1 The Complex Fourier Series

Let f ( x) be a periodic (real-valued) function with period 2L, which satisfies Direchlets conditions on
the interval [L, L]. Then it can be represented by the Fourier series
a0 X h n x n x i
f ( x) = + a n cos + b n sin (1.1)
2 n=1 L L

where
ZL
1
a0 = f ( x) dx, (1.2)
L
L
ZL n x
1
an = f ( x) cos , n = 1, 2, . . . (1.3)
L L
L
ZL n x
1
bn = f ( x) sin n = 1, 2, . . . . (1.4)
L L
L
From the Eulers identity
e ix = cos x + i sin x and e ix = cos x i sin x,
we can write
e ix + e ix e ix e ix
cos x = and sin x =
2 2i
Substituting these into (1.1) and then grouping the terms, we get
a0 X a ib
a n + ib n in x/L

e in x/L +
n n
f ( x) = + e .
2 n=1 2 2

Setting
a n ib n a n + ib n
c 0 = a 0 /2 and , cn =
cn = , n = 1, 2, . . .
2 2
in this, we finally obtain the the complex or exponential form of the Fourier series of f ( x) as

c n e in x/L ,
X
f ( x) = (L x L), (1.5)
n=
where the (complex) Fourier coefficient c n is given by
ZL
1
cn = f ( x) e inx/L dx, n = 0, 1, 2, . . . (1.6)
2L
L

Remark 1.1. At the points x of discontinuity, the Fourier series (1.5) converges to the average
f ( x 0) + f ( x + 0)
f ave = of the one-sided limits of f at x.
2

Remark 1.2. If the argument of f is a time variable t, then n = n/L are the frequencies.

Example 1.1. Consider the box wave with period 2L = 2

0<x<

1
f ( x) =
1 < x < 0
with f ( x + 2) = f ( x).
Then
Z Z

1 1 0
Z
inx inx inx
cn = f ( x) e dx = (1).e dx + (1) e dx
2 2
0

2
 e inx
" 0 #
1 e inx
= +

2 in

in
x=

x=0
1 h in

in
i
= 1 e e 1
2 ni
1 cos( n) 1 (1)n
= =
in in
i P n 1(1)n o inx
Therefore, f ( x) = n e .
n=
Example 1.2. Consider f ( x) = e x for < x < and f ( x + 2) = f ( x).
Then
Z
1 1 (1 in) x
Z
cn = e inx e x dx = e dx
2 2

1 e(1 in) x e e in e e in
= =
2 1 in 2(1 in)

x=
(1)n e e sinh (1)n
= =
(1 in) 2 1 in
Thus
(1) n
sinh X
f ( x) = e inx , x
n= 1 in
At x = 0 this gives the amusing formula
!
sinh X (1) n (1) n
X (1) n
sinh X
1= 1+ + =
n=1 1 in n=1 1 + in n=2 1 + n2

Parsevals identity for the Complex Fourier series

Z
1
| c n |2 = | f ( x)|2 dx.
X
(1.7)
n= 2

2 The Fourier Integral

Theorem 2.1 (Fourier Integral Theorem). Let f ( x) satisfy Dirichlet conditions, f be both inte-
Z Z
grable and absolutely integrable over the interval < x < so that f ( x) dx and | f ( x)| dx exist.

Then

Z
1
f ( x) = [ A cos( x) + B sin( x)] d , (2.1)

0
where the Fourier coefficients of f are
Z
1
A = f ( u) cos( u) du, (2.2)


Z
1
B = f ( u) sin( u) du. (2.3)

Example 2.1. Find the Fourier integral of the rectangular pulse function

1 1 < x < 1
f ( x) =
0 |x | > 1

3
 Z Z
sin cos x sin

and hence evaluate (a) d (b) d .

0 0
Solution. We have
Z Z1
1 sin( u) 1 2 sin()

1 1
A = f ( u) cos( u) du = cos( u) du = du = ,

x=1
1

and
Z Z1
1 1
B = f ( u) sin( u) du = sin( u) du = 0.

1
Therefore
Z Z Z
1 2 sin 2 2 sin cos( x)

f ( x) = [ A cos( x) + B sin( x)] d = cos( x) d = d .

0 0 0

Now f ( x) = 1 if x < 1 and f ( x) = 0 for x > 1, while the average of the left and right hand limits of f ( x)
1+0 1
at x = 1 is = . Therefore
2 2

/2 if x 1

Z
2 sin cos( x)

d = /4 if x = 1

0 if x > 1.

0

Z
sin

Taking x = 0 in this we get d =
2
0

3 Fourier Cosine and Sine Integrals

Suppose f is defined on the half-range [0, ) and 0 | f ( x)| dx converges. Then

R

Fourier cosine integral

Z
2
f ( x) = A cos( x) d

0
where
Z
2
A = f ( u) cos( u) du

0
Fourier sine integral
Z
2
f ( x) = B sin( x) d

0
where
Z
2
B = f ( u) sin( u) du.

0

4
Example 3.1. Find the Fourier cosine and sine integral representations of f ( x) = ekx , x > 0, k > 0.
Solution. We have
Z Z
2 2
A = f ( u) cos( u) du = eku cos( u) du

0 0
2 eku 2 k
= ( k cos u + sin u) =

k + 2 k 2 + 2
2

x=0
so that the Fourier cosine integral of f is
Z Z
2k 2k cos( u)
f ( x) = ekx = A cos( x) d = d
k 2 + 2
0 0

From this we see that

cos( u)
Z
d = ekx , x > 0, k > 0 (3.1)
0 k 2 + 2 2k
Now
Z Z
2 2
B = f ( u) sin( u) du = eku sin( u) du

0 0
2 eku 2
= ( k sin u cos u) =

k + 2 k 2 + 2
2

x=0
Therefore the Fourier sine integral of f is
Z Z
2 2 sin( u)

f ( x) = ekx = B sin( x) d = d .
2
k + 2
0 0

Here also we see that

Z
sin( u)

d = ekx , x > 0, k > 0 (3.2)
k 2 + 2 2
0
The integrals defined by (3.1) and (3.2) are called Laplaces Integrals.

Exercise 3.1. Find the Fourier sine integral of

f ( x) = if 0 < x < 1
2
= 0 if x > 1
Z
1 cos

Ans. f ( x) = sin x d

0

4 The Complex Fourier Transform

The Fourier Transform gives us a unique way of viewing any function as the sum of simple
sinusoids
The Fourier Transform is the extension of this idea of Fourier series representation to nonpe-
riodic functions

Definition 4.1. Let f ( x) be defined for all real < x < . Then its Fourier transform is given by
Z
1
F f ( x) = F () = p f ( x) e i x dx

(4.1)
2

5
Let f ( x) represent a signal. Then
its Fourier transform F () is known as complex frequency spectrum,
the graph of the magnitude |F ()| is called its amplitude spectrum,
the graph of the argument arg F () is called its phase spectrum.
Fourier Inversion formula:
Z
1
f ( x) = F 1 F () = p F ( ) e i x d .

(4.2)
2

We say that f ( x), F () is a Fourier pair.

5 Properties of the Fourier Transform

Let F () and G () be the Fourier transforms of f ( x) and g( x) respectively.

Property 5.1 (Linearity). F a f ( x) + b g( x) = aF () + bG ()

Property 5.2 (Duality). f ( x), F () is a Fourier pair if and only if F ( x), f () is a Fourier pair.

Definition 5.1. f ( x) is self-reciprocal if F () = f ().

1
Property 5.3 (Change of scale). F f (ax) =

F , a 6= 0
| a| a

Z
1
I = F f (ax) = p f (ax) e i x dx.

Proof. Let
2

Case (a): If a > 0, write ax = u so that u ranges from to as = x ranges from to . Therefore
Z
1 1 1
I= p f ( u) e i(/a)u dx = F
a 2 a a

Case (b): If a < 0 so that a > 0. Then write ax = v so that v ranges from to as = x ranges
from to . Therefore
Z
1
Z
dv

1 1 1
I= p f (v) e i(/a)v = p f ( z) e i(/a) z dz = F
2 a a 2 a a

Property 5.4 (Spatial-Shifting). F f ( x a) = e ia F ()

Proof. By the definition,

Z
1
F ( f )( x a) = p f ( x a) e i x dx

2

Z
1
= e i a p f ( x a ) e i ( x a ) d ( x a )
2

Z
1
= e i a p f ( u) e iu du = e ia F ()
2

Thus the effect of spatial shifting by a units to the right is to multiply the transform with e ia .

6
Property 5.5 (Frequency-Shifting). F e iax f ( x) = F ( + a)

Proof. By the definition,

Z Z
1 1
F e iax f ( x) = p [ e iax f ( x)] e i x dx = p f ( x) e i(+a) x dx = F ( + a)

2 2

Thus the effect of multiplying the spatial signal with e iax is to shift the frequency signal by a units
to the left.

Property 5.6 (Multiplication by x n ). F x n f ( x) = ( i )n F (n) (), n = 1, 2, ...

Z
1
Definition 5.2. Fourier Convolution] ( f g)( x) = p f ( u) g( x u) du, < x <
2

Property 5.7 (Convolution Theorem). F ( f g)( x) = F () G ()

Proof. By definition, we have

Z
1
F ( f g)( x) = p ( f g)( x) e i x dx

2

Z Z

1 1
e i x dx

=p p f ( u) g( x u) du
2 2

Z 1 Z

1 iu i ( x u )

=p f ( u) e p g( x u) e dx du
2 2

1 Z 1 Z

iu i ( x u )

= p f ( u) e du p g( x u) e dx
2 2

1 Z 1 Z

f ( u) e iu du g(v) e iv dv

= p p
2 2

= F () G ().
That is, the Fourier transform of the convolution of two functions equals the product of the respective
Fourier transforms.

1
Property 5.8 (Modulation). F f ( x) cos ax = [F ( a) + F ( + a)]

2

Proof. By the definition,

Z
1
F f ( x) cos ax = p f ( x) cos ax e i x dx

2

Z !
1 e iax + e iax
=p f ( x) e i x dx
2 2

Z Z

1 1 1
= p f ( x) e i(a) x dx + p f ( x) e i(+a) x dx
2 2 2

1
= [F ( a) + F ( + a)]
2
Modulation property is used in problems where a harmonic wave is modulated by a carrier wave.

7
Property 5.9 (Derivative). Suppose that
(a) f , f 0 , f 00 , ..., f (n1) all tend to 0 as | x| , and
Z Z
( k)
(b) | f ( x)| dx < , f ( x) dx < , k = 1, 2, ..., n.

Then F f (n) ( x) = ( i )n F (), n = 1, 2, ...

x
Z F ()
Property 5.10 (Integral). F f ( u) du = , < x <
i

The transform of the derivatives, multiplication by x n , integral property and convolution are useful
in solving differential equations.
Example 5.1 (Rectangular Pulse). Find the Fourier transform of
(
k, 0 < x < 1,
f ( x) =
0, elsewhere,
(
e iax , 0 < x < 1,
and hence of g( x) =
0, elsewhere.
Solution.
Z1 1
k e i x

1 ik
ke i x dx = p 1 e i .

F f ( x) = p

= p
2 2 i 2


0 x=0

i
1 e i .

With k = 1, this gives F f ( x) = p

2

Then by the frequency-shifting property, we get

i 1 e i ( + a )

iax
F g ( x) = F e f ( x) = F ( + a) =

p
( + a) 2
(
x, | x| < a,
Example 5.2 (Saw-tooth Signal). Find the Fourier transform of f ( x) =
0, elsewhere.
Solution. By defn.,
Za
1
F () = p x e i x dx
2
a
)a
e i x e i x
( ) (
1
= p ( x ) (1)
2 i ( i )2 x=a

a
1 ix 1

e i x

= p +
2 2
x=a
1 ia 1 ia 1

i a
=p + e + e i a
2 2 2
1 ia ia 1

i a i a
=p e +e e e i a
2 2

1 ia 2 cos a i 2 sin a

=p
2 2
s
2 a cos a sin a

= i , 6= 0
2

8
Alternately,
Za
1
F () = p x e i x dx
2
a
Za
1
=p x [cos x i sin x] dx
2
a

Za odd function Za even function

1 z }| { i z }| {
=p x cos x dx p x sin x dx
2 2
a a
| {z } | {z }
=0 =2 0a x sin x dx
R

s
2 n cos x o sin x a

= i ( x ) (1)
2
x=0
s
2 a cos a sin a

= i , 6= 0
2

Example 5.3 (Square Wave). Find the Fourier transform of

(
1, | x| < 1,
f ( x) =
0, elsewhere.

Hence
Z
sin t
(a) derive that dt = ;
t 2
0
sin x
(b) using the duality property, derive the Fourier transform of g( x) =
x
Solution.
Z1
1
F ( ) = p e i x dx
2
1
Z1
1
=p [cos x i sin x] dx
2
1

Z1 z even Z1 z odd
1 }| { i }| {
=p cos x dx p sin x dx
2 2
1 1
| {z } | {z }
=2 01 cos x dx
R =0

Z1
s s s
2 sin x 1 2 sin

2
= cos x dx = = , 6= 0
x=0
0

(a) By the inversion formula,

Z
1
f ( x) = F 1 F () = p F () e i x d

2

Z
s
1 2 sin i x

p e d = f ( x) for all < x < .
2

9
 In particular, for x = 0, this gives
Z
s
1 2 sin

p d = f (0) = 1
2

or
Z Z
sin sin

d = d =
2
0
Z
sin t

Replacing with a dummy variable t, this gives dt =
t 2
0
(b) Now by the duality property, F ( x), f () is a Fourier pair. That is
F F ( x) = f ()

r r
2 sin x 2
But F ( x) = = sin cx. Thus
x
(s ) (
2 sin x 1, | | < 1,
F = f () =
x 0, elsewhere
or

r
sin x || < 1,

,
F = 2
x
0, elsewhere.

Example 5.4 (Parabolic Signal). Find the Fourier transform of

(
1 x2 , | x| < 1,
f ( x) =
0, elsewhere.
Z
sin x x cos x x 3

Hence deduce that cos dx =
x3 2 16
0
Solution. By defn.,
Z1
1
F () = p (1 x2 ) e i x dx
2
1
Z1
1
=p (1 x2 )[cos x i sin x] dx
2
1
even function odd function
Z1 z Z1 z
1
}| {
1
}| {
=p (1 x2 ) cos x dx i p (1 x2 ) sin x dx
2 2
1 1
| {z } | {z }
=2 01 (1 x2 ) cos x dx
R =0
s
sin x cos x sin x 1

2

= (1 x2 ) (2 x) + (2)
2 3
x=0
s
2 2 cos 2 sin

= 0 + {0 0 0}
2 3
s
2 cos sin

= 2 , 6= 0
2 3

10
By the inversion formula,
Z
1
f ( x) = F 1 F () = p F () e i x d

2

Z
1 4 cos sin

p p e i x d = f ( x)
2 2 2 3

Z
cos sin

e i x d = f ( x)
2 3 2

Writing x = 12 in this, we see that

Z
cos sin 1 3 3

+ e i/2 d = f = =
2 3 2 2 24 8

Z
cos sin 3

[cos(/2) i sin(/2)] d =
2 3 8

Since the imaginary part on the left hand side is odd function of , we get
Z Z
cos sin 3 x cos x sin x x 3

cos(/2) d = or cos dx =
2 3 16 x 3 2 16
0 0
Multiplying this with 1 both sides, we get the required result.
(
1 | x |2 , | x| < 1,
Exercise 5.1. Find the Fourier transform of f ( x) =
0, elsewhere.
(
1 | x |, | x| < 1,
Exercise 5.2 (Triangule Pulse). Find the Fourier transform of f ( x) =
0, elsewhere.

2 1 cos
r
Ans. F () = , 6= 0
2
Example 5.5 (Negative Exponential Signal). Find the Fourier transform of
(
eax , x 0,
f ( x) =
0, elsewhere.
Hence find the amplitude and phase spectra of f ( x).
Solution. We have
Z Z
1 e(a+ i) x

1 1
F f ( x) = F () = p eax e i x dx = p e(a+ i) x dx = p

2 2 2 a + i

0 0 x=0
1 1 1 a

=p =p i
2 a + i 2 a2 + 2 a 2 + 2
Aliter:

Z Z
1 1
F f ( x ) = F ( ) = p eax e i x dx = p eax [cos x i sin x] dx

2 2
0 0

Z

1
Z
ax ax
=p e cos x dx i e sin x dx
2
0 0
1 a

=p i
2 a2 + 2 a 2 + 2

11
 1 1
The amplitude spectrum of f ( x) is |F ()| = p p , and the phase spectrum of f ( x) is
2 a 2 + 2

arg F () = tan1
a
Example 5.6 (Decaying Exponential Signal). Find the Fourier transform of f ( x) = e| x| , < x < .
Hence
Z
cos x
(a) derive that d = e| x| , and
1 + 2 2
0

(b) obtain the Fourier transform of xe| x| .

Solution.
Z Z
1 1
F () = p e| x| e i x dx = p e| x| (cos x i sin x) dx
2 2

even function odd function
Z z }| { Z z }| {
1 | x| 1
=p e cos x dx i p e| x| sin x dx
2 2

| {z }
=0
Z
s s
2 e x

2 2 1
e x cos x dx = ( cos x + sin x)

=p =
2 1 + 2 x=0 1 + 2
0

(a) By the inversion formula,

Z
1
f ( x) = F 1 F () = p F () e i x d

2

Z " Z i x
s #
1 2 1 1 e
=p e i x d = d
2 1 + 2 1 + 2

Z
1 cos x + i sin x
= d
1 + 2

even odd
Z z }| { Z z }| { Z
1 cos x 1 sin x 2 cos x

= d + i d = d
1 + 2 1 + 2 1 + 2
0
| {z }
=0

Z
cos x d
or = e| x| , < x < .
1 + 2 2
0
(b) By the multiplication by x property, we have
s ! s
0 d 2 1 2 2 i
F x f ( x) = ( i )F () = i

=
d 1 + 2 (1 + 2 )2

Example 5.7. Find the Fourier transform of

| x| <
(
cos x, 2,
f ( x) =
0, elsewhere.

Z(
cos 2x
)

Hence deduce that dx =
1 x2 2
0

12
Solution. By defn.,

Z2 Z2
1 i x 1
F ( ) = p cos x e dx = p cos x[cos x i sin x] dx
2 2

2
2

Z2 even function Z2 zodd function
1 z }| { 1 }| {
=p cos x cos x dx i p cos x sin x dx
2
2
2 2
| {z } | {z }
=2 /2/2 cos x cos x dx =0
R

Z2
1
=p 2 cos x cos x dx
2
0

Z2 Z2
1
= p cos( + 1) x dx + cos( 1) x dx

2
0 0

1 sin( + 1) x sin( 1) x /2

= p +
2 +1 1
x=0
(+1) (1)

1 sin 2 sin 2
=p +
2 +1 1
( ) s
1 cos 2 cos 2 2 cos 2
=p + = , 6= 1
2 +1 1 1 2

For = 1, we have

Z2 Z2
s s
2 1 1
r
1 2
F ( ) = p 2 cos x cos( x) dx = cos2 x dx = =
2 2 2 2 2
0 0

By the inversion formula,

Z
1
f ( x) = F 1 F () = p F () e i x d

2

Z (
cos
s )
1 2 2
p d = f (0)
2 1 2

Z (
cos
)
2
d = cos 0
1 2

Z(
cos
)
2
d =
1 2 2
0

Z cos

2
Replacing with dummy variable , we see that d =
1 2 2
0

13
Example 5.8 (HOT). Given 0 < a < 1, find the Fourier tranform of f ( x) = | x|a . Hence for a = 12 ,
p
show that f is self-reciprocal using the value (1/2) = .
Solution. For > 0, we have
Z Z
1 1
F () = p | x|a e i x dx = p | x|a (cos x i sin x) dx
2 2

even function odd function
Z z Z z Z
1 1 2
}| { }| {
=p | x|a cos x dx i p | x|a sin x dx = p | x|a cos x dx
2 2 2
0
| {z }
=0
Z
s s
2 2
Z
a i x
= xa cos x dx = Real Part of x e dx

0 0

xa e i x dx. In fact, write i x = t so that x = it and dx = dt
R
We evaluate I = i . Therefore
0

Z Z a
t dt
xa e i x dx = e t
i i
0 0
Z
a1 i a
= ta e t dt
i
0
a1 h ia
= cos + i sin (1 a)
i 2 2
a1 (1 a) h a a i
= cos + i sin
i 2 2
h a a i
= a1 (1 a) sin i cos
2 2
h a a i
= a1 (1 a) sin + i cos
2 2
h a a i
a1
= (1 a) sin + i cos
2 2
Z a
The Real Part of xa e i x dx = a1 (1 a) sin .
2
0
Hence s
2 a1 a
F () = (1 a) sin
2
With a = 12 , this gives
s s
2 1/2 2 p 1
F x1/2 = (1/2) sin 1/2 p = 1/2 .

=
4 2

Thus x1/2 is self-reciprocal.

14
 2
Example 5.9 (Gaussian Signal). Consider f ( x) = e x /2 , < x < . Show that f is self-reciprocal,
2 2
and hence find the Fourier tranforms of g( x) = xe x /2 and h( x) = xe x /2 . Also find the convolution
f g.
Solution. By defn.,
Z Z
1 2 1 2
F () = p e x /2 e i x dx = p e( x +2 x.i)/2 dx
2 2

Z 2 Z
1 [( x+ i )2 ( i )2 ]/2 e /2 2
=p e dx = p e( x+ i) /2 dx
2 2

even function
Z z }| {
2 1 2
= e /2 p e z /2 dz
2

| {z }
2
=2 0 e z /2 dz
R

Z
s
2 /2 2 du
=e eu p , where u = z2 /2
2u
0
Z
2 1
= e /2 p u1/2 eu du

0
1 1 1 p

2 2 2
= e /2 p = e /2 p = e /2
2
2
Thus f ( x) = F (). Hence f ( x) = e x /2 is self-reciprocal.

By the multiplication by x property:

d n 2 /2 o 2
F g( x) = F x f ( x) = ( i ) F 0 () = i = i e /2 = G (), say.

e
d
Again applying the same property, we get
d n 2 o 2
F h( x) = F xg( x) = ( i ) G 0 () = i i e /2 = (1 2 ) e /2 .

d
Now by the convolution theorem
2 /2 2 o 2 i d n 2 o
F ( f g)( x) = F () G () = e ( i ) e /2 = i e =
n
e . (5.1)
2 d
Due to the change of scale property,
p p p p p
x

2 2
F f p = 2 F 2 = 2 e( 2) /2 = 2 e .
2
So by the multiplication by x property, we finally get
x d np 2 o

F xf p = i 2e
2 d
or
d n 2 o 1 x

e = p F xf p . (5.2)
d i 2 2
From (5.1) and (5.2), it follows that
x 2
( f g)( x) = p e x /4
2 2

15
 2 2 2
Example 5.10. Given that F e x /2 = e /2 , find the Fourier transform of eax , a > 0.

2
Solution. Let f ( x) = eax . Note that
2 2
p 2 p
eax = e2ax /2 = e( 2ax) /2 = f ( 2ax).

Therefore by the change of scale property,

p 1

F f ( 2ax) = p

F p
2a 2a
or 2
p

2 1 /2 1 2
F eax = p e 4a

= p e 2a
2a 2a
2 2 2
Exercise 5.3. Given that F e x /2 = e /2 , find the Fourier transform of e x /3 .

2 2 2 2
Exercise 5.4. Given that F e x /2 = e /2 , find the Fourier transforms of e4( x3) and e x cos 3 x.

6 Fourier Sine and Cosine Transforms

Let f ( x) be defined for all x > 0. Then

( a) The Fourier Sine transform of f is given by

s
2
Z
Fs f ( x) = F s () = f ( x) sin x dx


0
The Sine Inversion Formula of f is given by
s
2
Z
1
f ( x) = Fs F s () = F s () sin x d


0
( b) The Fourier Cosine transform of f is given by
s
2
Z
F c f ( x ) = F c ( ) = f ( x) cos x dx


0
The Cosine Inversion Formula of f is given by
s
2
Z
f ( x) = F c1 F c () = F c () cos x d


0

Example 6.1. Find the cosine transform of

(
1, 0xa
f ( x) =
0, elsewhere,

where a > 0.
Solution.
Z Za
s s s s
2 sin x a 2 sin a

2 2
F c ( ) = f ( x) cos x dx = 1 cos x dx = = , 6= 0
x=0
0 0

16
Example 6.2. Find the cosine transform of f ( x) = eax , a > 0, x > 0.
Solution.
Z Z
s s
2 2
F c () = f ( x) cos x dx = eax cos x dx

0 0
s s
2 eax

2 a
(a cos x + sin x)

= = ,
a 2 + 2 x=0 a 2 + 2

Example 6.3. Find the sine transform of f ( x) = eax , a > 0, x > 0 and hence show that
Z
x sin mx ea
dx =
1 + x2 2
0
Solution.
Z Z
s s
2 2
F s () = f ( x) sin x dx = eax sin x dx

0 0
s s
2 eax

2
(a sin x cos x)

= =
a 2 + 2 x=0 a 2 + 2
By the inversion formula,
Z Z(
s s s )
2 2 2
f ( x) = F s () sin x d = sin x d
a 2 + 2
0 0
or
Z
sin x d eax
= f ( x) =
a 2 + 2 2 2
0
Changing x to m, to x and a = 1 we get
Z
x sin mx dx em
=
1 + x2 2
0

Example 6.4. Find the sine transform of f ( x) = 1/ x, x > 0.

Solution.
Z
s s
sin x 2
r
2
F s () = dx = =
x 2 2
0

2
Example 6.5. Show that f ( x) = e x /2 , < x < is self-reciprocal under the cosine transform.
2
Solution. We recall that f ( x) = e x /2 is self-reciprocal under the Fourier transform. That is
2 /2 2
F e x = e /2 = F ()

Z Z
2 1 2 1 2
e /2 = p e x /2 e i x dx = p e x /2 (cos x i sin x) dx
2 2

Z
s
2 2
Z
2 2
=p e x /2 cos x dx = e x /2 cos x dx.
2
0 0
2 2 2
In other words, F c e x /2 = e /2 . Hence f ( x) = e x /2 is self-reciprocal.

17
Cosine transform from the Sine transform:
Theorem 6.1. If Fs f ( x) = F s () and f ( x) 0 as x , then

F c f 0 ( x) = F s ()

(6.1)
F c x f ( x) = F s0 ()

(6.2)

Sine transform from the Cosine transform:

Theorem 6.2. If F c f ( x) = F c () and f ( x) 0 as x , then

Fs f 0 ( x) = F c ()

(6.3)
Fs x f ( x) = F c0 ()

(6.4)

2
Example 6.6. Given that f ( x) = e x /2 is self-reciprocal under the cosine transform, find the sine
2 2
transform of g( x) = xe x /2 and the cosine transform of h( x) = x2 e x /2 .
2 2 2 2
Solution. Given that F c e x /2 = e /2 . Let f ( x) = e x /2 and F () = e /2 . Then

0
Fs f ( x) = F s ()
2
Fs xe x /2 = F s ()

2 2
Fs xe x /2 = F s () = e /2 = G s (), say.

2
In other words, g( x) = xe x /2 is self-reciprocal under the sine transform.

Now by the multiplication by x property, we have

2 d d n 2 /2 o 2
F c x( xe x /2 ) = {G s ()} = e = (1 2 ) e /2

d d
2
Thus F c h( x) = (1 2 ) e /2 .

eax
Example 6.7. Find the sine transform of f ( x) = and hence the cosine transform of g( x) = eax
x
r Z ax
2 e
Solution. Let I = F s () = sin x dx.
x
0
Differentiating w. r. t. under the integral sign, this gives
s ax s s
dI 2 e 2 2 a
Z Z
= ( x cos x) dx = eax cos x dx =
d x a 2 + 2
0 0
On one hand, this gives
Z
s
2 a
F c eax = eax cos x dx =

.
a 2 + 2
0
On the other hand, separating the variables in the differential equation
s
dI 2 a
=
d a 2 + 2
and then integrating, we get the general solution
s
2
I= tan1 + A.
a
Then writing = 0 in this so that
s s
2 2 eax
Z
tan1 0 + A = sin 0 dx = 0
x
0
r
2
or A = 0. Thus F s () = tan1
a

18
 1
Exercise 6.1. Find the sine transform of f ( x) = and hence deduce the cosine transform of
x( x2 + a2 )
1
g ( x) =
x2 + a2

r r
1 1
Ans. F s () = (1 ea ), G c () = e a
a2 2 a 2

1
Exercise 6.2. Find the cosine transform of f ( x) = and hence derive the sine transform of
x2 + 1
x
g ( x) =
x2 + 1

r
Ans. F c () = G s () = e
2
Example 6.8 (HOT). Given 0 < a < 1, find the sine and cosine tranforms of f ( x) = xa .

xa e i x dx. In fact, write i x = t so that x = it and dx = dt
R
Solution. First, we evaluate I = i .
0
Therefore
Z Z a
t dt
xa e i x dx = e t
i i
0 0
Z
a1 i a
= ta e t dt
i
0
a1 h ia
= cos + i sin (1 a)
i 2 2
a1 (1 a) h a a i
= cos + i sin
i 2 2
h a a i
= a1 (1 a) sin i cos
2 2
h a a i
= a1 (1 a) sin + i cos
2 2
h a a i
a1
= (1 a) sin + i cos
2 2
Therefore
Z
s s
2 2 a1 a
F c xa = The Real Part of xa e i x dx = (1 a) sin

2
0

and
Z
s s
2 2 a1 a
Fs xa = The Imaginary Part of xa e i x dx = (1 a) cos

2
0
With a = 21 we see that cos

p1 Hence

4 = sin 4 = 2
s s
2 1/2 2 p 1
F c x1/2 = (1/2) sin 1/2 p = 1/2

=
4 2
and
s s
2 1/2 2 p 1
Fs x1/2 = (1/2) cos 1/2 p = 1/2 .

=
4 2

Thus x1/2 is self-reciprocal under both sine and cosine transforms.

19
7 Parsevals Identities

Theorem 7.1 (Parsevals Identity for Complex Fourier Transform).

Z Z
( f ( x))2 dx = |F ()|2 d (7.1)

Proof. By definition, we note that
Z Z 1 Z

i x

2
( f ( x)) dx = f ( x) p F () e d dx
2

Z 1 Z

e i x dx d

= F () p f ( x)
2

Z 1 Z


= F () p f ( x) e i x dx d
2

Z 1 Z


= F () p f ( x) e i x dx d
2

Z Z
= F () F () d = |F ()|2 d

Theorem 7.2 (Parsevals Identity for Sine Transform).
Z Z Z Z
f ( x) g( x) dx = F s () G s () d and | f ( x)|2 dx = |F s ()|2 d
0 0 0 0
Theorem 7.3 (Parsevals Identity for Cosine Transform).
Z Z Z Z
f ( x) g( x) dx = F c () G c () d and | f ( x)|2 dx = |F c ()|2 d
0 0 0 0

Z
Remark 7.1. The integral | h()|2 d of the squared magnitude of a function h() is known as the

energy of the signal h. Thus the energy of the signal f ( x) is the same as the the energy contained in
its transform.

Z

r
2 a dx
Example 7.1. Given that F c eax =

, prove that =
2 + a 2 ( x2 + a2 )( x2 + b2 ) 2ab(a + b)
0

Solution. Let f ( x) = eax , g( x) = ebx . Then

s
2 a
F c f ( x) = = F c (),


2 + a 2
s
2 b
F c g ( x) = = G c ().


2 + b 2
Hence by Parsevals identity for cosine transform, we have
Z Z
F c () G c () d = f ( x) g( x) dx
0 0
Z Z
2 ab 1
d = eax ebx dx =
(2 + a2 )(2 + b2 ) a+b
0 0

20
 Z Z
d dx
so that = or =
(2 + a2 )(2 + b2 ) 2ab(a + b) ( x2 + a2 )( x2 + b2 ) 2ab(a + b)
0 0
Example 7.2. Given that
s s
2 a 2 sin a
F c eax = and F c g( x) =


2 + a 2
( Z 2
1 (0 < x < a) sin ax dx (1 ea )
where g( x) = prove that =
0, elsewhere, x( x2 + a2 ) 2 a2
0
Solution. From the Parsevals identity for cosine transform, we have
Z Z
F c () G c () d = f ( x) g( x) dx
0 0
Z Za 2
s
2 a sin a 1 ea
d = eax 1 dx =
( 2 + a 2 ) a2
0 0

Z 2 Z 2
sin a d (1 ea ) sin ax dx (1 ea )
so that = or =
( 2 + a 2 ) 2 a2 x( x2 + a2 ) 2 a2
0 0

Example 7.3. Find the sine and cosine transforms of e x , and use Parsevals identities to prove that
Z Z
dx x2 dx
= =
( x2 + 1)2 4 ( x2 + 1)2
0 0

Solution. Note that

Z
s
2
F c () + i F s () = e x (cos x + i sin x) dx

0
Z Z
s s
2 2
= e x e i x dx = e(1 i) x dx

0 0
s s
2 e(1 i) x

2 1
= =

1 i 1 i


x=0
s
2 1

= +i
1 + 2 1 + 2
Comparing the real and imaginary parts on both sides, we get
s s
2 1 2
F c () = , F s () =
1 + 2 1 + 2

Then from Parsevals identity for cosine transform, we get

Z Z Z
2 d 1 d
= e2 x dx = or = ,
(2 + 1)2 2 (2 + 1)2 4
0 0 0

from which the first result follows by replacing with x. Similarly, from Parsevals identity for sine
transform, the second result follows.

21
 (
2 sin a
r
1 (a x a)
Example 7.4. Given that F f ( x) = = F () where f ( x) =


0, elsewhere,
Z
sin2 ax dx a
prove that =
x2 2
0
Solution. From the Parsevals identity for the complex Fourier transform,
Z Z Z Za
2 sin2 a
|F ()|2 d = ( f ( x))2 dx d = 1. dx
2
a
Z Z
sin2 a sin2 a a
d = a d =
2 2 2
0

Replacing with x, the result follows.

(
2 4 sin2 (a/2)
r
a | x| ( a x a )
Exercise 7.1. If F f ( x) = = F () where f ( x) =


2 0, elsewhere,
Z Z
sin x 2 sin x 4

show that dx = and dx =
x 2 x 3
0 0

22