Fourier Analysis

A function is said to be periodic if f (x+t) = f(x), x; ∀𝑥 being real number

Definition: If a function
i) f(x) is defined on an interval (−𝜋, 𝜋)
ii) f(x) is periodic and
iii) f(x) is sectionally continuous on (−𝜋, 𝜋), then under these three conditions the Fourier series of
𝑎0
𝒇 (𝒙)𝑖𝑠 𝑔𝑖𝑣𝑒𝑛 𝑏𝑦 𝑓(𝑥) = + ∑(𝑎𝑛 𝑐𝑜𝑠 𝑥 + 𝑏𝑛 𝑠𝑖𝑛 𝑛 𝑥) 𝑤ℎ𝑒𝑟𝑒,
2
𝜋
1
𝑎0 = ∫ 𝑓(𝑥)𝑑𝑥
𝜋
−𝜋
fn → function
∀ → 𝐹𝑜𝑟 𝐴𝑙𝑙

𝜋
1
𝑎𝑛 = ∫ 𝑓(𝑥) cos 𝑛 𝑥 𝑑 𝑥
𝜋
−𝜋

𝜋
1
𝑏𝑛 = ∫ 𝑓(𝑥) sin 𝑛 𝑥 𝑑 𝑥
𝜋
−𝜋
Fourier co-efficient: If given fn is odd , an = 0
Odd f (x) = - f (x)
• If given f n is even, bn = 0
even f(-x)=f(x)

Find the Fourie’s series of f (x) = x2, - π < x < π

Soln: Fourier series of any function of the form
𝑎0
𝑓(𝑥) = + ∑(𝑎𝑛 𝑐𝑜𝑠 𝑥 + 𝑏𝑛 𝑠𝑖𝑛 𝑛 𝑥)
2

1 𝜋 2
𝑤ℎ𝑒𝑟𝑒, 𝑎0 = ∫ 𝑥 𝑑𝑥 𝑎
𝜋 −𝜋
∫ 𝑓(𝑥) 𝑑 𝑥
2 π −𝑎
𝑎
= ∫ x2d x
π 0 2 ∫ 𝑓(𝑥) 𝑑 𝑥 𝑤ℎ𝑒𝑛 𝑓(𝑥)𝑖𝑠 𝑒𝑣𝑒𝑛
={ 0
2π2 0; 𝑖𝑓 𝑓(𝑥)𝑖𝑠 𝑜𝑑𝑑
=
3 (ans.)

1 𝜋 2
𝑎𝑛 = ∫ 𝑥 cos 𝑛 𝑥 𝑑𝑥
𝜋 −𝜋
𝜋
2 𝜋 2 2 2 sin 𝑛 𝑥 sin 𝑛 𝑥
= ∫ 𝑥 cos 𝑛 𝑥 𝑑 𝑥 = [𝑥 − ∫ 2𝑥 ∙ 𝑑𝑥]
𝜋 0 𝜋 𝑛 𝑛 0
𝜋
2 𝑥2 𝑥 2 cos 𝑛 𝑥
= [( sin 𝑛𝜋) + (2 2 cos 𝑛 𝑥 + )]
𝜋 𝑛 𝑛 𝑛2 0
4 cos 𝑛 𝜋 4
= = (−1)𝑛 ∵ 𝑐𝑜𝑠 𝑛 𝜋 = (−1)𝑛
𝑛2 𝑛2
1 𝜋 2
𝑏𝑛 = ∫ 𝑥 sin 𝑛 𝑥 𝑑 𝑥 = 0
𝜋 −𝜋 since x2 sin n x
∴ The required Fourier Series is is ever function
 ∞
𝜋2 4
𝑓(𝑥) = + ∑ 2 (−1)𝑛 cos 𝑛 𝜋
3 𝑛
𝑛=1
2
𝜋 cos 2 𝑥 cos 3 𝑥 cos 4 𝑥
= + 4 (− cos 𝑥 + − + ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ )
3 22 32 42

𝝅 − 𝟒, −𝝅 ≤ 𝒙 < 𝟎
Find the Fourier series of 𝒇(𝒙) = {
𝝅 + 𝒙, 𝟎 ≤ 𝒙 ≤ 𝝅
Solution: Here, 𝑓(−𝑥) = 𝑓(𝑥), i.e. the given function is an even function
∞
𝑎0
𝐿𝑒𝑡 𝑡ℎ𝑒 𝑓𝑜𝑢𝑟𝑖𝑒𝑟 𝑠𝑒𝑟𝑖𝑒𝑠 𝑏𝑒 𝑓(𝑥) = + ∑(𝑎𝑛 cos 𝑛 x + 𝑏𝑛 sin 𝑛 𝑥)
2
𝑛=1
1 𝜋
Where, 𝑎0 + ∫ 𝑓(𝑥) 𝑑 𝑥
𝜋 −𝜋

1 0 𝜋
= [∫ (𝜋 − 𝑥)𝑑 𝑥 + ∫ (𝜋 + 𝑥)𝑑 𝑥 ]
𝜋 −𝜋 0

0 0
1 1 1
= [(𝜋𝑥 − 𝑥 2 ) + (𝜋𝑥 + 𝑥 2 ) ]
𝜋 2 −𝜋 2 −𝜋

1 2 1 2 1 3𝜋 2
= (𝜋 + 𝜋 + 𝜋 2 + 𝜋 2 ) = =3
𝜋 2 2 𝜋

1 𝜋
𝒂𝒏 = ∫ 𝑓(𝑥)𝑐𝑜𝑠 𝑛 𝑥 𝑑 𝑥
𝜋 −𝜋

𝜋
1
= [∫ (𝜋 − 𝑥) cos 𝑛 𝑥 𝑑 𝑥 + ∫ (𝜋 + 𝑥) cos 𝑛 𝑥 𝑑 𝑥 ]
𝜋 −𝜋 0

1 0 1 0 1 𝜋 1 𝜋
= ∫ 𝜋 cos 𝑛 𝑥 𝑑 𝑥 − ∫ 𝑥 cos 𝑛 𝑥 𝑑 𝑥 + ∫ 𝜋 cos 𝑛 𝑥 𝑑 𝑥 + ∫ 𝑥 cos 𝑛 𝑥 𝑑 𝑥
𝜋 −𝜋 𝜋 −𝜋 𝜋 0 𝜋 0

1 sin 𝑛 𝑥 0 1 x sin 𝑛 𝑥 cos 𝑛 𝑥 0 sin 𝑛 𝑥 𝜋 1 x sin 𝑛 𝑥 cos 𝑛 𝑥 𝜋

= 𝜋[ ] − [ + ] + [ ] + [ + ]
𝜋 𝑛 −𝜋 𝜋 𝑛 𝑛2 −𝜋 𝑛 0 𝜋 𝑛 𝑛2 0

1 cos 𝑛 𝜋 cos 𝑛 𝜋 1 1 1 1
= − 2
+ 2
+ 2
− 2
= [ 2 {(−1)𝑛 − 1} + 2 {(−1)𝑛 − 1}]
𝜋𝑛 𝜋𝑛 𝜋𝑛 𝜋𝑛 𝜋 𝑛 𝑛

1
= [(−1)𝑛 − 1]
𝑛2

and; bn = 0, since f(x) is an even function

The Fourier series is,
∞
3𝜋 2
𝑓(𝑥) = + ∑ 2 {(−1)𝑛 − 1} 𝑐𝑜𝑠 𝑛 𝑥
2 𝜋𝑛
𝑛=1
 Full range Fourier series

f (x)= x 2 ; 0 < 𝑥 < 2𝜋

we know,
∞
𝑎0
𝑓(𝑥) = + ∑ 𝑎𝑛 cos 𝑛 𝑥 + 𝑏 𝑛 sin 𝑛 𝑥
2
𝑛=1

1 2𝜋 2
𝑁𝑜𝑤, 𝑎0 = ∫ 𝑥 𝑑𝑥
𝜋 0
1 1 8𝜋 2
. . 8𝜋 3 =
𝜋 3 3
1 2𝜋 2
𝑎n = ∫ 𝑥 cos 𝑛 𝑥 𝑑 𝑥
𝜋 0
2𝜋
1 𝑥 2 sin 𝑛 𝑥 2 x cos 𝑛 𝑥 2𝜋 2 sin 𝑛 𝑥 2𝜋
= {[ ] +[ ] − [ ] }
𝜋 𝑛 0
𝑛2 0 𝑛3 0

1 4𝜋 2 sin 2 𝜋 𝑛 4𝜋 cos 2 𝜋 𝑛 2 sin 2 𝜋 𝑛

= [ + + ]
4 𝑛 𝑛2 𝑛3
4 4
= 2
cos 2 𝜋𝑛 = 2 cos 𝑛 (2𝜋)
𝑛 𝑛
Since the function is even, bn = 0
The Fourier series of f(x) is
∞
8π2 4
𝐹. 𝑆. = + ∑ 2 cos n (2π). cos n x
6 n
𝑛=1

Half Range Fourier Series:

If a function f(x) is defined on (0, π) we extend it on ( - π, 0) either as an ever function or as an odd function
i) If, f(x) is extended as an ever function then, bn = 0
2 𝜋
𝑎0 = ∫ 𝑓(𝑥)𝑑 𝑥
𝜋 0
2 𝜋
𝑎𝑛 = ∫ 𝑓(𝑥) cos 𝑛 𝑥 𝑑 𝑥
𝜋 0
∞
𝑎0
∴ Fourier series, 𝑓(𝑥) = + ∑ 𝑎𝑛 cos 𝑛 𝑥
2 [cosine series]
𝑛=1

ii) If, f(x) is extended as an odd function then, a0 = 0

2 𝜋
𝑏𝑛 = ∫ 𝑓(𝑥) sin 𝑛 𝑥 𝑑 𝑥
𝜋 0
an = 0
∞

∴ Fourier series, 𝑓(𝑥) = ∑ 𝑎𝑛 sin 𝑛 𝑥 ;

𝑛=1
it is a sine series
 𝝅
𝟏, 𝟎 < 𝒙 <
𝟐
Find the Fourier series of 𝒇 (𝒙) = { 𝝅
𝟎, < 𝒙 < 𝝅
𝟐
Solution: Here, f(x) is defined on (0, π) we extend this function on (-π, 0) as an even function. Then bn = 0 and the
F.S. is,
∞
𝑎0
𝑓(𝑥) = + ∑ 𝑎𝑛 cos 𝑛 𝑥
2
𝑛=1

2 𝜋
𝑊ℎ𝑒𝑟𝑒, 𝑎0 = ∫ 𝑓(𝑥) 𝑑 𝑥
𝜋 0

𝜋
2 2 2 𝜋
= [∫ 1 𝑑 𝑥 + 0] = . = 1
𝜋 0 𝜋 2

2 𝜋
𝑎𝑛 = ∫ 𝑓(𝑥) cos 𝑛 𝑥 𝑑 𝑥
𝜋 0

𝜋
2 2
= [∫ 1 cos 𝑛 𝑥 𝑑 𝑥 + 0]
𝜋 0

2 𝑛𝜋
= 𝑠𝑖𝑛
𝜋𝑛 2

 the required F.S is

∞
1 2 𝑛𝜋
𝑓(𝑥) = + ∑ sin cos 𝑛 𝑥
2 𝜋𝑛 2
𝑛=1

➢ Fourier series in any interval ( - l , l ): Suppose a function f (x) is defined on ( - l , l ). But we know that F.S.
can be found only for those functions defined in ( - π, π ). To find the required F.S. we change a variable
from x to y. So that f (x) changes to g (y) and as x varies from – l to l , y varies from – π to π
𝑥 𝑦
=
2𝑙 2𝜋
𝑥𝜋
∴𝑦=
𝑙
The F.S. of f (x) in ( - l , l ) is the same as the F.S. of g (y) in ( - π, π ). Now for the F.S. of g (y) in ( - π, π ):
1 𝜋
∴ 𝑎0 = ∫ 𝑔(𝑦)𝑑 𝑦 𝑥𝜋
𝜋 −𝜋 𝑦=
𝑙
1 𝑙 𝜋 𝜋
= ∫ 𝑓(𝑥) 𝑑𝑥 ∴ 𝑑𝑦= 𝑑𝑥
𝜋 −𝑙 𝑙 𝑙

1 𝑙
∴ 𝑎0 = ∫ 𝑓(𝑥) 𝑑 𝑥
𝑙 −𝑙

1 𝜋
𝑎𝑛 = ∫ 𝑔(𝑦) cos 𝑛 𝑦 𝑑 𝑦
𝜋 −𝜋
 1 𝑙 nπx 𝜋
= ∫ 𝑓(𝑥) cos ( ) ∙ 𝑑𝑥
𝜋 −𝑙 l 𝑙

1 𝑙 nπx
∴ 𝑎𝑛 = ∫ 𝑓(𝑥) cos ( )𝑑 𝑥
𝑙 −𝑙 l

1 𝜋
𝑏𝑛 = ∫ 𝑔(𝑦) sin 𝑛 𝑦 𝑑 𝑦
𝜋 −𝜋

1 𝑙 nπx 𝜋
= ∫ 𝑓(𝑥) sin ( ) ∙ 𝑑𝑥
𝜋 −𝑙 l 𝑙

1 𝑙 nπx
∴ 𝑏𝑛 = ∫ 𝑓(𝑥) sin ( )𝑑 𝑥
𝑙 −𝑙 l

∞
𝑎𝑜 𝑛𝜋𝑥 𝑛𝜋𝑥
∴ 𝑇ℎ𝑒 𝐹. 𝑆. = + ∑(𝑎𝑛 cos + 𝑏𝑛 𝑠𝑖𝑛 )
2 𝑙 𝑙
𝑛=1

𝑑𝑢
∗ ∫𝑢𝑣 𝑑 𝑥 = 𝑢∫𝑣 𝑑 𝑥 − ∫[ ∫ 𝑣 𝑑 𝑥] 𝑑 𝑥
𝑑𝑥

𝑢
∗ ∫ 𝑑𝑥=
𝑣

Fourier Transform
∞
1
𝐹[𝑓(𝑥)] = 𝐹 (∝) = ∫ 𝑓(𝑡)𝑒 𝑖∝𝑡 𝑑 𝑡
√2𝜋 −∞

∞
𝑜𝑟, 𝐹[𝑓(𝑥)] = 𝐹 (∝) = ∫ 𝑓(𝑥)𝑒 𝑖∝𝑥 𝑑 𝑥;
−∞
𝑥2
−
𝐹𝑖𝑛𝑑 𝑡ℎ𝑒 𝐹𝑜𝑢𝑟𝑖𝑒𝑟 𝑡𝑟𝑎𝑛𝑠𝑓𝑜𝑟𝑚 𝑜𝑓𝑓(𝑥) = 𝑒 2

Solution: From the definition of F.T. we have,

∞
1
𝐹[𝑓(𝑥)] = ∫ 𝑓(𝑥)𝑒 𝑖∝𝑥 𝑑 𝑥
√2𝜋 −∞ 𝑥2
− +𝑖 ∝𝑥
2 2
1 ∞ −𝑥
= ∫ 𝑒 2 𝑒 𝑖∝𝑥 𝑑 𝑥 1 1
√2𝜋 −∞ = (𝑥 2 − 2𝑖 ∝ 𝑥 + (𝑖 ∝)2 + (𝑖 ∝)2
2 2
∝2
1 ∞ −1(𝑥−𝑖∝)2 1 1
√2𝜋
∫−∞
𝑒 2 𝑒2 𝑑𝑥 = − (𝑥 − 𝑖 ∝)2 + (−∝2 )
2 2

Let, 1 ∝2
= − (𝑥 − 𝑖 ∝)2 −
𝑥−𝑖∝ 2 2
=𝑡
√2
 𝑑𝑥
⇒ = 𝑑𝑡
√2

∝2
𝑒− 2 ∞
2
𝐹[𝑓(𝑥)] = ∫ 𝑒 −𝑡 𝑑 𝑡 t2 = z
√𝜋 −∞
2tdt=dz
∝2
−
𝑒 2 ∞ 1 1
= ∫−∞ 𝑒 −𝑧 2 𝑑𝑧 𝑑𝑡 = 𝑑𝑧
√𝜋 √𝑧 2𝑡
∝2
1
−
𝑒 2 ∞ 1
= 𝑑𝑧
= 2 ∫0 𝑒 −𝑧 𝑑𝑧 2√𝑧
√𝜋 2√𝑧

∝2
− 1
𝑒 2 ∞ −
= ∫0 𝑒 −𝑧 𝑧 2 𝑑 𝑧
√𝜋

∝2
− 1
𝑒 2 ∞ −1
= ∫0 𝑒 −𝑧 𝑧 2 𝑑 𝑧
√𝜋

∝2 ∝2
− −
∝2
𝑒 1
2 𝑒 2
= .√ = . √𝜋 = 𝑒 − 2
√𝜋 2 √𝜋

❖ Find the F.T. of 𝒇(𝒙) = 𝒆−𝒂|𝒙| , - ∞ < x < ∞

−𝑥, 𝑥 < 0
Solution: We know that, |𝑥| = {
𝑥, 𝑥 ≥ 0

From the definition of F.T. we have,

∞
1
𝐹[𝑓(𝑥)] = ∫ 𝑓(𝑥) 𝑒 𝑖∝𝑛 𝑑 𝑥
√2𝜋 −∞

0 ∞
1 1
= ∫ 𝑒 𝑎𝑥 𝑒 𝑖∝𝑛 𝑑 𝑥 + ∫ 𝑒 −𝑎𝑥 𝑒 𝑖∝𝑛 𝑑 𝑥
√2𝜋 −∞ √2𝜋 0

0 ∞
1 1
= ∫ 𝑒 (𝑎+𝑖∝)𝑥 𝑑 𝑥 + ∫ 𝑒 (−𝑎+𝑖∝)𝑥 𝑑 𝑥
√2𝜋 −∞ √2𝜋 0

0 ∞
1𝑒 (𝑎+𝑖∝)𝑥 1 𝑒 (−𝑎+𝑖∝)𝑥
= [ ] + [ ]
√2𝜋 𝑎 + 𝑖 ∝ −∞ √2𝜋 − 𝑎 + 𝑖 ∝ 0

1 1 1 1
= . [1 − 0] + . [ 0 − 1]
√2𝜋 𝑎 + 𝑖 ∝ √2𝜋 (−𝑎 + 𝑖 ∝)

1 1 1 1 𝑎 − 𝑖 ∝ +𝑎 + 𝑖 ∝ 1 2𝑎 √2𝑎
= [ + ]= [ 2 2 2 ]= 2 2
=
√2𝜋 𝑎 + 𝑖 ∝ 𝑎 − 𝑖 ∝ √2𝜋 𝑎 − (𝑖 ∝ ) √2𝜋 𝑎 + ∝ √𝜋(𝑎2 +∝2 )
Gamma Function:
∞
𝑇ℎ𝑒 𝐺𝑎𝑚𝑚𝑎 𝑓 𝑛 𝑖𝑠 𝑑𝑒𝑛𝑜𝑡𝑒𝑑 𝑏𝑦 𝜞𝒏 𝑎𝑛𝑑 𝑖𝑠 𝑑𝑒𝑓𝑖𝑛𝑒𝑑 𝑏𝑦 𝜞𝒏 = ∫ 𝒆−𝒙 𝒙𝒏−𝟏 𝒅 𝒙 ; 𝒏 > 𝟎
𝟎
Beta Function:

𝑇ℎ𝑒 𝑏𝑒𝑡𝑎 𝒇𝒏 𝑖𝑠 𝑑𝑒𝑛𝑜𝑡𝑒𝑑 𝑏𝑦 𝜷 ( 𝒎 , 𝒏 ) 𝑎𝑛𝑑 𝑖𝑠 𝑑𝑒𝑓𝑖𝑛𝑒𝑑 𝑏𝑦,

𝟏
𝛽 ( 𝑚 , 𝑛 ) = ∫ 𝒙𝒎−𝟏 (𝟏 − 𝒙)𝒏−𝟏 𝒅 𝒙 ; 𝒎 > 𝟎, 𝒏 > 𝟎
𝟎

Properties of Gamma and Beta:

1. 𝚪 𝒏 + 𝟏 = 𝒏 𝚪𝒏 6. 𝜷 ( 𝒎 , 𝒏 ) = 𝛃 ( 𝐧 , 𝐦 ) ∞ 𝒚𝒏−𝟏
8. 𝜷 ( 𝒎 , 𝒏 ) = ∫𝟎 (𝟏+𝒚)𝒎+𝒏
𝒅𝒚
𝝅
2. 𝚪 𝒏 + 𝟏 = 𝒏 !
7. ∫𝟎 𝒔𝒊𝒏𝒑 𝜽 𝒄𝒐𝒔𝒒 𝜽 𝒅 𝜽
𝟐 ∞
𝒚 𝒎− 𝟏
𝟏 =∫ 𝒅𝒚
3. 𝚪 𝟐 = √𝝅 𝟎 (𝟏 + 𝒚)𝒎+𝒏
𝒑+𝟏 𝒒+𝟏
𝚪 𝟐 𝟐
=
4. 𝚪𝟏 = 𝟏 𝒑+𝒒+𝟐
𝟐𝚪 𝟐
𝚪𝒎𝚪𝒏
5. 𝜷 ( 𝒎 , 𝒏 ) = 𝚪 𝒎+𝒏

∞
𝟐 √𝝅
𝑷𝒓𝒐𝒗𝒆 𝒕𝒉𝒂𝒕, ∫ 𝒆−𝒕 𝒅 𝒕 =
𝟎 𝟐

Let, t2 = x
t ∞ o
x ∞ o
⇒2𝑡𝑑𝑡 =𝑑𝑥

⇒𝑑𝑡=𝑑𝑥

1
⇒𝑑𝑡= 𝑑𝑥
2𝑡

1
= 𝑑𝑥
2√𝑥

∞
1 1 ∞ 1 1 ∞ 1
𝐼 = ∫ 𝑒 −𝑥 𝑑 𝑥 = ∫ 𝑒 −𝑥 𝑥 − 2 𝑑 𝑥 = ∫ 𝑒 −𝑥 𝑥 2 − 1 𝑑 𝑥
0 2√𝑥 2 0 2 0

1 1 1 √𝜋
= Γ = √𝜋 =
2 2 2 2 (proved)

∞ 𝟏 5 𝟑 𝟏
𝟐
∫ 𝒆−𝒚 𝒚𝟓 𝒅 𝒚 ∫ x 2 (𝟏 − 𝒙)𝟐 𝒅 𝒙 ∫ 𝒙𝟕 (𝟏 − 𝒙)𝟑 𝒅 𝒙
𝟎 𝟎 𝟎

∞ 𝝅
∫ 𝒙𝟓 𝒆− 𝟒𝒙 𝒅 𝒙 𝟐
𝟎
∫ 𝒔𝒊𝒏𝟒 𝜽 𝒄𝒐𝒔𝟓 𝜽 𝒅𝜽
𝟎
 Fourier analysis & special functions

𝝅 𝒑+𝟏 𝒒+𝟎
𝟐 𝜞 𝟐 𝜞 𝟐
𝒑 𝒒
𝑷𝒓𝒐𝒗𝒆 𝒕𝒉𝒂𝒕, ∫ 𝒔𝒊𝒏 𝜽 𝒄𝒐𝒔 𝜽 𝒅 𝜽 =
𝒑+𝒗+𝟐
𝟎 𝟐𝜞
𝟐
𝜋
2
𝑆𝑜𝑙: 𝐼 = ∫ 𝑠𝑖𝑛𝑝 𝜃 𝑐𝑜𝑠 𝑞 𝜃 𝑑 𝜃
0

𝜋
2
= ∫ sinp−1 𝜃 cos q−1 𝜃 sin 𝜃 cos 𝜃 𝑑 𝜃
0

𝜋
2 𝑝−1 𝑞−1
= ∫ (sin2𝜃) 2 (cos 2 𝜃) 2 sin 𝜃 cos 𝜃 𝑑𝜃
0

𝜋
2 𝑝−1 𝑞−1
= ∫ (sin2𝜃) 2 (1 − sin2 𝜃) 2 sin 𝜃 cos 𝜃 𝑑𝜃
0

Let, Z = sin2 ϴ => d z = 2 sin ϴ cos ϴ d ϴ

When, ϴ = 0 => z = sin2 0 = 0

𝜋 𝜋 2
𝛳 = => 𝑧 = ( 𝑠𝑖𝑛 ) = 1
2 2
1 𝑝−1 𝑞+1 1 1 1 (𝑝+1)−1 𝑞+1
∴𝐼= ∫ 𝑧 2 (1 − 𝑧) 2 𝑑𝑧= ∫ 𝑧 2 (1 − 𝑧)( 2 )−1 𝑑 𝑧
0 2 2 0

𝑝+1 𝑞+1
1 𝑝+1 𝑞+1 1 Γ 2 Γ 2
= 𝛽( , )=
2 2 2 2 2 Γ 𝑝 + 𝑞 + 2 (proved)
2

𝜋
2
# ∫ sin6 𝑥 𝑑 𝑥
0

Orthogonal Functions

Legendre Polynomials: Legendre differential equation is

𝑑2 𝑦 𝑑𝑦
(1 − 𝑥 2 ) − 2𝑥 + 𝑛(𝑛 + 1) 𝑦 = 0
𝑑𝑥 2 𝑑𝑥

𝑊ℎ𝑒𝑟𝑒 𝒏 𝑖𝑠 𝑎 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 𝑖𝑛𝑡𝑒𝑔𝑒𝑟𝑥 2 𝑦” + 𝑥𝑦 ′ + (𝑥 2 − 𝑛2 ) 𝑦 = 0

Legendre Polynomial is
 Fourier analysis & special functions
𝑛
[ ]
2
(1)𝑟 (2𝑛 − 2𝑟)
𝑃𝑛 (𝑥) = ∑
2𝑛 𝑟! (𝑛 − 𝑟)! (𝑛 − 2𝑟)!
𝑟=0

𝑛
, 𝑤ℎ𝑒𝑛 𝑛 𝑖𝑠 𝑒𝑣𝑒𝑛
𝑛 2
𝑊ℎ𝑒𝑟𝑒, [ ] = 𝑛 − 1
2 , 𝑤ℎ𝑒𝑛 𝑛 𝑖𝑠 𝑜𝑑𝑑
{ 2

1 𝑑𝑛
𝑃𝑛 (𝑥) = ∙ (𝑥 2 − 1); 𝑤ℎ𝑒𝑟𝑒 𝑃𝑛 (𝑥)𝑖𝑠 𝐿𝑒𝑔𝑒𝑛𝑑𝑟𝑒 𝑝𝑜𝑙𝑦𝑛𝑜𝑚𝑖𝑎𝑙
2𝑛 ∙𝑛! 𝑑𝑥 𝑛

Proof of Rodrigue’s formuls: Let, 𝑣 = (𝑥 2 − 1)𝑛 ………………………………(I)

∴ 𝑣1 = 𝑛(𝑥 2 − 1)𝑛−1 ∙ 2𝑥

(𝑥 2 − 1)𝑛
⇒ 𝑣1 = 2 𝑛 𝑥
(𝑥 2 − 1)

(𝑥 2 − 1)𝑛
⇒ 𝑣1 = 2 𝑛 𝑥
(𝑥 2 − 1)

⇒ (𝑥 2 − 1)𝑣1 = 2 𝑛 𝑥 (𝑥 2 − 1)𝑛

⇒ (𝑥 2 − 1)𝑣1 = 2 𝑛 𝑥 𝑣

Differentiating both sides with respect to x,

(𝑥 2 − 1)𝑣2 + 2 𝑥 𝑣1 = 2𝑛 (𝑥𝑣1 + 𝑣)

⇒ (𝑥 2 − 1)𝑣2 + (2 − 2𝑛) 𝑥 𝑣1 − 2 𝑛 𝑣 = 0

Now applying Leibnitz’s theorem;

Leibnitz’s theorem:

Let u and v be two functions of x then the n-th derivative of their product given by

(𝑢𝑣)𝑛 = 𝑢𝑣𝑛 + 𝑛𝑐1 𝑢1 𝑉𝑛−1 + 𝑛𝑐2 𝑢2 𝑉𝑛−2 + ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ +𝑈𝑛 𝑉 ∙∙

Where the suffixes in u & v denote the order of differentiation w.r to x.

(𝑥 2 − 1)𝑣𝑛+1 + 𝑛𝑐1 (2𝑥) 𝑣𝑛+1 + 𝑛𝑐2 (2)𝑣𝑛 + (2 − 2𝑛)(𝑥𝑣𝑛+1 + 𝑛𝑐1 (1)𝑣𝑛 ) − 2𝑛 𝑣𝑛 = 0

𝑛2 − 𝑛
⇒ (𝑥 2 − 1)𝑣𝑛+2 + 2𝑛𝑥 𝑣𝑛+1 + ∙ 2𝑣𝑛 + (2 − 2𝑛 ) 𝑥 𝑣𝑛+1 + (2 − 2𝑛) 𝑛 𝑣𝑛 − 2𝑛 𝑣𝑛 = 0
2

⇒ (𝑥 2 − 1)𝑣𝑛+2 + (2𝑛 + 2 − 2𝑛) 𝑥 𝑣𝑛+1 + (𝑛2 − 𝑛 + 2𝑛 − 2𝑛2 − 2𝑛) 𝑣𝑛 = 0

⇒ (𝑥 2 − 1)𝑣𝑛+2 + 2𝑥 𝑣𝑛+1 − 𝑛(𝑛 + 1)𝑣𝑛 = 0

𝑑𝑛+2 𝑑𝑛+1 𝑣 𝑑𝑛 𝑣
⇒ (1 − 𝑥 2 ) − 2𝑥 + 𝑛 (𝑛 + 1) =0
𝑑𝑥 𝑛+2 𝑑𝑥 𝑛+1 𝑑𝑥 𝑛
 Fourier analysis & special functions

𝑑2 𝑑𝑛 𝑣 𝑑 𝑑𝑛 𝑣 𝑑𝑛 𝑣
⇒ (1 − 𝑥 2 ) ( ) − 2𝑥 ( ) + 𝑛 (𝑛 + 1) =0
𝑑𝑥 2 𝑑𝑥 𝑛 𝑑𝑥 𝑑𝑥 𝑛 𝑑𝑥 𝑛

𝑑2 𝑣𝑛 𝑑 𝑣𝑛
⇒ (1 − 𝑥 2 ) − 2𝑥 + 𝑛 (𝑛 + 1)𝑣𝑛 = 0 ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ (2)
𝑑𝑥 2 𝑑𝑥

In (2) we see that Legendre differential equation is satisfied by vn . But we know, Pn (x) is s solution of Legendre
equation

𝑑𝑛𝑣
 Pn (x) must be a constant multiple of vn; i.e. 𝑃𝑛 (𝑥) = 𝑐 𝑣𝑛 = 𝑐 𝑑𝑥 𝑛

𝑑𝑛
⇒ 𝑝𝑛 (𝑥) = 𝑐 (𝑥 2 − 1)𝑛 ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ (3)
𝑑𝑥 𝑛

⇒ 𝑝𝑛 (𝑥) = 𝑐[(𝑥 2 − 1)𝑛 ]𝑛

= 𝑐[(𝑥 + 1)𝑛 {(𝑥 − 1)𝑛 }𝑛 + 𝑛𝑐1 {(𝑥 + 1)𝑛 } {(𝑥 − 1)𝑛 }𝑛−1 +∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ +(𝑥 − 1)𝑛 {(𝑥 + 1)𝑛 }𝑛 ]

𝑛!
= 𝑐 [(𝑥 + 1)𝑛 ∙ 𝑛 ! + 𝑛 ∙ (𝑥 − 1)𝑛(𝑥 + 1)𝑛−1 +∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ +(𝑥 − 1)𝑛 𝑛! ]
1!

Now putting x = 1, we get, 𝑃𝑛 (1) = 𝑐[𝑛! 2𝑛 + 0]

𝑛]
⇒ 1 = 𝑐[𝑛! 2
Dn (ax+b)m
1 𝑚!(𝑎𝑥+𝑏)𝑚−𝑛
∴𝑐= = 𝑎𝑛 (𝑚−𝑛)!
𝑛
2 𝑛!

1 𝑑𝑛
⇒ 𝑃𝑛 (𝑥) = 𝑛
(𝑥 2 − 1)𝑛 [Proved]
2 𝑛! 𝑑𝑥 𝑛

Recurrence relation for Pn (x):

1. (𝑛 + 1) 𝑃𝑛+1 (𝑥) = (2𝑛 + 1)𝑥 𝑃𝑛 (𝑥) − 𝑛 𝑃𝑛−1 (𝑥)

Proof: From the generating function of the Legendre polynomial Pn (x), we have

∞
1
2 )−2
(1 − 2𝑥ℎ + ℎ = ∑ ℎ𝑛 𝑃𝑛 (𝑥) … … … … . (1)
𝑛=0

∞
3
𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡𝑖𝑎𝑡𝑖𝑛𝑔 (1)𝑤. 𝑟. 𝑡𝑜 ℎ, 𝑤𝑒 𝑔𝑒𝑡 (1 − 2𝑥ℎ + ℎ2 )−2 (−2𝑥 + 2ℎ) = ∑ 𝑛 ℎ𝑛−1 𝑃𝑛 (𝑥)
𝑛=0

1 ∞
(1 − 2𝑥ℎ + ℎ2 )−2
⇒ ∙ (𝑥 − ℎ) = ∑ 𝑛 ℎ𝑛−1 𝑃𝑛 (𝑥)
(1 − 2𝑥ℎ + ℎ2 )
𝑛=0

∞
1
⇒ (1 − 2𝑥ℎ + ℎ2 )−2 (𝑥 − ℎ) = (1 − 2𝑥ℎ + ℎ 2)
∑ 𝑛 ℎ𝑛−1 𝑃𝑛 (𝑥)
𝑛=0

∞ ∞
𝑛 2)
⇒ ∑ ℎ 𝑃𝑛 (𝑥) ∙ (𝑥 − ℎ) = (1 − 2𝑥ℎ + ℎ ∑ 𝑛 ℎ𝑛−1 𝑃𝑛 (𝑥)
𝑛=0 𝑛=0
 Fourier analysis & special functions
∞ ∞ ∞ ∞ ∞
𝑛 𝑛+1 𝑛−1
⇒ ∑ 𝑥 ℎ 𝑃𝑛 (𝑥) − ∑ ℎ 𝑃𝑛 (𝑥) = ∑ 𝑛 ℎ 𝑃𝑛 (𝑥) ∑ 2 𝑥 𝑛 ℎ 𝑃𝑛 (𝑥) + ∑ 𝑛 ℎ𝑛+1 𝑃𝑛 (𝑥)
𝑛

𝑛=0 𝑛=0 𝑛=0 𝑛=0 𝑛=0

Now equating the co-efficient of hn from both sides,

𝑥 𝑝𝑛 (𝑥) − 𝑛 𝑃𝑛−1 (𝑥) = (𝑛 + 1) 𝑃𝑛+1 (𝑥) − 2𝑛𝑥 𝑃𝑛 (𝑥)

⇒ (2𝑛 + 1)𝑥 𝑝𝑛 (𝑥) − (𝑛 + 1) 𝑃𝑛+1 (𝑥) = 𝑛 𝑃𝑛−1 (𝑥)

∴ (𝑛 + 1) 𝑝𝑛+1 (𝑥) = (2𝑛 + 1)𝑥 𝑃𝑛 (𝑥) − 𝑛 𝑃𝑛−1 (𝑥)

[Proved]

2. 𝑷𝒓𝒐𝒗𝒆 𝑻𝒉𝒂𝒕, (𝟐𝒏 + 𝟏) 𝑷𝒏 (𝒙) = 𝑷′𝒏+𝟏 (𝒙) − 𝑷′𝒏−𝟏 (𝒙)

Proof: From Rodrigues Formula for

Pn (x), we have

1 𝑑𝑛 1
𝑃𝑛 (𝑥) = 𝑛 𝑛
(𝑥 2 − 1)𝑛 = 𝑛 ∙ 𝐷 𝑛 (𝑥 2 − 1)𝑛
2 ∙ 𝑛! 𝑑𝑥 2 ∙ 𝑛!

Replacing n by (n+1), we get

1
𝑃𝑛+1 (𝑥) = 𝐷 𝑛 (𝑥 2 − 1) 𝑛 + 1
2𝑛+1 (𝑛 + 1)!

′
1
𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡𝑖𝑎𝑡𝑖𝑛𝑔 𝑤. 𝑟. 𝑡𝑜 𝒙, 𝑤𝑒 𝑔𝑒𝑡 𝑃𝑛+1 (𝑥) = 𝐷 𝑛+1 {2𝑥(𝑛 + 1)(𝑥 2 − 1)𝑛 }
2𝑛+1 (𝑛 + 1)!
1
= ∙ 𝐷 𝑛+1 {𝑥(𝑥 2 − 1)𝑛 }
2𝑛 ∙ 𝑛!

1
= ∙ 𝐷 𝑛 {(𝑥 2 − 1)𝑛 + 2𝑛𝑥 2 𝑥(𝑥 2 − 1)𝑛−1 }
2𝑛 ∙ 𝑛!

1 2𝑛
= ∙ 𝐷 𝑛 (𝑥 2 − 1)𝑛 + 𝑛 𝐷 𝑛 [𝑥 2 (𝑥 2 − 1)𝑛−1
2𝑛 ∙ 𝑛! 2 ∙ 𝑛!

2𝑛
= 𝑃𝑛 (𝑥) + 𝐷 𝑛 [{(𝑥 2 − 1) + 1}(𝑥 2 − 1)𝑛−1 ]
2𝑛 ∙ 𝑛!

′ (𝑥)
2𝑛
∴ 𝑃𝑛+1 − 𝑃𝑛 (𝑥) = 𝐷 𝑛 {(𝑥 2 − 1)𝑛 + (𝑥 2 − 1)𝑛−1 }
2𝑛 ∙ 𝑛!

2𝑛 2𝑛
= 𝑑𝑛 (𝑥 2 − 1)𝑛 + 𝑛 𝐷 𝑛 (𝑥 2 − 1)𝑛−1
2𝑛∙ 𝑛! 2 ∙ 𝑛!

1
= 2𝑛 𝑃𝑛 (𝑥) + 𝐷 𝑛 (𝑥 2 − 1)𝑛−1
2𝑛−1 ∙ (𝑛 − 1)!
1
= 2𝑛 𝑃𝑛(𝑥) + 𝐷 { 𝐷 𝑛−1 (𝑥 2 − 1)𝑛−1 }
2𝑛−1 (𝑛 − 1)!

= 2𝑛 𝑃𝑛 (𝑥) + 𝐷 𝑃𝑛−1 (𝑥)

 Fourier analysis & special functions

′ (𝑥)
= 2𝑛 𝑃𝑛 (𝑥) + 𝑃𝑛−1

′ ′ (𝑥)
∴ (2𝑛 + 1) 𝑃𝑛 (𝑥) = 𝑃𝑛−1 (𝑥) − 𝑃𝑛−1

3. 𝑷𝒓𝒐𝒗𝒆 𝒕𝒉𝒂𝒕, 𝑷′𝒏+𝟏 (𝒙) + 𝑷′𝒏−𝟏 (𝒙) = 𝟐 𝒙 𝑷′𝒏 (𝒙) + 𝑷𝒏 (𝒙)

Proof: From the generating function for Pn (x), we get,

∞
1
2 )−2
(1 − 2𝑥ℎ + ℎ = ∑ ℎ𝑛 𝑃𝑛 (𝑥)
𝑛=0

Differentiating both sides w.r. to x,

∞
1 3
− (1 − 2𝑥ℎ + ℎ2 )−2 (−2ℎ) = ∑ ℎ𝑛 𝑃′ 𝑛 (𝑥)
2
𝑛=0

∞
1
2 )−2 2)
⇒ ℎ(1 − 2𝑥ℎ + ℎ = (1 − 2𝑥ℎ + ℎ ∑ ℎ𝑛 𝑃′ 𝑛 (𝑥)
𝑛=0

∞ ∞ ∞ ∞
𝑛+1 𝑛 ′ 𝑛+1
⇒ ∑ℎ 𝑃𝑛 (𝑥) = ∑ ℎ 𝑃 𝑛 (𝑥) − ∑ 2𝑥ℎ 𝑃 𝑛 (𝑥) + ∑ ℎ𝑛+2 𝑃′ 𝑛 (𝑥)
′

𝑛=0 𝑛=0 𝑛=0 𝑛=0

now equating co-efficient of hn+1, Pn (x) = P’n+1 (x) – 2x P’n (x) + P’n-1 (x)

 P’n+1 (x) + P’n-1 (x) = Pn (x) + 2 x P’n (x)

[Proved]

Beta-Gamma Functions
1
Definition: The Beta Function denoted by β ( m , n ) is defined as 𝛽 ( 𝑚 , 𝑛 ) = ∫0 𝑥 𝑚−1 (1 − 𝑥)𝑛−1 𝑑 𝑥 ,

Where m > 0 , n > 0 ……………………(1)

Which is convergent for positive m and n, Evidently β ( m , n ) is symmetric in m and n for the transformation x = 1 –
y, that is β ( m , n ) = β ( n , m )

Thus substituting x=1–y in (1) we get,

𝑥=0 𝑥=1
} }𝑑 𝑥 = − 𝑑 𝑦
𝑦=1 𝑦=0

0
∴ 𝛽 ( 𝑚 , 𝑛 ) = ∫ (1 − 𝑦)𝑚−1 𝑦 𝑛−1 ∙ − 𝑑 𝑦
1

1
= ∫ 𝑦 𝑛−1 (1 − 𝑦)𝑚−1 𝑑 𝑦 = 𝛽 ( 𝑛 , 𝑚 )
0
 Fourier analysis & special functions

The Gamma function denoted by Γn is defined as

∞
Γ𝑛 = ∫ 𝑒 −𝑥 𝑥 𝑛−1 𝑑 𝑥, 𝑤ℎ𝑒𝑟𝑒 𝑛 > 0 … … … … … (2)
0

Thus Integral converges for n > 0 . The restriction n > 0 is imposed because the integral diverges at the lower limit
for other values of n

∞
𝐼𝑛 𝑃𝑎𝑟𝑡𝑖𝑐𝑢𝑙𝑎𝑟, Γ1 = ∫ 𝑒 −𝑥 𝑑 𝑥 = [−𝑒 −𝑥 ]∞
0 =1
1

Relation between beta and gamma functions:

There is a very useful formula which expresses the beta function in terms of gamma function. From definition, we
have,

∞
Γ𝑚 = ∫ 𝑒 −𝑡 𝑡 𝑚−1 𝑑𝑡
0

Putting t = x2 so that dt = 2 x d x

∞
2
𝑇ℎ𝑒𝑛 Γ𝑚 = 2 ∫ 𝑒 −𝑥 𝑥 2𝑚−1 𝑑𝑥 ⋯ ⋯ ⋯ ⋯ (1)
0

∞
𝑠𝑖𝑚𝑖𝑙𝑎𝑟𝑙𝑦 𝑝𝑢𝑡𝑡𝑖𝑛𝑔 𝒕 = 𝒚𝟐 𝑖𝑛 𝜞𝒏 = ∫ 𝒆−𝒕 𝒕𝒏−𝟏 𝒅𝒕
𝟎

∞
2
𝑤𝑒 𝑔𝑒𝑡 𝚪𝒏 = 2 ∫ 𝑒 −𝑦 𝑦 2𝑛−1 𝑑𝑦 ⋯ ⋯ ⋯ ⋯ (2)
0

∞ ∞
2 2
∴ Γ𝑚 Γ𝑛 = 4 ∫ 𝑒 −𝑥 𝑥 2𝑚−1 𝑑𝑥 ∫ 𝑒 −𝑦 𝑦 2𝑛−1 𝑑𝑦
0 0

∞ ∞
2 −𝑦 2 )
= 4 ∫ ∫ 𝑒 −(𝑥 𝑥 2𝑚+1 𝑦 2𝑛−1 𝑑𝑥 𝑑𝑦 ⋯ ⋯ ⋯ ⋯ (3)
0 0

Now changing to polar coordinates x = r cos  , y = sin  and d x d y = r d  d r, To cover the region in (3) which is
𝝅
the entire first quadrant r must vary from o to ∞ and  from o to 𝟐 Thus (3) become Γm Γn

𝜋
∞
2 2
= 4 ∫ ∫ 𝑒 −𝑟 𝑟 2𝑚−1 cos2𝑚−1 𝜃 𝑟 2𝑛−1 sin2𝑛−1 𝜃 ∙ 𝑟 𝑑 𝜃 ∙ 𝑑𝑟
0 0

𝜋
∞
2
−𝑟 2 2(𝑚+𝑛)−1
= [2 ∫ 𝑒 𝑟 𝑑 𝑟][2 ∫ cos2𝑚−1 𝜃 sin2𝑛−1 𝜃 𝑑 𝜃
0 0

= Γ(𝑚 + 𝑛) ∙ Γ(𝑚, 𝑛)

Γm Γn
∴ β (m, n) =
Γ(m + n)

which is the required relation between beta and gamma functions

 Fourier analysis & special functions

∞
Γ𝑛
= ∫ 𝑒 −𝑘𝑥 𝑥 𝑛−1 𝑑 𝑥
𝑘𝑛 0

Examples:
1
1. (i) Evaluate 𝐼 = ∫0 𝑥 7 (1 − 𝑥)3 𝑑 𝑥

1 1
𝑆𝑜𝑙𝑢𝑡𝑖𝑜𝑛: ∫ 𝑥 7 (1 − 𝑥)3 𝑑 𝑥 = ∫ 𝑥 8−1 (1 − 𝑥)4−1 𝑑 𝑥
0 0

Γ8 Γ4
𝛽(8, 4) =
Γ(8 + 4)

𝛤8 ∙ 3 ∙ 2 ∙ 1 3 ∙2 ∙ 1 1
= = =
11 ∙ 10 ∙ 9 ∙ 8 𝛤8 11 ∙ 10 ∙ 9 ∙ 8 1320 (𝑎𝑛𝑠. )

∞
2 √𝜋
(𝑖𝑖) 𝑃𝑟𝑜𝑣𝑒 𝑡ℎ𝑎𝑡, ∫ 𝑒 −𝑡 𝑑𝑡 =
0 2

∞
2
𝑃𝑟𝑜𝑜𝑓: 𝐿𝑒𝑡, 𝐼 = ∫ 𝑒 −𝑡 𝑑𝑡 … … … … … … … … … . (𝐼)
0

Putting 𝑡2 = 𝑥 𝑠𝑜 𝑡ℎ𝑎𝑡 2𝑡𝑑𝑡 =𝑑𝑥

1 𝑑𝑥 1 𝑑𝑥
∴ 𝑑𝑡 = =
2 𝑡 2 √𝑥

1 1
𝑜𝑟, 𝑑𝑡 = 𝑥 −2 𝑑𝑥
2

𝑡=0 𝑡=∞
Limits: } }
𝑥=0 𝑥=∞

Thus from (I), we have,

∞ ∞
2 1 1
𝐼 = ∫ 𝑒 −𝑡 𝑑𝑡 = ∫ 𝑒 −𝑥 𝑥 −2 𝑑𝑥
0 0 2

1 ∞ 1
= ∫ 𝑒 −𝑥 𝑥 2−1 𝑑𝑥
2 0

∞
1 1
= Γ 𝑠𝑖𝑛𝑐𝑒 ∫ 𝑒 −𝑥 𝑥 𝑛−1 𝑑 𝑥 = Γ𝑛
2 2 0

1 √𝜋
= √𝜋 =
2 2
 Fourier analysis & special functions

∞
2 √𝜋
∴ ∫ 𝑒 −𝑡 𝑑 𝑡 =
0 2
(proved)

𝟏
𝒅𝒙
(𝒊𝒊𝒊) 𝑬𝒙𝒑𝒓𝒆𝒔𝒔 𝒕𝒉𝒆 𝒊𝒏𝒕𝒆𝒈𝒓𝒂𝒍 ∫ 𝒊𝒏 𝒕𝒆𝒓𝒎𝒔 𝒐𝒇 𝒃𝒆𝒕𝒂 𝒇𝒖𝒏𝒄𝒕𝒊𝒐𝒏
𝟎 √𝟏 − 𝒙𝟑

1
𝑑𝑥
𝑆𝑜𝑙𝑢𝑡𝑖𝑜𝑛: 𝐼 = ∫ ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ (𝐼)
0 √1 − 𝑥 3

By definition of beta function, we have

1
𝛽(𝑚, 𝑛) = ∫ 𝑥 𝑚−1 (1 − 𝑥)𝑛−1 𝑑 𝑥, 𝑚 > 0, 𝑛>0
0

Putting x3 = t in (I) so that

3x2 d x = d t

1 2
𝑥 = 𝑡3 ∴ 𝑥2 = 𝑡3

1
𝑑𝑥 = 2 𝑑𝑡
3𝑡 3

𝑥=0 𝑥=1 1
Limits: } } 𝑑𝑥 = 2 𝑑 𝑡
𝑡=0 𝑡=1 3𝑡 3

Thus, from (I), we get,

1
1 1
𝐼= ∫ 1 2 𝑑𝑡
0 (1 − 𝑡)2 3𝑡 3

1 1 −2 1
= ∫ 𝑡 3 (1 − 𝑡)−2 𝑑 𝑡
3 0

1 1 2 1
= ∫ 𝑡 −3 (1 − 𝑡)−2 𝑑 𝑡
3 0

1 1 1 1
= ∫ 𝑡 3−1 (1 − 𝑡)2−1 𝑑 𝑡
3 0

1 1 1
= 𝛽 ( , )
3 3 2
1
𝑑𝑥 1 1 1
∴𝐼=∫ = 𝛽 ( , )
0 √1 − 𝑥 3 3 3 2
 Fourier analysis & special functions

Bessel Function

The differential equation of the form

𝑑2 𝑦 𝑑𝑦
𝑥2 +𝑥 + (𝑥 2 − 𝑛2 )𝑦 = 0; is all Bessel’s differential equation of order r Bessel differential equation is
𝑑𝑥 2 𝑑𝑥

𝑥𝑦" + 𝑥𝑦′ + (𝑥 2 − 𝑛2 )𝑦 = 0 ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ (1)

Here, x=0 is a regular singular point so let the series solution of (1)

𝑦 = ∑ 𝑎𝑟 𝑥 𝑐+𝑟 , 𝑎0 ≠ 0 ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ (2)
𝑟=0

∴ 𝑦 = ∑ 𝑎𝑟 (𝑐 + 𝑟) 𝑥 𝑐+𝑟−1
′

𝑟=0

𝑎𝑛𝑑, 𝑦" = ∑ 𝑎𝑟 (𝑐 + 𝑟)(𝑐 + 𝑟 − 1) 𝑥 𝑐+𝑟−2

𝑟=0

Now putting the values of y, y’ and y” we get,

∞ ∞ ∞
2 𝑐+𝑟−2 𝑐+𝑟−1 (𝑥 2 2)
𝑥 ∑ 𝑎𝑟 (𝑐 + 𝑟) (𝑐 + 𝑟 − 1) 𝑥 + 𝑥 ∑ 𝑎𝑟 (𝑐 + 𝑟) 𝑥 + −𝑛 ∑ 𝑎𝑟 𝑥 𝑐+𝑟 = 0
𝑟=0 𝑟=0 𝑟=0

∞ ∞ ∞ ∞
𝑐+𝑟 𝑐+𝑟 𝑐+𝑟+2
⇒ ∑ 𝑎𝑟 (𝑐 + 𝑟) (𝑐 = 𝑟 − 1)𝑥 + ∑ 𝑎𝑟 (𝑐 + 𝑟) 𝑥 + ∑ 𝑎𝑟 𝑥 − ∑ 𝑛2 𝑎𝑟 𝑥 𝑐+𝑟 = 0
𝑟=0 𝑟=0 𝑟=0 𝑟=0

∞ ∞
2 2 }𝑥 𝑐+𝑟
⇒ ∑ 𝑎𝑟 {(𝑐 + 𝑟) − (𝑐 + 𝑟)(𝑐 + 𝑟) − 𝑛 + ∑ 𝑎𝑟 𝑥 𝑐+𝑟+2 = 0
𝑟=0 𝑟=0

∞ ∞
2 2 }𝑥 𝑐+𝑟
⇒ ∑ 𝑎𝑟 {(𝑐 + 𝑟) − 𝑛 + ∑ 𝑎𝑟 𝑥 𝑐+𝑟+2 = 0
𝑟=0 𝑟=0

∞ ∞

⇒ 𝑎0 (𝑐 2 −𝑛 2) 𝑐
𝑥 + 𝑎1 {(𝑐 + 1) − 𝑛 2 2}
𝑥 𝑐+1 2
+ ∑ 𝑎𝑟 {(𝑐 + 𝑟) − 𝑛 }𝑥 2 𝑐+𝑟
+ ∑ 𝑎𝑟 𝑥 𝑐+𝑟+2 = 0
𝑟=0 𝑟=0

∞ ∞

⇒ 𝑎0 (𝑐 2 −𝑛 2) 𝑐
𝑥 + 𝑎1 {(𝑐 + 1) − 𝑛 2 2}
𝑥 𝑐+1
+ ∑ 𝑎𝑟+2 {(𝑐 + 𝑟 + 2) − 𝑛 }𝑥 2 2 𝑐+𝑟+2
+ ∑ 𝑎𝑟 𝑥 𝑐+𝑟+2 = 0
𝑟=0 𝑟=0

equating to zero the co-efficient of lowest power of x (i.e. the co-efficient of xc), the indicial equn a0 ( c2- n2 ) = 0

⇒ 𝑐 2 − 𝑛2 = 0 ∵ 𝑎0 ≠ 0

∴ 𝑐 = ±𝑛

equating to zero the co-efficient of x c+1

𝑎1 {(𝑐 + 1)2 − 𝑛2 } = 0
 Fourier analysis & special functions

𝑠𝑖𝑛𝑐𝑒 (𝑐 + 1)2 − 𝑛2 ≠ 0 𝑓𝑜𝑟 𝑐 = ±𝑛

Equating to zero the coefficient of x c+r+2 then the recurrence relation is,

𝑎𝑟+2 {(𝑐 + 𝑟 + 2)2 − 𝑛2 }𝑎𝑟 = 0

− 𝑎𝑟
⇒ 𝑎𝑟+2 =
(𝑐 + 𝑟 + 2)2 − 𝑛2

− 𝑎𝑟
⇒ 𝑎𝑟+2 = ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ (3)
(𝑐 + 𝑟 + 2 − 𝑛)(𝑐 + 𝑟 + 2 + 𝑛)

Now putting r = 0, 1, 2, 3, 4, …………………………in (3), we get

− 𝑎0 − 𝑎1
𝑎2 = , 𝑎3 = =0
(𝑐 + 2 − 𝑛)(𝑐 + 4 + 𝑛) (𝑐 + 3 − 𝑛)(𝑐 + 3 + 𝑛)

− 𝑎2 𝑎0
𝑎1 = = =0
(𝑐 + 4 − 𝑛)(𝑐 + 4 + 𝑛) (𝑐 + 4 − 𝑛)(𝑐 + 4 + 𝑛)(𝑐 + 2 − 𝑛)(𝑐 + 2 + 𝑛)

− 𝑎3
𝑎5 = = 0 ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ (4)
(𝑐 + 5 − 𝑛)(𝑐 + 5 + 𝑛)

and so on  a1 = a3 = a5 = ……………………... = 0

Now from (2) we get

y = x c (a0 + a1 x + a2 x 2 + a3 x3 + ……………….. )
𝑎0 𝑥 2 𝑎0 𝑥 4
= 𝑥 𝑐 {𝑎0 − + −. … … … … … … … … … }
(𝑐 + 2 − 𝑛)(𝑐 + 2 + 𝑛) (𝑐 + 4 − 𝑛)(𝑐 + 4 + 𝑛)(𝑐 + 2 − 𝑛)(𝑐 + 2 + 𝑛)

𝑥2 𝑥4
= 𝑎0 𝑥 𝑐 {1 − + −. … … … … } ⋯ ⋯ ⋯ ⋯ (5)
(𝑐 + 2 − 𝑛)(𝑐 + 2 + 𝑛) (𝑐 + 4 − 𝑛)(𝑐 + 4 + 𝑛)(𝑐 + 2 − 𝑛)(𝑐 + 2 + 𝑛)

Now in (5) replacing c by n, - n and replacing a0 by a and b, we get

𝑥2 𝑥4
𝑦 = 𝑎𝑥 𝑛 {1 − + −. … … … … … … . } ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ (6)
2(2𝑛 + 2) 2 ∙ 4 ∙ (2𝑛 + 2)(2𝑛 + 4)

𝑥2 𝑥4
𝑎𝑛𝑑, 𝑦 = 𝑏𝑥 −𝑛 {1 − + −. … … … … … … . } ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ (7)
2(2 − 2𝑛) 2 ∙ 4 (2 − 2𝑛)(4 − 2𝑛)

1
𝑇ℎ𝑒 𝑝𝑎𝑟𝑡𝑖𝑐𝑢𝑙𝑎𝑟 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 𝑜𝑓 (1) 𝑖𝑠 𝑜𝑏𝑡𝑎𝑖𝑛𝑒𝑑 𝑓𝑟𝑜𝑚 (6) 𝑏𝑦 𝑡𝑎𝑘𝑖𝑛𝑔 𝑎 = . 𝑇ℎ𝑖𝑠 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 𝑖𝑠 𝑐𝑎𝑙𝑙𝑒𝑑
2𝑛 𝛤𝑛 +1

𝐵𝑒𝑠𝑠𝑒𝑙 ′ 𝑠𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑡ℎ𝑒 𝑓𝑖𝑟𝑠𝑡 𝑘𝑖𝑛𝑑 𝑜𝑓 𝑜𝑡ℎ𝑒𝑟 𝑛. 𝐼𝑡 𝑖𝑠 𝑑𝑒𝑛𝑜𝑡𝑒𝑑 𝑏𝑦 𝐽𝑛 (𝑥)

𝑥𝑛 𝑥2 𝑥4
𝐽𝑛 (𝑥) = {1 − + −. … … … … … … . }
2𝑛 Γ𝑛 + 1 22 (𝑛 + 1) 22 ∙ 23 (𝑛 + 1)(𝑛 + 2)

𝑥 𝑛 𝑥 2 𝑥 4
(2) (2) (2)
= { + −. … … … … … … . }
Γ𝑛 + 1 1 ∙ (𝑛 + 1) 1 ∙ 2 ∙ (𝑛 + 1)(𝑛 + 2)
 Fourier analysis & special functions

𝑥 𝑛 𝑥 𝑛+2 𝑥 𝑛+4
( ) ( ) ( )
={ 2 − 2 + 2 −. … … … … … … . }
1 ∙ Γ𝑛 + 1 1 ∙ (𝑛 + 1) Γ𝑛 + 1 1 ∙ 2 ∙ (𝑛 + 2)(𝑛 + 1)Γ𝑛 + 1

𝑥 𝑛 𝑥 𝑛+2 𝑥 𝑛+4
(2) (2) (2 )
={ − + −. … … … … … … . }
Γ1 Γ𝑛 + 1 Γ2 Γ𝑛 + 2 Γ3 ∙ Γ𝑛 + 3

∞ 𝑥 𝑛+2𝑟
(−1)𝑟 ( )
=∑ 2
Γ𝑟 + 1 Γ𝑛 + 𝑟 + 1
𝑟=0

∞ 𝑥 𝑛+2𝑟
(−1)𝑟 ( )
∴ 𝐽𝑛 (𝑥) = ∑ 2
r ! Γ𝑛 + 𝑟 + 1
𝑟=0

1
𝑇ℎ𝑒 𝑝𝑎𝑟𝑡𝑖𝑐𝑢𝑙𝑎𝑟 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 𝑜𝑓 (1)𝑖𝑠 𝑜𝑏𝑡𝑎𝑖𝑛𝑒𝑑 𝑏 𝑓𝑟𝑜𝑚 (7)𝑏𝑦 𝑡𝑎𝑘𝑖𝑛𝑔 𝑏 = 𝑎𝑛𝑑 𝑝𝑟𝑜𝑐𝑒𝑒𝑑𝑖𝑛𝑔
2𝑛 Γ𝑛 +1

𝑇ℎ𝑖𝑠 𝑖𝑠 𝑐𝑎𝑙𝑙𝑒𝑑 𝑡ℎ𝑒 𝑔𝑒𝑛𝑒𝑟𝑎𝑙 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 𝑜𝑓 𝐵𝑒𝑠𝑠𝑒𝑙 ′ 𝑠𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑜𝑟𝑑𝑒𝑟 𝑛

y = A Jn (x) + B J-n (x)

Where A and arbitrary constants. Here n is not integer.

J-n (x) = (-1)n Jn (x) , Prove it.

Proof: From the definition of Bessel’s function

∞ 𝑥 𝑛+2𝑟
(−1)𝑟 ( )
𝐽𝑛 (𝑥) = ∑ 2
𝑟 ! Γ𝑛 + 𝑟 + 1
𝑟=0

Now replacing n by -n

∞ 𝑥 𝑛+2𝑟
(−1)𝑟 ( )
𝐽−𝑛 (𝑥) = ∑ 2 ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ (1)
𝑟 ! Γ𝑛 + 𝑟 + 1
𝑟=0

Since n is integer so for r = 0, 1, 2, ………., n+1

1
then Γ − 𝑛 + 𝑟 + 1 is infinite and hence Γ−𝑛+𝑟+1

in this case (1) can be written as

∞ 𝑥 −𝑛+2𝑟
(−1)𝑟 ( )
𝐽−𝑛 (𝑥) = ∑ 2 ∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙∙ (2)
𝑟! Γ−𝑛+𝑟+1
𝑟=𝑛

now let 𝑟−𝑛 = 𝑘 ⇒𝑟 =𝑛+𝑘

 Fourier analysis & special functions

∞ 𝑥 −𝑛+2𝑛+2𝑘
(−1)𝑛+𝑘 ( )
𝐹𝑟𝑜𝑚 (2) ⇒ 𝐽−𝑛 (𝑥) = ∑ 2
(𝑛 + 𝑘) ! Γ − 𝑛 + 𝑛 + 𝑘 + 1
𝑘=0

where, r = n, k = 0, r = ∞ , k = - ∞

∞ 𝑥 𝑛+2𝑘
(−1)𝑘 ( )
= (−1)𝑛 ∑ 2
(𝑛 + 𝑘) ! Γ k + 1
𝑘=0

∞ 𝑥 𝑛+2𝑘 as
(−1)𝑘 ( )
= (−1)𝑛 ∑ 2
(𝑛 + 𝑘) ! Γ k + 1
𝑘=0 Γ𝑛 = (𝑛 − 1)!

∞ 𝑥 𝑛+2𝑘 ∴ Γ𝑛 + 1 = 𝑛!
(−1)𝑘 ( )
= (−1)𝑛 ∑ 2
𝑘! Γ n + k + 1
𝑘=0 𝑎𝑛𝑑 (𝑛 + 𝑘)! = Γ𝑛 + 𝑘 + 1

= (-1)n Jn (x)

 J-n (x) = (-1)n Jn (x)

[Proved]

𝑷𝒓𝒐𝒗𝒆 𝒕𝒉𝒂𝒕, 𝟐 𝑱′𝒏 (𝒙) = 𝑱𝒏−𝟏 (𝒙) − 𝑱𝒏+𝟏 (𝒙)

Proof: we have,

∞ 𝑥 𝑛+2𝑟
(−1)𝑟 ( )
𝐽𝑛 (𝑥) = ∑ 2
𝑟 ! Γ𝑛 + 𝑟 + 1
𝑟=0

∞ 𝑥 𝑛+2𝑟−1 1
(−1)𝑟 ( ) (𝑛 + 2𝑟) ∙
′ (𝑥)
𝐽𝑛 =∑ 2 2
𝑟 ! Γ𝑛 + 𝑟 + 1
𝑟=0

∞ 𝑥 𝑛+2𝑟−1
1 (−1)𝑟 {(𝑛 + 𝑟) + 𝑟} ( )
= ∑ 2
2 𝑟 ! Γ𝑛 + 𝑟 + 1
𝑟=0

∞ 𝑥 𝑛+2𝑟−1 ∞ 𝑥 𝑛+2𝑟−1
(−1)𝑟 (𝑛 + 𝑟) ( ) (−1)𝑟 ∙ 𝑟 ∙ ( )
2𝐽′ 𝑛 (𝑥) = ∑ 2 +∑ 2
𝑟 ! Γ𝑛 + 𝑟 + 1 𝑟 ! Γ𝑛 + 𝑟 + 1
𝑟=0 𝑟=0

∞ 𝑥 𝑛+2𝑟−1 ∞ 𝑥 𝑛+2𝑟−1
(−1)𝑟 ( ) (−1)𝑟 ( )
=∑ 2 +∑ 2
𝑟 ! Γ𝑛 + 𝑟 (𝑟 − 1)! Γ𝑛 + 𝑟 + 1
𝑟=0 𝑟=0

∞ 𝑥 (𝑛−1)+2𝑟 ∞ 𝑥 𝑛+2𝑟−1
(−1)𝑟 ( ) (−1)𝑟 ( ) 1
=∑ 2 +∑ 2 [ = 0]
𝑟 ! Γ(n − 1) + 𝑟 + 1 (𝑟 − 1)! Γ𝑛 + 𝑟 + 1 −1!
𝑟=0 𝑟=1
 Fourier analysis & special functions

∞ 𝑥 𝑛+2𝑟−1
(−1)𝑟 ( )
= 𝐽𝑛−1 (𝑥) + ∑ 2
(𝑟 − 1)! Γ𝑛 + 𝑟 + 1
𝑟=1

Let, r – 1 = k,  r = k + 1

𝑟 = 1, 𝑘 = 0
Limits: when
𝑟=∞ 𝑘=∞

∞ 𝑥 𝑛+2𝑘+2−1
(−1)𝑘+1 ( )
= 𝐽𝑛−1 (𝑥) + ∑ 2
𝑘 ! Γ𝑛 + 𝑘 + 2
𝑘=0

∞ 𝑥 (𝑛+1)+2𝑘
(−1)𝑘 ( )
= 𝐽𝑛−1 (𝑥) + (−1)1 ∑ 2
𝑘 ! Γ(𝑛 + 1) + (𝑘 + 1)
𝑘=0

∴ 2𝐽′ 𝑛 (𝑥) = 𝐽𝑛−1 (𝑥) − 𝐽𝑛+1 (𝑥)

[Proved]

Fourier Transform
Defination:

Let f(x) be a function defined on ( - ∞ , ∞ ) and is piecewise continuous differentiable in each finite interval and is
absolutely integrable on ( - ∞ , ∞ ), then the Fourier Transform of f(x) denoted by F [f(x)] and written in

∞
1
𝐹 [ 𝑓(𝑥)] = 𝐹(∝) = ∫ 𝑓(𝑥)𝑒 𝑖∝𝑥 𝑑𝑥
√2𝜋 −∞

𝑇ℎ𝑒 𝑖𝑛𝑣𝑒𝑟𝑠𝑒 𝑜𝑓 𝐹(∝)𝑖𝑠 𝑑𝑒𝑛𝑜𝑡𝑒𝑑 𝑏𝑦 𝐹 −1 [𝑓(∝)]𝑎𝑛𝑑 𝑖𝑠 𝑑𝑒𝑓𝑖𝑛𝑒𝑑 𝑏𝑦 𝐹 −1 [𝑓(∝)] = 𝑓(𝑥)

∞
1
= ∫ 𝑓(∝)𝑒 −𝑖∝𝑥 𝑑 ∝
√2𝜋 −∞

∞
𝐹𝑜𝑢𝑟𝑖𝑒𝑟 𝑇𝑟𝑎𝑛𝑠𝑓𝑜𝑟𝑚 𝑖𝑠 𝑎𝑙𝑠𝑜 𝑔𝑖𝑣𝑒𝑛 𝑎𝑠, 𝐹[𝑓(𝑥)] = 𝐹(∝) = ∫ 𝑓(𝑥)𝑒 𝑖∝𝑥 𝑑 𝑥
−∞

∞
1
𝑎𝑛𝑑 𝑡ℎ𝑒 𝑐𝑜𝑟𝑟𝑒𝑠𝑝𝑜𝑛𝑑𝑖𝑛𝑔 𝑖𝑛𝑣𝑒𝑟𝑠𝑒 𝐹𝑜𝑢𝑟𝑖𝑒𝑟 𝑡𝑟𝑎𝑛𝑠𝑓𝑜𝑟𝑚 𝐹 −1 [𝐹(∝)] = 𝑓(𝑥) = ∫ 𝐹(∝)𝑒 −𝑖∝𝑥 𝑑 ∝
√2𝜋 −∞

Fourier transform of some elementary functions:

− 𝒙𝟐⁄
Find the Fourier transform of 𝒇(𝒙) = 𝒆 𝟐

Solution: From the definition of F.T. we have,

∞
1
𝐹 [𝑓 (𝑥 )] = ∫ 𝑓(𝑥 )𝑒 𝑖∝𝑥 𝑑 𝑥
√2𝜋 −∞
 Fourier analysis & special functions

2
1 ∞ −𝑥
= ∫ 𝑒 2 ∙ 𝑒 𝑖∝𝑥 𝑑 𝑥
√2𝜋 −∞

− ∝2
1 ∞ −1(𝑥−𝑖∝)2
∫ 𝑒 2
√2𝜋 −∞
𝑒 2 𝑑𝑥

Let,
𝑥−𝑖∝
=𝑡
√2

𝑑𝑥
⇒ = 𝑑𝑡
√2

∝2
−2 ∞
𝑒 −𝑡 2
∞ 𝑥2
√𝜋
𝐹 [𝑓 (𝑥 )] = ∫ 𝑒 𝑑𝑡 ∫ 𝑒2 𝑑𝑥=
2
√𝜋 −∞ 0

∝2
−
𝑒 2
= ∙ √𝜋
√𝜋

∝2
−2
= 𝑒

• Find the F.T. of 𝒇(𝒙) = 𝒆−𝒂|𝒙| , - ∞ < x < ∞

Solution:
We know that,
−𝑥, 𝑥 < 0
|𝑥| = {
𝑥, 𝑥 ≥ 0

From the definition of F.T. we have,

∞
1
𝐹[𝑓(𝑥)] = ∫ 𝑓(𝑥) 𝑒 𝑖∝𝑥 𝑑 𝑥
√2𝜋 −∞
0 ∞
1 𝑎𝑥 𝑖∝𝑥
1
= ∫ 𝑒 𝑒 𝑑𝑥+ ∫ 𝑒 −𝑎𝑥 𝑒 𝑖∝𝑥 𝑑 𝑥
√2𝜋 −∞ √2𝜋 0

0 ∞
1 (𝑎+𝑖∝)𝑥
1
= ∫ 𝑒 𝑑𝑥+ ∫ 𝑒 (−𝑎+𝑖∝)𝑥 𝑑 𝑥
√2𝜋 −∞ √2𝜋 0

√2 a
= ( 2 )
𝜋 a +∝2
Ans: