phase spectrum

Recall that the Fourier series of a periodic function of period T0 (and frequency
ω0 = 2π/T0 ) is
∞ 
2nπt 2nπt
X 
1
f (t) = 2 a0 + an cos + bn sin
n=1 T0 T0
with certain formulae, namely (5.8) and (5.9), for the coefficients an and bn .
We can write the Fourier series in terms of ω0 as
∞
X
1
f (t) = a
2 0
+ (an cos nω0 t + bn sin nω0 t)
n=1
X∞
1
= a
2 0
+ An cos(nω0 t + δn )
n=1

24
where An is called the amplitude and δn the phase. To find formulae for the numbers
An and δn , n = 1, 2, 3, . . ., we use elementary trigonometry. Now

An cos(nω0 t + δn ) = An cos nω0 t cos δn − An sin nω0 t sin δn

Comparing the right hand side of the above expression with

an cos nω0 t + bn sin nω0 t

we see that we need

An cos δn = an and An sin δn = −bn

These expressions imply that q

An = a2n + b2n
and
bn
tan δn = −
an
The numbers δn , 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 num-
bers together form the spectrum and (with the frequency ω0 ) constitute one way of
describing a periodic function.

25