Fourier integral representation:

Let f(x) be a piecewise continuous and absolutely integrable function of x in


(, ). i, e.,  f ( x) dx exists.

Then f(x) can be represented by an integral as

1
f ( x) 
   A(S ) cos sx  B(s)sin nx  ds
0
(1)

 
where A( s )  

f (t )cos st dt , B( s)   f (t )sin st dt.


Such an integral representation is called the Fourier integral representation of f(x).

The integral on RHS of (1) converges to f(x0) if f(x) is continuous at x0 and to

average of left and right hand limits if f(x) is discontinuous at x0.

Proof: Consider a periodic function fl(x) defined in (-l, l) such that fl(x) =f(x) for
l  x  l
Then, fl(x) can be represented by a Fourier series as

 n x n x 
fl ( x)  a0    an cos  bn sin  (2)
n 1  l l 

Where

n t n t
l l l
1 1 1
2l l l l l l
a0  f l (t ) dt , an  f l (t )cos dt , bn  f l (t )sin dt
l l

Substituting in (2) we get,

   l
n t  n x  1 n t  n x 
l l
1 1
fl ( x)   fl (t ) dt     fl (t )cos dt  cos    fl (t )sin dt  sin 
2l l n1 
 l
l l  l  l
l l  l 

1  n t n x n t n x
l l
1
fl ( x)   fl (t ) dt    fl (t )(cos cos  sin sin ) dt
2l l l n1 l l l l l

n
Let sn 
l
 1 sn
Then sn  sn1  sn  or 
l l 
On substitution we get

1 sn 1 
l l
fl (t ) dt    fl (t )(cos snt cos sn x  sin snt sin sn x) dt sn
2  l
fl ( x) 
 n1 l

 1 l
 1  
l l

f l ( x)    fl (t ) dt  sn   cos sn x  fl (t )cos snt dt  sin sn x  f l (t )sin snt )  sn
 2 l   n1  l l 

Let l  , then fl ( x)  f ( x) .

l 
 l f (t ) dt  sn  0 sn  0.
 l 
Now taking the limit as l  , we get

1  
  
f ( x)   cos sx  f (t )cos stdt  sin sx  f (t )sin stdt  ds
0   

1
   A( s)cos sx  B( s)sin sx  ds
0


Where A( s )  

f (t ) cos st dt
 
B( s)  

f (t ) sin st dt

Note (1): The Fourier integral representation of f ( x ) can also be expressed as

1 
f ( x)   f (t )  cos st cos sx  sin st sin sx  dt ds
 0 

 
1

2 
 
f (t ) cos s( x  t ) dt ds      (3)

cos s( x  t ) is an even function of s.

 
1
Also,
2 
 
f (t ) sin s( x  t ) dt ds  0      (4)

sin s( x  t ) is an odd function of s.

(3)  i(4) gives
 
1
f ( x)   f (t )  cos s( x  t )  i sin s ( x  t  dt ds
2  

 
1

2 
 
f (t ) eis ( xt ) dt ds

Which is called the complex form of Fourier integral representation of f ( x ).

Note (2): If f ( x ) is an even function of x, then

 
A( s )  

f (t ) cos st dt  2  f (t ) cos st dt
0


B( s)  

f (t )sin st dt  0

Therefore Fourier integral becomes

 2
f ( x) 
  A(s) cos sx ds
0

2 

   f (t ) cos st
0 0
cos sx dt ds

Which is called the Fourier Cosine integral representation of f ( x ).

If f ( x ) is an odd function of x, then


A( s )  

f (t ) cos st dt  0

 
A( s )  

f (t ) sin st dt  2  f (t ) sin st dt
0

2
f ( x) 
  B(s) sin sx ds
0

2 

   f (t ) sin st
0 0
sin sx dt ds

Which is called the Fourier Sine integral representation of f ( x ).

Fourier transforms
Consider the Fourier integral representation of the function f ( x ) given by
 
1
f ( x) 
2 
 
f (t ) eis ( xt ) dt ds

1  1  

   2  f (t ) e ist ) dt  ds
isx
e
2   
 
1
Let F ( s ) 
2


f (t )e ist dt      (1)


1
Then f ( x)   F ( s )e ds      (2)
isx

2 

The integral defined by (1) is called the Fourier transform of the function f ( x ) and
is denoted by F  f ( x). Given F (s)  F  f ( x) , the formula (2) defined f ( x),
which is called the inverse Fourier transform of F ( s ) and is denoted by
F 1 ( F ( s) .
Note (1): A function f ( x ) is said to be self-reciprocal under Fourier transforms if
F  f ( x)  F (s).

Note (2): If F  f ( x)  F (s), then f ( x)  F F (s)

1
is called spectral
representation of F ( s ) and F ( s ) is called spectral density of f ( x ). Here s is
called the frequency of the transform. The graph of F ( s) is called amplitude
2
spectrum of f ( x ) and F ( s ) is called energy of the spectrum.

Properties of the Fourier transforms:

(1) Fourier transform is linear
i.e., if c1 and c2 are constants then
F c1 f ( x)  c2 g ( x)  c1F  f ( x)  c2 F g ( x)
Proof: Follows from definition and linearity of the integral.
(2) If F  f ( x)  F (s) then F e  iax
f ( x)  F (s  a)

1
Proof: Consider F ( s  a ) 
2


f (t )e  i ( sa )t dt

1

2


f (t )e ist eiat dt
 

  f (t )e  e
1  ist
 iat
dt
2 

 F e f ( x)
iax

(3) If F  f ( x)  F (s) then F  f ( x  a)  e

 isa
F (s)
Proof: Exercise
1 s
(4) If F  f ( x)  F (s) then F  f (ax)  F 
a a
Proof: Exercise
(5) If F  f ( x)  F (s) then
1
(i) F  f ( x) cos ax   F ( s  a)  F ( s  a) 
2
i
(ii) F  f ( x)sin ax   F (s  a)  F (s  a) 
2
(iii)  
F f ( x)  F ( s )
(iv) F  f ( x)  F (s)
(v)  
F f ( x)  F ( s )
dn
(vi) F  x f ( x)  i
n n
F (s)
ds n
(vii) F  f ( x)  (is) F  f ( x)
( n) n

Convolution:
For functions f ( x) & g ( x), we define the convolution of f ( x) & g ( x), denoted by

1
 f  g  ( x) as  f  g  ( x)   f (t ) g ( x  t )dt; provided the integral exists.
2 

Note that f  g  g  f .

Convolution Theorem:
F  f  g  ( x)  F  f ( x) F g ( x).

Proof:
Consider

F  f  g  ( x) 
1

2 
( f  g )( x)e isx dx

1

 1   isx
   2 
2 
 f (t ) g ( x  t ) dt  e dx

1

 1  
 
 ist  is ( x t )
 f (t ) e  g ( x  t ) e dx  dt
2   2   
Let x  t  u . Then

 1  
F  f  g  ( x) 
1
 
 ist
f (t )e  g (u )e is (u )du  dt
2   2  
 1   1  
 
 ist  is ( u )
 f (t ) e dt  g (u ) e du 
 2    2   
 F  f ( x) F  g ( x).

Parseval’s Identity:
 
If F  f ( x)  F (s) then  | f ( x) |
2
dx   | F (s) |
2
ds .
 

Proof:
If F  f ( x)  F ( s), F{g ( x)}  G ( s), then
F ( s)G ( s )  F  f ( x)  g ( x) or
f ( x)  g ( x)  F 1 F ( s)G ( s)
 
1 1
i.e., 
2 
f (t ) g ( x  t ) dt  
2 
F ( s )G ( s )eisx ds

For x  0, we get
 



f (t ) g (t )dt   F (s)G(s)ds


Let g ( x)  f ( x) or g (  x)  f ( x) .

 
ThenG ( s)  F  g ( x)  F f ( x)  F  f ( x)  F ( s).
  
  f (t ) g (t )dt   f (t ) f (t )dt   F (s) F (s)ds
  
 
  | f (t ) |2 dt   | F ( s) |
2
ds
 

Problems:

1, | x | a
1) Find the Fourier transform of f ( x)   . Hence deduce that
 0, | x | a
 
 
2
sin t  sin t 
 t

dt 
2
and  t 

 
 dt 
2
.

Solution:

1
F  f ( x)   f ( x)eisx dx
2 
a
.
1  eisx 
a
1 2 sin as
2 a
 isx
 1e dx      F ( s)
2  is  a  s
  
1 1 2 sin as
 f ( x)   F ( s )eisx    cos sx  i sin sx ds
2  2   s

2 sin as  sin as 
  cos sxds  sin sx is an odd function of s 
0 s  s 

 0, | x | a


sin as  
 cos sxds  f ( x)   , | x | a
0
s 2  2
1    
 2  2  0   4 , | x | a .
  
For x  0, f ( x)  1.

sin as 
 ds  .
0
s 2

t dt
Let as  t or s  . ds  .
a a
On substitution, we get

sin t dt 
0 t / a a  2

sint 
or 0 t dt 
2
.

By Parseval's identity,
 

 | f ( x) | dx   | F ( s) |
2 2
ds
 
 2  2
2  sin as  4  sin as 
a
i.e.,  1dx   
 0  s 
 ds    ds
a 
 s 

a
2
 sin as 
0  s  ds  2


2
 sin t 
For a  1, we get    ds  .
0
t  2
 
cos xt
a
a x
2) Find the Fourier transform of e , a  0 and hence evaluate dt and
0
2
 t2
a x
F{xe }
 
1 2
e 
a x a x  isx a x
F{e } e dx  e cos sxdx
2  0

e ax
2
 

  a cos sx  s sin sx
 a2  s2 0

2 e o 2 a
 (0  2 2 (a))   F ( s)
 a s  a2  s2

1
 f ( x)   F ( s) e
isx
ds
2 

1 a

2 

a2  s2
(cos sx  i sin sx)ds


2 a cos sx
  2 2 ds
 0 a s
 
cos sx cos xt  f ( x)   a x
  2 2 ds   2 2 dt   e
0
a s 0
a t 2a 2a

Since
dn
F{x f ( x)}   i 
n n
F ( s)
ds n
1 d 2 d a 2  2as 
for n  1, F{xe }   i 
a x a x
F{e }  i 
  (a 2  s 2 )2 
i
ds  ds a 2  s 2
x2

 a2 x2
3) Find the Fourier transform of e , a  0 and hence show that e 2
is
self-reciprocal under Fourier transform.
  
1 1
e e
 a2 x2  a 2 x 2  isx (a x  isx )
} dx 
2 2
F{e e dx
2  2 
2 2
  is  s
1 ax    2
 
 e
2a  4a
dx
2 
s2
 2  is 
2
4a  ax  
e  
 e
2a 
dx
2 

is dt
Let ax   t then dx 
2a a
s2 s2
 
4 a2  4 a2 
e dt 2e
e  e dt
t t

2 2
=
2 
a a 2 

t 2  y or t  y
Put 1 21
 dt  y dy
2
s2 s2 s2 s2
   
4 a2 
e y  12
e e 1 e 4 a2 4 a2
e 4 a2

2 0 2
2 y dy     
a 2  2  a 2 a 2

1
For a  or 2a  1
2
2 2
x s
F{e a x }  F{e } e
2 2
2 2

2
s
e 2
is self-reciprocal under Fourier transform.

1  x x 1
4) Find the Fourier transform of f ( x)   and hence evaluate
 0 x 1
 2
 sin x 
0  x  dx
  1
1 1
  (1  x )(cos sx  i sin sx) dx
 isx
F{ f ( x)}  f ( x) e dx 
2  2 1
1
2

  (1  x)cos sx dx
0
(1  x )sin sx is an odd function of x.

2 (1  cos s)
1
2  sin sx cos sx 
  (1  x)  2    F ( s)
  s s 0  s2

1
 f ( x)   F (s) e
isx
ds
2 

1 2 (1  cos s)

2  
 s2
(cos sx  i sin sx)ds


 (1  x ) x 1

2 (1  cos s) 
  cos sx ds  f ( x )  2
0 s2 2  0 x 1


(1  cos sx)
  
For x  0,  2
ds  (1  0) 
0 s 2 2
2
 2sin ( s / 2) 
 2
ds 
0 s 2
Let s / 2  t or s  2t ds  2dt
 
2
 2 sin 2 t   sin t 
0 4t 2
2dt  or
2  0 
 t 
 dt 
2
Exercises:
1. Obtain the Fourier transform of
1  x 2 x  1 
 x cos x  sin x 
f ( x)   and hence evaluate    cos( x / 2)dx
 0 x  1 0  x 3

x x a
2. Find the Fourier transform of f ( x)  
0 x a0
a 2  x 2 x a
3. Find the Fourier transform of f ( x)  
 0 x a
Fourier Cosine and Sine transforms

Consider the Fourier cosine integral representation of a function f(x)


2
f ( x) 
  f (t )cos st cos sx dtds ,
0 0
x0

2  2 
 

 0   0
  f (t )cos st dt  cos sx ds


2
Let Fc  f ( x)   f (t )cos st dt  Fc (s )....(1)
 0

2
Then f ( x) 
  F  f ( x) cos sx ds
0
c .......(2)

The transform Fc{ f ( x)}defined by (1) is called the Fourier cosine transform of
f(x).The formula (2) is called the inverse Fourier cosine transform of
Fc{ f ( x)}  Fc (s) and is denoted by f ( x)  Fc1 Fc ( s ). Similarly, using the
Fourier sine integral representation of f(x) given by

2
f ( x) 
  f (t )sin st sin sx dt ds, we can define the Fourier
0 0
sine transform of f(x)


2
denoted by Fs  f ( x) as Fs  f ( x)   f (t )sin st dt  F (s)
 0
s

Then the inverse Fourier sine transform of Fs ( s) is defined as


2
 F (s)sin sx ds
1
f ( x)  F {Fs ( s)} 
s
 0
s

Definition: A function f(x) is said to be self reciprocal under Fourier cosine (sine)
transform if Fc  f ( x)  f (s)  Fs  f ( x)  f (s) 
Properties of Fourier cosine/ sine transforms
(1) Both Fourier cosine and sine transforms are linear.
 Fc{c1f (x)  c 2g(x)}  c1Fc{f (x)}  c 2 Fc{g(x)}
Fs{c1f (x)  c 2g(x)}  c1Fs{f (x)}  c 2 Fs{g(x)} where c1 and c 2 are cons tan ts
1
(2). Fc{f (x)cosax}   Fc (s  a)  Fc (s  a) 
2
1
Fc{f (x)sin ax}   Fs (s  a)  Fs (a  s) 
2
1
(3). Fs {f (x)cosax}   Fs (s  a)  Fs (s  a) 
2
1
Fs{f (x)sin ax}   Fc (s  a)  Fs (s  a) 
2
The proof of (2) and (3) follows directly from trigonometric identities.
1
(4). Fc{f (ax)}  Fc (s / a)
a
1
Fs{f (ax)}  Fs (s / a)
a
(5). If f (x)  0 as x  , then
2
Fc{f (x)}   f (0)  s Fs (s)

Fs{f (x)}   s Fc (s)

2
 0
Pr oof : Consider Fc{f (x)}  f (x)cossx dx

Integrating by parts, we get

2 

FC  f '( x)   f ( x) cos sx |0   f ( x) sin sx ( s)dx 


 0 
2 

  0  f (0)  s  f ( x) sin sx dx 
 0 
2
 sFS  f ( x)  f (0) provided f ( x)  0 as x  

Also,

2
FS  f '( x)   f '( x)sin sx dx
 0

Integrating by parts, we get

2 

FS  f '( x)   f ( x ) sin sx |
 0 f ( x ) cos sx ( s ) dx 

0


2  
  0  0  sF ( s )    sFC ( s).
 
C
2 
2
(6) FC  f ''( x)   f '(0)  s 2 FC ( s)

2
FS  f ''( x)  sf (0)  s 2 FS ( s)

provided f ( x) and f '( x)  0 as x  

dFS dF
(7) FC  xf ( x)  and FS xf ( x)   C
ds ds

If FC  f ( x)  FC (s), FC  g ( x)  GC ( s ),

(8)
FS  f ( x)  FS (s) and FS  g ( x)  GS ( s) exist,
 

then  FC (s) GC (s) ds   FS (s) GS (s) ds   f ( x) g ( x)dx
0 0
0

  

 FC ( s) ds   FC ( s) ds   f ( x) dx
2 2 2
and
0 0 0

which is called Parsival’s identity.

Problems:

   
(1) Find FC e ax , FS e ax and hence find FC xe ax   and F xe 
S
 ax

Solution: By definition,

2
FC e  
 ax  ax
e cos sx dx
0


2  e  ax  2 a
      ,a  0
  a 2  s 2 
a cos sx s sin sx
0  a 2
 s 2

2
FS e  
 ax  ax
e sin sx dx
0


2  e  ax  2 s
    a sin sx  s cos sx    , a  0.
 a  s
2 2
0  a s
2 2

Therefore,

d  2 s  2 a2  s2
FC  xe   ds FS e   ds   a 2  s 2    (a 2  s 2 )2
 ax d  ax

 

d  2 a 
FS  xe ax    FC e ax    
d 2 2as
 
ds ds   a 2  s 2   (a 2  s 2 ) 2

2).  
Find Fc e  a x & Fs xe  a x .
2 2

 2 2

Solution : We have from definition,

  
Fc e   a2 x2
 
2  a2 x2
 0
e cos sxdx 
2  a 2 x2 isx
 0
e e dx  F e  a2 x2
 
s2
1  4 a2
 e
2a
d  1  4 a2 
2 2

   
s s
d s  2
 a2 x2  a2 x2
 Fs xe   Fc e   e  e 4a
ds ds  2a  2 2a 3

Note that

  x   s
 
2 2
1  a2 x2
for a  , Fc e  Fc e 2 e 2
2  
  x 
 
2 2
s

 a2 x2
& Fs xe  Fs  xe 2   xe 2 .
 
x2

e 2
is self reciprocal under Fourier cosine transform and
x2

2
xe is self reciprocal under Fourier sine transform .

3) Find Fc  x a1 & Fs  x a1 ,0  a  1.


Consider  x a 1e isx dx;
0

dy
Let isx  y or dx  . Then
is
  a
1 ( a ) a ( a )  i 2  (a)
x 0 y e dy  i a s a  (i) s a   e  s a
a 1  isx a 1  y
e dx 
 is 
a
0

 a   ( a )  a   ( a )
 cos   a  i sin   a
 2  s  2  s
Thus
 
 a   ( a )  a   ( a )
x cos sxdx  i  x a 1 cos sxdx  cos   a  i sin   a
a 1

0 0  2  s  2  s
Equating real and imaginary parts, we get

2 ( a )  a 
Fc  x  
2 a 1
 0
a 1
x cos sxdx  cos   and
 s a
 2 

2  ( a )  a 
Fs  x  
2 a 1
 0
a 1
x sin sxdx  sin   .
 sa  2 
Note:

 1  1  1   1 
For a  , Fc  x a 1  Fc   
1
 Fs         
2  x s  x  2 
1
 is self reciprocal under Fourier sine and cosine transforms.
x
Note:

2
Consider Fc  f ( x) 
 0
f ( x)cos sxdx  Fc ( s ).


2
Then f ( x)  Fc Fc ( s) 
 0
1
Fc ( s)cos sxds      (1)

Interchanging s and x in (1), we get


2
Fc ( x)cos sxdx  Fc Fc ( s )
 0
f (s) 

Similarly, if Fs  f ( x)  Fs ( s) then, Fs Fs ( s)  f ( s).

4) Find

 1   x 
Fc  2 
and Fs  2 
.
1  x  1  x 
Solution: We have shown that

Fc e  ax  
2 a
 Fc ( s ).
 a  s2
2


 2 a  
 Fc  Fc ( x)  Fc  2 
 e  as .
  a x 
2
 
 a    as
Or Fc  2 2 
 e .
a  x  2
 1   -s
For a  1, Fc  2 
 e .
1  s  2
 x  d  -s  -s
Fs     e  e .
1  x 
2
ds 2 2
Exercises:
1) Find the Fourier cosine transform of e ax cos ax, e ax sin ax and hence find the
1 x2
Fourier cosine transforms of and .
 x 4
 k 4
  x 4
 k 4

1 e  ax
Find the Fourier sine transform of and , a  0.
x x