Differential Equations & Transforms

(BMAT102L)

Module 6
(Infinite) Fourier Transforms

Dr. T. Phaneendra
Professor of Mathematics
(Higher Academic Grade)

School of Advanced Sciences

Vellore Institute of Technology
Vellore-632014
Tamil Nadu, India

E-mail: phaneendra.t@vit.ac.in

May 12, 2022

Contents

1 Complex Fourier Transform 2

1.1 Complex Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 Properties of the Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Fourier Convolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2 Fourier Sine and Cosine Transforms 10

2.1 Fourier Sine and Cosine Transforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.2 Parseval’s Identities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

1
Chapter 1

Complex Fourier Transform

1.1 Complex Fourier Transform

Let 𝑓 (𝑥) be defined for all real −∞ < 𝑥 < ∞. Then its Fourier transform is given by
∞
1∫
𝑓 (𝑥) 𝑒 −𝑖ω𝑥 d𝑥, −∞ < ω < ∞.

ℱ 𝑓 (𝑥) = 𝐹 (ω) = √ (1.1.1)
2π −∞

Let 𝑓 (𝑥) represent a spatial signal. Then

(a) its Fourier transform ℱ(ω) is known as complex frequency spectrum,
(b) the graph of the magnitude |ℱ(ω)| is called its amplitude spectrum,
(c) the graph of the argument arg ℱ(ω) is called its phase spectrum.
Fourier Inversion formula:
∞
1∫
𝑓 (𝑥) = ℱ −1 𝐹 (ω) = √

𝐹 (ω) 𝑒 𝑖ω𝑥 dω. (1.1.2)
2π −∞

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

1.2 Properties of the Fourier Transform

Theorem

1.2.1 (Linearity). Let 𝐹 (ω) and 𝐺 (ω) be the Fourier transforms of 𝑓 (𝑥) and 𝑔(𝑥) respectively.
Then ℱ 𝑎 𝑓 (𝑥) + 𝑏𝑔(𝑥) = 𝑎𝐹 (ω) + 𝑏𝐺 (ω), where 𝑎 and 𝑏 are scalars, not both zeros.

Example 1.2.1. Find the Fourier transform of the negative exponential signal
(
𝑒 −𝑎𝑥 , 𝑥 ≥ 0,
𝑓 (𝑥) =
0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒.

Hence find the amplitude and phase spectra of the signal 𝑓 (𝑥).

Solution. We have

∞ ∞
1∫ 1∫
𝑒 −𝑎𝑥 𝑒 −𝑖ω𝑥 d𝑥 = √ 𝑒 −(𝑎+𝑖ω) 𝑥 d𝑥

ℱ 𝑓 (𝑥) = 𝐹 (ω) = √
2π 0 2π 0
∞
1 𝑒 −(𝑎+𝑖ω) 𝑥 1 1 1 h 𝑎 n ω oi
=√ · − =√ =√ − 𝑖 ·
2π 𝑎 + 𝑖ω 𝑥=0 2π 𝑎 + 𝑖ω 2π 𝑎 2 + ω2 𝑎 2 + ω2

Note that the amplitude spectrum of 𝑓 (𝑥) is |𝐹 (ω)| = √1 · √ 1 and the phase spectrum of 𝑓 (𝑥) is
2π 𝑎2 +ω2
arg 𝐹 (ω) = tan−1 ω𝑎 ·

Example 1.2.2. Find the Fourier transform of the saw-tooth signal


 𝑥, |𝑥| < 𝑎,



𝑓 (𝑥) =

 0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒.



2
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Complex Fourier Transform Professor of Mathematics

Solution. We have

𝑎 𝑎
1∫ 1∫
𝐹 (ω) = √ 𝑥𝑒 −𝑖ω𝑥 d𝑥 = √ 𝑥 [cos ω𝑥 − 𝑖 sin ω𝑥] d𝑥
2π −𝑎 2π −𝑎
odd function even function
𝑎 z }| { 𝑎 z }| {
1∫ 1∫
=√ 𝑥 cos ω𝑥 −𝑖 · √ 𝑥 sin ω𝑥 d𝑥
2π −𝑎 2π −𝑎
| {z } | {z }
=0 ∫ 𝑎
=2 𝑥 sin ω𝑥 d𝑥
0
√︂   𝑎
√︂  
2 n cos ω𝑥 o sin ω𝑥 2 𝑎 cos ω𝑎 sin ω𝑎
= −𝑖 𝑥 − − (1) − =𝑖 − ,ω ≠ 0
π ω ω2 𝑥=0 π ω ω2

Example 1.2.3. Find the Fourier transform of the parabolic signal


 1 − 𝑥 2 , |𝑥| < 1,



𝑓 (𝑥) =

 0,
 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒.

∞  
∫ sin 𝑥 − 𝑥 cos 𝑥 𝑥  3π
Hence deduce that cos d𝑥 = ·
0 𝑥3 2 16

Solution. We have

1 1
1∫ 1∫
𝐹 (ω) = √ (1 − 𝑥 2 )𝑒 −𝑖ω𝑥 d𝑥 = √ (1 − 𝑥 2 ) [cos ω𝑥 − 𝑖 sin ω𝑥] d𝑥
2π −1 2π −1
even function odd function
1 z }| { 1 z }| {
1∫ 1∫
=√ (1 − 𝑥 2 ) cos ω𝑥 −𝑖 · √ (1 − 𝑥 2 ) sin ω𝑥 d𝑥
2π −1 2π −1
| {z } | {z }
∫ 1 =0
=2 (1−𝑥 2 ) cos ω𝑥 d𝑥
0
√︂   n cos ω𝑥 o  1
2 2 sin ω𝑥 sin ω𝑥
= (1 − 𝑥 ) − − (−2𝑥) − + (−2) −
π ω ω2 ω3 𝑥=0
√︂   1
√︂  
2 2 cos ω sin ω 2 cos ω sin ω
= 0− + − (0 − 0 − 0) = −2 − ,ω ≠ 0
π ω2 ω3 𝑥=0 π ω2 ω3

By the inversion formula (1.1.2),

∞
1∫
𝑓 (𝑥) = ℱ −1 𝐹 (ω) = √

𝐹 (ω)𝑒 𝑖ω𝑥 dω
2π −∞
" √︂  #
∞
1∫ 2 cos ω sin ω
⇒ √ −2 − 𝑒 𝑖ω𝑥 dω = 𝑓 (𝑥)
2π −∞ π ω2 ω3
∫ ∞  cos ω sin ω  π
⇒ 2
− 3
𝑒 𝑖ω𝑥 dω = − · 𝑓 (𝑥)·
−∞ ω ω 2

Writing 𝑥 = 1/2 in this, we see that

∫ ∞    
cos ω sin ω 𝑖ω/2 π 1
+ 𝑒 dω = − 𝑓
−∞ ω2 ω3 2 2

3
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Complex Fourier Transform Professor of Mathematics

∫ ∞  
cos ω sin ω h ω  ω i π 3 3π
or 2
+ 3
cos − 𝑖 sin dω = − · = − ·
−∞ ω ω 2 2 2 4 8

Since the imaginary part on the left hand side is odd function of ω, we get

∫ ∞   ω
cos ω sin ω 3π
2
− 3
cos dω = − ·
0 ω ω 2 16

Multiplying with −1, and replacing ω with 𝑥 in the integral, we get

∞  
∫ sin 𝑥 − 𝑥 cos 𝑥 𝑥  3π
3
cos d𝑥 = ·
0 𝑥 2 16

Theorem 1.2.2 (Duality). h 𝑓 (𝑥), 𝐹 (ω)i is a Fourier transform pair if and only if h𝐹 (𝑥), 𝑓 (−ω)i is a
Fourier pair.

Example 1.2.4. Find the Fourier transform of the rectangular pulse


 𝑘, |𝑥| < 𝑎,



𝑓 (𝑥) =

 0,
 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒,


where 𝑘 and 𝑎 are positive numbers. Hence show that

∫ ∞ sin 𝑎ω cos ω𝑥 π
(𝑎) dω = · 𝑓 (𝑥),
0 ω 2
∫ ∞ sin 𝑥 π
(𝑏) d𝑥 = ,
0 𝑥 2
sin 𝑥
Then, obtain the Fourier transform of 𝑔(𝑥) = 𝑥 ·

Solution. We see that

𝑒 −𝑖ω𝑥 − 𝑒 −𝑖ω𝑎
𝑎 𝑎  𝑖ω𝑎 
1∫ 𝑘 𝑘 𝑒
𝐹 (ω) = √ 𝑘 · 𝑒 −𝑖ω𝑥 d𝑥 = √ − =√
2π −𝑎 2π 𝑖ω 𝑥=−𝑎 2π 𝑖ω
  √︂  
𝑘 2 sin ω𝑎 2 sin ω𝑎
=√ =𝑘 .
2π ω π ω

(a) By the inversion formula (1.1.2),

∞
1∫
𝑓 (𝑥) = ℱ −1 𝐹 (ω) = √

𝐹 (ω)𝑒 𝑖ω𝑥 dω
2π −∞
∞ √︂   ∞  
1∫ 2 sin ω𝑎 𝑖ω𝑥 ∫ sin ω𝑎 𝑖ω𝑥 π
⇒ √ 𝑘 𝑒 dω = 𝑓 (𝑥) ⇒ 𝑒 dω = · 𝑓 (𝑥)
2π −∞ π ω −∞ ω 𝑘
∫ ∞  sin ω𝑎  π
⇒ (cos ω𝑥 + 𝑖 sin ω𝑥) dω = · 𝑓 (𝑥)
−∞ ω 𝑘

Comparing the real parts on both sides,

∫ ∞  
sin ω𝑎 cos ω𝑥 π
dω = · 𝑓 (𝑥), −∞ < 𝑥 < ∞.
−∞ ω 𝑘

4
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Complex Fourier Transform Professor of Mathematics

(b) In particular, for 𝑥 = 0, 𝑎 = 1, this gives

∫ ∞   ∫ ∞  
sin ω π π sin ω π
dω = · 𝑓 (0) = · 𝑘 = π or dω = ·
−∞ ω 𝑘 𝑘 0 ω 2

Replacing ω with 𝑥 in the integral, we get

∫ ∞  
sin 𝑥 π
d𝑥 = ·
0 𝑥 2

To find the Fourier transform of 𝑔(𝑥) = sin𝑥 𝑥 , we employ the duality property that h𝐹 (𝑥), 𝑓 (−ω)i is a Fourier
 √︃

pair. That is, ℱ 𝐹 (𝑥) = 𝑓 (−ω). But 𝐹 (𝑥) = 𝑘 π2 sin𝑥𝑎𝑥 for any 𝑎 > 0. Thus

( √︂ ) 
 𝑘, |ω| < 𝑎,


2 sin 𝑎𝑥 
ℱ 𝑘 · = 𝑓 (ω) =
π 𝑥 
 0,
 elsewhere


or with 𝑎 = 1, we get

  √︁
 π2 , |ω| < 1,
 

sin 𝑥 
ℱ =
𝑥 
 0,
 elsewhere.


Example 1.2.5. Show that the Gaussian signal 𝑓 (𝑥) = 𝑒−𝑥 /2 , −∞ < 𝑥 < ∞ is self-reciprocal.
2

Solution. We say that 𝑓 (𝑥) is self-reciprocal if 𝐹 (ω) = 𝑓 (ω).

∞ ∞
1∫ 1∫
𝑒 −𝑥
2 /2
· 𝑒 −𝑖ω𝑥 d𝑥 = √ 𝑒 −( 𝑥
2 +2𝑥.𝑖ω)/2
𝐹 (ω) = √ d𝑥
2π −∞ 2π −∞
∞ ∞
𝑒 −ω /2∫
2
1∫
𝑒 −[ ( 𝑥+𝑖ω)
2 −(𝑖ω) 2 ]/2
𝑒 −( 𝑥+𝑖ω)
2 /2
=√ d𝑥 = √ d𝑥
2π −∞ 2π −∞
∞
1∫
= 𝑒 −ω
2 /2
𝑒 −𝑧 /2 𝑑𝑧 = 𝑒 −ω /2 ·
2 2
·√
2π −∞
| {z }
=1

Thus 𝑓 (𝑥) = 𝐹 (ω). Hence 𝑓 (𝑥) = 𝑒 −𝑥

2 /2
is self-reciprocal.

Theorem 1.2.3 (Change of Scale). Let 𝐹 (ω) be the Fourier transform of 𝑓 (𝑥), and 𝑎 ≠ 0. Then

 1 ω
ℱ 𝑓 (𝑎𝑥) = ·𝐹 · (1.2.1)
|𝑎| 𝑎

∫ ∞
𝑓 (𝑎𝑥)𝑒 −𝑖ω𝑥 d𝑥.

Proof. Let 𝐼 = ℱ 𝑓 (𝑎𝑥) = √1
2π
−∞
Case (a): If 𝑎 > 0, write 𝑎𝑥 = 𝑢 so that d𝑥 = 𝑑𝑢/𝑎 and 𝑢 ranges from −∞ to ∞ as 𝑥 ranges from −∞ to ∞.
Therefore,

∞
1 1∫ 1 ω
𝑓 (𝑢)𝑒 −𝑖 (ω/𝑎)𝑢 𝑑𝑢 =

𝐼 = ℱ 𝑓 (𝑎𝑥) = · √ ·𝐹 ·
𝑎 2π −∞ 𝑎 𝑎

5
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Complex Fourier Transform Professor of Mathematics

Case (b): If 𝑎 < 0 so that −𝑎 > 0. Then write −𝑎𝑥 = 𝑣 so that d𝑥 = −𝑑𝑣/𝑎 and 𝑣 ranges from ∞ to −∞ as 𝑥
ranges from −∞ to ∞. Therefore,

−∞
1 1∫
𝑓 (−𝑣)𝑒 𝑖 (ω/𝑎) 𝑣 𝑑𝑢.

𝐼 = ℱ 𝑓 (𝑎𝑥) = · √
𝑎 2π ∞

Now, write −𝑣 = 𝑝 so that 𝑑𝑣 = −𝑑𝑝 and 𝑝 ranges from −∞ to ∞ as 𝑣 ranges from ∞ to −∞. Therefore,
the above equation becomes

∞
1 1∫ 1 ω
𝐼=− ·√ 𝑓 ( 𝑝)𝑒 −𝑖 (ω/𝑎) 𝑝 𝑑𝑝 = − ·𝐹 ·
𝑎 2π −∞ 𝑎 𝑎

This completes the proof. 

Example 1.2.6. Given that the Fourier transform of 𝑒−𝑥 /2 is 𝑒−ω /2 , find the Fourier transform of 𝑒−𝑏𝑥 ,
2 2 2

where 𝑏 > 0.

Solution. Let 𝑓 (𝑥) = 𝑒 −𝑥

2 /2
. Then

ℱ 𝑓 (𝑎𝑥) = 𝐹 (ω) = 𝑒 −ω /2 .
 2

Since

√ √
𝑒 −𝑏𝑥 = 𝑒 −2𝑏𝑥
2 2 /2
= 𝑒− 2𝑏𝑥 2 /2
= 𝑓 ( 2𝑏𝑥),

by (1.2.1), we see that

√
 
1 ω
= 𝑒 −ω /4𝑏 .
 2
ℱ 𝑓 ( 2𝑏𝑥) = √ · 𝐹 √
2𝑏 2𝑏

Theorem 1.2.4 (Spacial-Shifting). Let 𝐹 (ω) be the Fourier transform of 𝑓 (𝑥), and 𝑎 ≠ 0. Then

ℱ 𝑓 (𝑥 − 𝑎) = 𝑒 −𝑖𝑎ω 𝐹 (ω).

(1.2.2)

Theorem 1.2.5 (Frequency-Shifting). Let 𝐹 (ω) be the Fourier transform of 𝑓 (𝑥), and 𝑎 ≠ 0. Then

ℱ 𝑒 −𝑖𝑎𝑥 𝑓 (𝑥) = 𝐹 (ω + 𝑎).


(1.2.3)

Theorem 1.2.6 (Modulation). Let 𝐹 (ω) be the Fourier transform of 𝑓 (𝑥), and 𝑎 ≠ 0. Then

 1
ℱ 𝑓 (𝑥) cos 𝑎𝑥 = [𝐹 (ω − 𝑎) + 𝐹 (ω − 𝑎)]. (1.2.4)
2

6
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Complex Fourier Transform Professor of Mathematics

Example 1.2.7. Find the Fourier transform of



 1, 0 < 𝑥 < 1,


𝑓 (𝑥) =

 0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒.



Hence obtain the Fourier transform of


 𝑒 −𝑖𝑎𝑥 , 0 < 𝑥 < 1,



𝑔(𝑥) =

 0,
 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒.


Solution. First,

𝑒 −𝑖ω𝑥
1 1
1∫ 1 𝑖
𝑒 −𝑖ω𝑥 d𝑥 = √ · − 1 − 𝑒 −𝑖ω .


ℱ 𝑓 (𝑥) = √ =− √
2π 0 2π 𝑖ω 𝑥=0 ω 2π

Then by the frequency-shifting property (1.2.3), we get

𝑖 1 − 𝑒 −𝑖 (ω+𝑎)


−𝑖𝑎𝑥
 
ℱ 𝑔(𝑥) = ℱ 𝑒 𝑓 (𝑥) = 𝐹 (ω + 𝑎) = √ ·
(ω + 𝑎) 2π

Theorem 1.2.7 (Multiplication by 𝑥 𝑛 ). Let 𝐹 (ω) be the Fourier transform of 𝑓 (𝑥). Then

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


(1.2.5)

Example 1.2.8. Find the Fourier transform of the decaying exponential signal 𝑓 (𝑥) = 𝑒−| 𝑥 | for all −∞ <
𝑥 < ∞. Hence
∫ ∞
cos ω𝑥 π
(𝑎) derive that dω = · 𝑒 − |𝑥 | , and
0 1+ω 2 2
(𝑏) obtain the Fourier transform of 𝑥𝑒 − | 𝑥 | .

Solution. Note that

∞ ∞
1∫ 1∫
𝐹 (ω) = √ 𝑒 − |𝑥 | 𝑒 −𝑖ω𝑥 d𝑥 = √ 𝑒 −| 𝑥 | (cos ω𝑥 − 𝑖 sin ω𝑥) d𝑥
2π −∞ 2π −∞
even function odd function
∞ z }| { ∞ z }| {
1∫ 1∫
=√ 𝑒 − |𝑥 | cos ω𝑥 d𝑥 −𝑖 √ 𝑒 −|𝑥 | sin ω𝑥) d𝑥
2π −∞ 2π −∞
| {z } | {z }
∫ ∞ =0
=2 𝑒−| 𝑥| cos ω𝑥 d𝑥
0
∞ √︂ ∞
√︂
2∫ −𝑥 2 𝑒 −𝑥 (− cos ω𝑥 + ω sin ω𝑥) 2 1
=√ 𝑒 cos ω𝑥 d𝑥 = · = · .
2π 0 π 1 + ω2 𝑥=0 π 1 + ω2

7
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Complex Fourier Transform Professor of Mathematics

(a) By the inversion formula (1.1.2),

"√︂ #
∞ ∞
−1
 1∫ 𝑖ω𝑥 1∫ 2 1
𝑓 (𝑥) = ℱ 𝐹 (ω) = √ 𝐹 (ω)𝑒 dω = √ · 𝑒 𝑖ω𝑥 dω
2π −∞ 2π −∞ π 1 + ω2
∞   ∞  
1∫ 1 𝑖ω𝑥 1∫ 1
= 𝑒 dω = (cos ω𝑥 + 𝑖 sin ω𝑥) dω
−∞ 1 + ω π −∞ 1 + ω2
π 2
∞ 
2∫ cos ω𝑥 
= dω
π 0 1 + ω2

∫ ∞
cos ω𝑥 dω π − | 𝑥 |
or = · 𝑒 , −∞ < 𝑥 < ∞.
0 1 + ω2 2
(b) By the multiplication by 𝑥 property (1.2.5) with 𝑛 = 1, we have

√︂ ! √︂
 0𝑑 2 1 2 2𝑖ω
ℱ 𝑥 𝑓 (𝑥) = 𝑖 · 𝐹 (ω) = 𝑖 · · =− ·
dω π 1 + ω2 π (1 + ω2 ) 2

√︂
2 2𝑖ω
or ℱ 𝑥𝑒 − | 𝑥 | = −

· .
π (1 + ω2 ) 2

1.3 Fourier Convolution

∫ ∞
We define ( 𝑓 ∗ 𝑔) (𝑥) = √1 𝑓 (𝑢)𝑔(𝑥 − 𝑢) d𝑢, −∞ < 𝑥 < ∞.
2π
−∞

Theorem 1.3.1 (Convolution Theorem). ℱ ( 𝑓 ∗ 𝑔) (𝑥) = 𝐹 (ω) · 𝐺 (ω).



Proof. By definition, we have

∞
1∫
( 𝑓 ∗ 𝑔) (𝑥)𝑒 −𝑖ω𝑥 d𝑥

ℱ ( 𝑓 ∗ 𝑔) (𝑥) = √
2π −∞
∞  ∞ 
1∫ 1∫
=√ √ 𝑓 (𝑢)𝑔(𝑥 − 𝑢)𝑑𝑢 𝑒 −𝑖ω𝑥 d𝑥
2π −∞ 2π −∞
∞  ∞ 
1∫ 1∫
=√ 𝑓 (𝑢)𝑒 −𝑖ω𝑢 √ 𝑔(𝑥 − 𝑢)𝑒 −𝑖ω( 𝑥−𝑢) d𝑥 d𝑢
2π −∞ 2π −∞
 ∫ ∞  ∞ 
1 1∫
= √ 𝑓 (𝑢)𝑒 −𝑖ω𝑢 d𝑢 √ 𝑔(𝑥 − 𝑢)𝑒 −𝑖ω( 𝑥−𝑢) d𝑥
2π −∞ 2π −∞
 ∞  ∫ ∞ 
1∫ −𝑖ω𝑢 1 −𝑖ω𝑣
= √ 𝑓 (𝑢)𝑒 d𝑢 √ 𝑔(𝑣)𝑒 d𝑣
2π −∞ 2π −∞
= 𝐹 (ω) · 𝐺 (ω).

That is, the Fourier transform of the convolution of two functions equals the product of the respective
Fourier transforms. 

Theorem 1.3.2 (Fourier transform of Derivatives). Suppose that

(a) 𝑓 , 𝑓 0, 𝑓 00, ..., 𝑓 (𝑛−1) all tend to 0 as |𝑥| → ∞, and

∫ ∞ ∫ ∞
(b) | 𝑓 (𝑥)| d𝑥 < ∞, 𝑓 (𝑘) (𝑥) d𝑥 < ∞, 𝑘 = 1, 2, ..., 𝑛.
−∞ −∞

8
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Complex Fourier Transform Professor of Mathematics

If 𝐹 (ω) be the Fourier transform of 𝑓 (𝑥), then

ℱ 𝑓 (𝑛) (𝑥) = (𝑖ω) 𝑛 𝐹 (ω), for 𝑛 = 1, 2, ...



Exercise 1.3.1. Find the complex Fourier transform of the triangular pulse


 𝑒 𝑖ω𝑥 , 𝑎 < 𝑥 < 𝑏,



𝑓 (𝑥) =

 0,
 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒.


√︃
Ans. 2
π · 1−cos ω
ω2
, for ω ≠ 0

Exercise 1.3.2. Find the complex Fourier transform of


 1 − |𝑥|, |𝑥| < 1,



𝑓 (𝑥) =

 0,
 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒.


Exercise 1.3.3. Find the complex Fourier transform of

 1
, |𝑥| < 𝑎,

 √2π𝑎


𝑓 (𝑥) =

 0,
 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒.


Exercise 1.3.4. Find the complex Fourier transform of

 1
, |𝑥| < 𝑎,

 √2π𝑎


𝑓 (𝑥) =

 0,
 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒.


Exercise 1.3.5. Given that the Fourier transform of 𝑒−𝑥 /2 is 𝑒−ω /2 , find the Fourier transform of
2 2

𝑒 −4( 𝑥−3) and 𝑒 −𝑥 cos 3𝑥.

2 2

9
Chapter 2

Fourier Sine and Cosine Transforms

2.1 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

√︂ ∫ ∞
 2
ℱ𝑠 𝑓 (𝑥) = 𝐹𝑠 (ω) = 𝑓 (𝑥) sin ω𝑥 d𝑥
π 0
The Sine Inversion Formula of f is given by
√︂ ∫ ∞
2
𝑓 (𝑥) = ℱ𝑠−1 𝐹𝑠 (ω) =

𝐹𝑠 (ω) sin ω𝑥 dω
π 0
(𝑏) The Fourier Cosine transform of f is given by
√︂ ∫ ∞
 2
ℱ𝑐 𝑓 (𝑥) = 𝐹𝑐 (ω) = 𝑓 (𝑥) cos ω𝑥 d𝑥
π 0
The Cosine Inversion Formula of f is given by
√︂ ∫ ∞
−1
 2
𝑓 (𝑥) = ℱ𝑐 𝐹𝑐 (ω) = 𝐹𝑐 (ω) cos ω𝑥 dω
π 0

Example 2.1.1. Find the cosine transform of


 1, 0 ≤ 𝑥 ≤ 𝑎



𝑓 (𝑥) =

 0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒,



where 𝑎 > 0.

Solution. We have
√︂ ∫ ∞ √︂ ∫ 𝑎 √︂ 𝑎
√︂
2 2 2 sin ω𝑥 2 sin ω𝑎
𝐹𝑐 (ω) = 𝑓 (𝑥) cos ω𝑥 d𝑥 = cos ω𝑥 d𝑥 = = ·
π 0 π 0 π ω 𝑥=0 π ω

Example 2.1.2. Find the cosine transform of 𝑓 (𝑥) = 𝑒−𝑎𝑥 , 𝑎 > 0, 𝑥 > 0.

Solution. We have
√︂ ∫ ∞ √︂ ∫ ∞
2 2
𝐹𝑐 (ω) = 𝑓 (𝑥) cos ω𝑥 d𝑥 = 𝑒 −𝑎𝑥 · cos ω𝑥 d𝑥
π 0 π 0
∞
√︂ √︂
2 𝑒 −𝑎𝑥 2 𝑎
= · (−𝑎 cos ω𝑥 + ω sin ω𝑥) = · 2 ,
π 𝑎 2 + ω2 𝑥=0 π 𝑎 + ω2

Example 2.1.3. Find the sine transform of 𝑓 (𝑥) = 𝑒−𝑎𝑥 , 𝑎 > 0, 𝑥 > 0, and and hence deduce that
∞
∫ 𝑥 sin 𝑚𝑥 π𝑒 −𝑎
d𝑥 = ·
0 1 + 𝑥2 2

10
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Sine and Cosine Transform Professor of Mathematics

Solution. We have
√︂ ∫ ∞ √︂ ∫ ∞
2 2
𝐹𝑠 (ω) = 𝑓 (𝑥) sin ω𝑥 d𝑥 = 𝑒 −𝑎𝑥 · sin ω𝑥 d𝑥
π 0 π 0
∞
√︂ √︂
2 𝑒 −𝑎𝑥 2 ω
= · 2 (−𝑎 sin ω𝑥 − ω cos ω𝑥) = · 2
π 𝑎 +ω 2
𝑥=0 π 𝑎 + ω2

By the inversion formula,

(√︂ )
√︂ ∫ ∞ √︂ ∫ ∞
2 2 2 ω
𝑓 (𝑥) = 𝐹𝑠 (ω) sin ω𝑥 dω = · sin ω𝑥 dω
π 0 π 0 π 𝑎 2 + ω2

or
∞
∫ ω sin ω𝑥 dω π π𝑒 −𝑎𝑥
= 𝑓 (𝑥) = ·
0 𝑎 2 + ω2 2 2

Changing 𝑥 to 𝑚, ω to 𝑥 and 𝑎 = 1 we get

∞
∫ 𝑥 sin 𝑚𝑥 d𝑥 π𝑒 −𝑚
= ·
0 1 + 𝑥2 2

Example 2.1.4. Find the sine transform of 𝑓 (𝑥) = 1/𝑥, 𝑥 > 0.

Solution. We have
√︂ ∫ ∞ √︂ √︂
2 sin ω𝑥 2 π π
𝐹𝑠 (ω) = d𝑥 = · = ·
π 0 𝑥 π 2 2

Example 2.1.5. 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 /2
is self-reciprocal under the Fourier transform. That is

∞ ∞
1∫ 1∫
𝑒 −ω
2 /2
𝑒 −𝑥
2 /2
· 𝑒 −𝑖ω𝑥 d𝑥 = √ 𝑒 −𝑥
2 /2
=√ (cos ω𝑥 − 𝑖 sin ω𝑥) d𝑥
2π −∞ 2π −∞
even function odd function
∞ z }| { ∞ z }| {
1∫ −𝑥 2 /2 1∫ −𝑥 2 /2
=√ 𝑒 cos ω𝑥 d𝑥 − 𝑖 √ 𝑒 sin ω𝑥 d𝑥
2π −∞ 2π −∞
| {z }
=0
∞ √︂ ∫ ∞
2∫ −𝑥 2 /2 2 −𝑥 2 /2
=√ 𝑒 cos ω𝑥 d𝑥 = 𝑒 cos ω𝑥 d𝑥.
2π 0 π 0

In other words, ℱ𝑐 𝑒 −𝑥 /2 = 𝑒 −ω /2 . Hence 𝑓 (𝑥) = 𝑒 −𝑥 /2 is self-reciprocal.

 2 2 2

Theorem 2.1.1 (Cosine transform from the Sine transform). If ℱ𝑠 𝑓 (𝑥) = 𝐹𝑠 (ω) and 𝑓 (𝑥) →


0 as 𝑥 → ∞, then

ℱ𝑐 𝑓 0 (𝑥) = ω𝐹𝑠 (ω)


(2.1.1)

ℱ𝑐 𝑥 𝑓 (𝑥) = 𝐹𝑠0 (ω)


(2.1.2)

11
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Sine and Cosine Transform Professor of Mathematics

Theorem 2.1.2 (Sine transform from the Cosine transform). If ℱ𝑐 𝑓 (𝑥) = 𝐹𝑐 (ω) and 𝑓 (𝑥) →


0 as 𝑥 → ∞, then

ℱ𝑠 𝑓 0 (𝑥) = −ω𝐹𝑐 (ω)


(2.1.3)

ℱ𝑠 𝑥 𝑓 (𝑥) = −𝐹𝑐0 (ω)


(2.1.4)

Example 2.1.6. Given that 𝑓 (𝑥) = 𝑒−𝑥 /2 is self-reciprocal under the cosine transform, find the sine
2

transform of 𝑔(𝑥) = 𝑥𝑒 −𝑥
2 /2
and the cosine transform of ℎ(𝑥) = 𝑥 2 𝑒 −𝑥
2 /2
.

Solution. We have Given that ℱ𝑐 𝑒 −𝑥 /2 = 𝑒 −ω /2 . Let 𝑓 (𝑥) = 𝑒 −𝑥 /2 and 𝐹 (ω) = 𝑒 −ω /2 . Then by Theorem
 2 2 2 2

3, we have

ℱ𝑠 𝑓 0 (𝑥) = −ω𝐹𝑐 (ω)



⇒ ℱ𝑠 − 𝑥𝑒 −𝑥 /2 = −ω𝐹𝑤 (ω)
 2

⇒ ℱ𝑠 𝑥𝑒 −𝑥 /2 = ω𝐹𝑐 (ω) = ω𝑒 −ω /2 = 𝐺 𝑠 (ω), say.

 2 2

In other words, 𝑔(𝑥) = 𝑥𝑒 −𝑥

2 /2
is self-reciprocal under the sine transform.

Now by the multiplication by 𝑥 property, we have

𝑑 𝑑 n −ω2 /2 o
ℱ𝑐 𝑥(𝑥𝑒 −𝑥 /2 ) =
2
= (1 − ω2 )𝑒 −ω /2
 2
{𝐺 𝑠 (ω)} = ω𝑒
dω dω

Thus ℱ𝑐 ℎ(𝑥) = (1 − ω2 )𝑒 −ω /2 .
 2

𝑒−𝑎𝑥
Example 2.1.7. Find the sine transform of 𝑓 (𝑥) = 𝑥 and hence the cosine transform of 𝑔(𝑥) = 𝑒 −𝑎𝑥 ·
∞
𝑒 −𝑎𝑥
√︃ ∫  
Solution. We have Let 𝐼 = 𝐹𝑠 (ω) = 2
π sin ω𝑥 d𝑥. Differentiating w. r. t. ω under the integral
0 𝑥
sign, this gives
√︂ ∫ ∞ √︂ ∫ ∞ √︂
𝑒 −𝑎𝑥
 
𝑑𝐼 2 2 2 𝑎
= (𝑥 cos ω𝑥) d𝑥 = 𝑒 −𝑎𝑥 cos ω𝑥 d𝑥 = · ·
dω π 0 𝑥 π 0 π 𝑎 2 + ω2

On one hand, this gives

∫ ∞ √︂
 −𝑎𝑥 −𝑎𝑥 2 𝑎
ℱ𝑐 𝑒 = 𝑒 cos ω𝑥 d𝑥 = · .
0 π 𝑎 2 + ω2

On the other hand, separating the variables in the differential equation

√︂
𝑑𝐼 2 𝑎
= · 2
dω π 𝑎 + ω2

and then integrating, we get the general solution

√︂ ω
2
𝐼= · tan−1 + 𝐴.
π 𝑎

12
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Sine and Cosine Transform Professor of Mathematics

Then writing ω = 0 in this so that

√︂ √︂ ∫ ∞
𝑒 −𝑎𝑥
 
2 2
· tan−1 0 + 𝐴 = sin 0 d𝑥 = 0
π π 0 𝑥

√︃
· tan−1
2 ω


or 𝐴 = 0. Thus 𝐹𝑠 (ω) = π 𝑎 ·

2.2 Parseval’s Identities

Theorem 2.2.1 (Parseval’s Identity for Complex Fourier Transform).

∫ ∞ ∫ ∞
( 𝑓 (𝑥)) 2 d𝑥 = |𝐹 (ω)| 2 dω (2.2.1)
−∞ −∞

Proof. By definition, we note that

∫ ∞ ∫ ∞  ∞ 
1∫
( 𝑓 (𝑥)) 2 d𝑥 = 𝑓 (𝑥) √ 𝐹 (ω) 𝑒 𝑖ω𝑥 dω d𝑥
−∞ −∞ 2π −∞
∫ ∞  ∞ 
1∫
= 𝐹 (ω) √ 𝑓 (𝑥) 𝑒 𝑖ω𝑥 d𝑥 dω
−∞ 2π −∞
∫ ∞  ∞ 
1∫ −𝑖ω𝑥
= 𝐹 (ω) √ 𝑓 (𝑥) 𝑒 d𝑥 dω
−∞ 2π −∞
∫ ∞  ∞ 
1∫ −𝑖ω𝑥
= 𝐹 (ω) √ 𝑓 (𝑥) 𝑒 d𝑥 dω
−∞ 2π −∞
∫ ∞ ∫ ∞
= 𝐹 (ω) 𝐹 (ω) dω = |𝐹 (ω)| 2 dω
−∞ −∞

This completes the proof. 

Theorem 2.2.2 (Parseval’s Identity for Sine and Cosine Transforms).

∫ ∞ ∫ ∞
(𝑎) 𝑓 (𝑥) 𝑔(𝑥) d𝑥 = 𝐹𝑠 (ω) 𝐺 𝑠 (ω) dω and
0 0
∫ ∞ ∫ ∞
| 𝑓 (𝑥)| 2 d𝑥 = |𝐹𝑠 (ω)| 2 dω, in particular,
0 0
∫ ∞ ∫ ∞
(𝑏) 𝑓 (𝑥) 𝑔(𝑥) d𝑥 = 𝐹𝑐 (ω) 𝐺 𝑐 (ω) dω and
0 0
∫ ∞ ∫ ∞
| 𝑓 (𝑥)| 2 d𝑥 = |𝐹𝑐 (ω)| 2 dω, in particular.
0 0

∫ ∞
The integral |ℎ(𝜉)| 2 𝑑𝜉 of the squared magnitude of a function ℎ(𝜉) 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.
√︃
Example 2.2.1. Given that ℱ𝑐 𝑒−𝑎𝑥 = π2 · ω2𝑎+𝑎2 , prove that


∫ ∞
d𝑥 π
= ·
0 (𝑥 2 + 𝑎 2 ) (𝑥 2 + 𝑏2 ) 2𝑎𝑏(𝑎 + 𝑏)

13
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Sine and Cosine Transform Professor of Mathematics

Solution. Let 𝑓 (𝑥) = 𝑒 −𝑎𝑥 , 𝑔(𝑥) = 𝑒 −𝑏𝑥 . Then

√︂
 2 𝑎
ℱ𝑐 𝑓 (𝑥) = · = 𝐹𝑐 (ω),
π ω2 + 𝑎 2
√︂
 2 𝑏
ℱ𝑐 𝑔(𝑥) = · 2 = 𝐺 𝑐 (ω).
π ω + 𝑏2

Hence by Parseval’s identity for cosine transform, we have

∫ ∞ ∫ ∞
𝐹𝑐 (ω) 𝐺 𝑐 (ω) dω = 𝑓 (𝑥) 𝑔(𝑥) d𝑥
0 0
∞ ∞
2∫ 𝑎𝑏 ∫ 1
⇒ · dω = 𝑒 −𝑎𝑥 𝑒 −𝑏𝑥 d𝑥 =
π 0 (ω2 + 𝑎 2 ) (ω2 + 𝑏 2 ) 0 𝑎+𝑏

∫ ∞
dω π
or = ·
0 (ω2 + 𝑎 2 ) (ω2 + 𝑏2 ) 2𝑎𝑏(𝑎 + 𝑏)

Changing ω to 𝑥, this gives

∫ ∞
d𝑥 π
= ·
0 (𝑥 2 + 𝑎 2 ) (𝑥 2 + 𝑏 2 ) 2𝑎𝑏(𝑎 + 𝑏)

Example 2.2.2. Given that

√︂ √︂
 −𝑎𝑥 2 𝑎  2 sin ω𝑎
ℱ𝑐 𝑒 = · 2 and ℱ𝑐 𝑔(𝑥) = ·
π ω +𝑎 2 π ω

where

(0 < 𝑥 < 𝑎)

1


𝑔(𝑥) =

 0, elsewhere,


∞
π (1 − 𝑒 −𝑎 )
∫ 2
sin 𝑎𝑥 d𝑥
prove that = ·
0 𝑥(𝑥 2 + 𝑎 2 ) 2𝑎 2

Solution. From the Parseval’s identity for cosine transform, we have

∫ ∞ ∫ ∞
𝐹𝑐 (ω) 𝐺 𝑐 (ω) dω = 𝑓 (𝑥) 𝑔(𝑥) d𝑥
0 0
√︂ ∫ ∞
1 − 𝑒 −𝑎
∫ 𝑎 2
2 𝑎 sin ω𝑎
⇒ · dω = 𝑒 −𝑎𝑥 · 1 d𝑥 =
π 0 ω(ω2 + 𝑎 2 ) 0 𝑎2

∞
π (1 − 𝑒 −𝑎 )
∫ 2
sin ω𝑎 dω
or = ·
0 ω(ω + 𝑎 )
2 2 2𝑎 2

∞
π (1 − 𝑒 −𝑎 )
∫ 2
sin 𝑎𝑥 d𝑥
Changing ω to 𝑥, this gives = ·
0 𝑥(𝑥 + 𝑎 )
2 2 2𝑎 2

14
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Sine and Cosine Transform Professor of Mathematics

Example 2.2.3. Find the sine and cosine transforms of 𝑒−𝑥 , and use Parseval’s identities to prove that
∞ ∞
∫ d𝑥 π ∫ 𝑥 2 d𝑥
= = ·
0 (𝑥 2 + 1) 2 4 0 (𝑥 2 + 1) 2

Solution. Note that

√︂ ∫ ∞
2
𝐹𝑐 (ω) + 𝑖 𝐹𝑠 (ω) = · 𝑒 −𝑥 (cos ω𝑥 + 𝑖 sin ω𝑥) d𝑥
π 0
√︂ ∫ ∞ √︂ ∫ ∞
2 −𝑥 𝑖ω𝑥 2
= · 𝑒 𝑒 d𝑥 = · 𝑒 −(1−𝑖ω) 𝑥 d𝑥
π 0 π 0
√︂ ∞ √︂
2 𝑒 −(1−𝑖ω) 𝑥 2 1
= · − = ·
π 1 − 𝑖ω 𝑥=0 π 1 − 𝑖ω
√︂   ω 
2 1
= · + 𝑖 ·
π 1 + ω2 1 + ω2

Comparing the real and imaginary parts on both sides, we get

√︂ √︂
2 1 2 ω
𝐹𝑐 (ω) = · , 𝐹𝑠 (ω) = · ·
π 1 + ω2 π 1 + ω2

Then from Parseval’s identity for cosine transform, we get

∞ ∞ ∞
2∫ dω ∫ 1 ∫ dω π
= 𝑒 −2𝑥 d𝑥 = or = ,
π 0 (ω2 + 1) 2
0 2 0 (ω2 + 1) 2 4

from which the first result follows by replacing ω with 𝑥. Similarly, from Parseval’s identity for sine trans-
form, the second result follows.
√︃
Example 2.2.4. Given that ℱ 𝑓 (𝑥) =
 2 sin ω𝑎
π · ω = 𝐹 (ω)
where

(−𝑎 ≤ 𝑥 ≤ 𝑎)

1


𝑓 (𝑥) =

 0, elsewhere,


∞
∫ sin2 𝑎𝑥 d𝑥 π𝑎
prove that = ·
0 𝑥2 2

Solution. From the Parseval’s identity for the complex Fourier transform,

∞ ∞ ∞ 𝑎
∫ ∫ 2∫ sin2 ω𝑎 ∫
|𝐹 (ω)| 2 dω = ( 𝑓 (𝑥)) 2 d𝑥 ⇒ · dω = 1. d𝑥
−∞ −∞ π −∞ ω2 −𝑎
∞ ∞
∫ sin2 ω𝑎 ∫ sin2 ω𝑎 π𝑎
⇒ dω = π𝑎 ⇒ dω = ·
−∞ ω2 0 ω2 2

Replacing ω with 𝑥, the result follows.

Exercise 2.2.1. Find the sine transform of 𝑓 (𝑥) = 1

𝑥 ( 𝑥 2 +𝑎2 )
and hence the cosine transform of 𝑔(𝑥) =
1
𝑥 2 +𝑎2
·

15
BMAT102L - Differential Equations & Transforms Dr. T. Phaneendra
Module 6: Sine and Cosine Transform Professor of Mathematics

· (1 − 𝑒 −𝑎ω ), 𝐺 𝑐 (ω) = · 𝑒 −𝑎ω

√︁ π √︁ π
Ans. 𝐹𝑠 (ω) = 1
𝑎2 2
1
𝑎 2

Exercise 2.2.2. Find the cosine transform of 𝑓 (𝑥) = 1

𝑥 2 +1
and hence the sine transform of 𝑔(𝑥) = 𝑥
𝑥 2 +1
·

· 𝑒 −ω
√︁ π
Ans. 𝐹𝑐 (ω) = 𝐺 𝑠 (ω) = 2
√︃
4 sin2 (ω𝑎/2)
Exercise 2.2.3. If ℱ 𝑓 (𝑥) =
 2
π · ω2
= 𝐹 (ω)
where

 𝑎 − |𝑥| (−𝑎 ≤ 𝑥 ≤ 𝑎)



𝑓 (𝑥) =

 0,
 elsewhere,

∫ ∞  sin 𝑥  2 π ∫
∞ 
sin 𝑥
4
π
show that d𝑥 = and d𝑥 = ·
0 𝑥 2 0 𝑥 3

Exercise 2.2.4. Find the cosine transform of


 cos 𝑥, 0 ≤ 𝑥 < 𝑎,



𝑓 (𝑥) =

 0,
 𝑥 > 𝑎.


16