BESSEL FUNCTION

Introductions:
Bessel’s functions, named after the mathematician Friedrich Bessel
was first defined by Daniel Bernoulli. There are two types of Bessel’s
functions, namely, Bessel function of Type-1 and Bessel function of
Type-2 and both of them are the two independent solutions of a second
order ordinary differential equation which is known as Bessel’s
equation. Between these two types of functions, Bessel function of first
kind is the most important one as this function related to various
physical phenomenon. In this chapter we will define Bessel’s equation,
Bessel’s function and discuss various properties.
Bessel’s Equation:
The Differential equation is of the form

2
𝑑2 𝑦 𝑑𝑦
𝑥 2
+ 𝑥 + (𝑥 2 − 𝑛2 )𝑦 = 0
𝑑𝑥 𝑑𝑥
𝑑2 𝑦 1 𝑑𝑦 𝑛2
or, + + (1 − ) 𝑦 = 0 … … … … … … . (1)
𝑑𝑥 2 𝑥 𝑑𝑥 𝑥2

is called Bessel’s equation of order n, n being a non- negative constant.

Solution of Bessel’s differential equation:

We now solve (1) in series by using the well-known method of
Frobenius.
Let

𝑦 = ∑ 𝑎𝑚 𝑥 𝑚+𝑘 , 𝑎0 ≠ 0 … … … … … (2)
𝑚=0

be the series solution of (1). Then

 ∞
𝑑𝑦
= ∑ 𝑎𝑚 (𝑚 + 𝑘)𝑥 𝑚+𝑘−1
𝑑𝑥
𝑚=0
∞
𝑑2 𝑦
2
= ∑ (𝑚 + 𝑘)(𝑚 + 𝑘 − 1)𝑎𝑚 𝑥 𝑚+𝑘−2
𝑑𝑥
𝑚=0

Substituting these values in (1), we get

∞ ∞

𝑥 2 ∑ (𝑚 + 𝑘)(𝑚 + 𝑘 − 1)𝑎𝑚 𝑥 𝑚+𝑘−2 + 𝑥 ∑ 𝑎𝑚 (𝑚 + 𝑘)𝑥 𝑚+𝑘−1

𝑚=0 𝑚=0
∞

+ (𝑥 2 − 𝑛2 ) ∑ 𝑎𝑚 𝑥 𝑚+𝑘 = 0
𝑚=0

or,
∞

∑ 𝑎𝑚 {(𝑚 + 𝑘)(𝑚 + 𝑘 − 1) + (𝑚 + 𝑘) − 𝑛2 }𝑥 𝑚+𝑘

𝑚=0
∞

+ ∑ 𝑎𝑚 𝑥 𝑚+𝑘+2 = 0
𝑚=0

or,
∞ ∞

∑ 𝑎𝑚 (𝑘 + 𝑚 + 𝑛)(𝑘 + 𝑚 − 𝑛)𝑥 𝑚+𝑘 + ∑ 𝑎𝑚 𝑥 𝑚+𝑘+2

𝑚=0 𝑚=0
= 0 … … … … . (3)
Equating the co-efficient of lower power of 𝑥 i.e. 𝑥 𝑘 to zero, we get the indicial
equation as
𝑎0 (𝑘 + 𝑛)(𝑘 − 𝑛) = 0
or, 𝑘 = −𝑛, 𝑛.
So, the roots of the indicial equation are distinct and not differing by an integer.
Next equating to zero, the co-efficient of 𝑥 𝑘+1 gives
𝑎1 (𝑘 + 1 + 𝑛)(𝑘 + 1 − 𝑛) = 0
or, 𝑎1 = 0 𝑓𝑜𝑟 𝑘 = 𝑛, −𝑛
Lastly equating to zero, the co-efficient of 𝑥 𝑘+𝑚 , we get the recurrence relation
as
𝑎𝑚 (𝑘 + 𝑚 + 𝑛)(𝑘 + 𝑚 − 𝑛) + 𝑎𝑚−2 = 0
𝑎𝑚−2
∴ 𝑎𝑚 = − … … … … … (4)
(𝑘 + 𝑚 + 𝑛)(𝑘 + 𝑚 − 𝑛)
Then putting 𝑛 = 3,5,7, … in (4) and using 𝑎1 = 0, we find
𝑎1 = 𝑎3 = 𝑎5 = ⋯ = 0
Now two cases arise:
Case I: When 𝒌 = 𝒏.
Now putting 𝑛 = 2,4,6, … in (4) successively, we get
1
𝑎2 = − 𝑎
(𝑘 + 2 + 𝑛)(𝑘 + 2 − 𝑛) 0
1
𝑎4 = − 𝑎
(𝑘 + 4 + 𝑛)(𝑘 + 4 − 𝑛) 2
1
= 𝑎
(𝑘 + 4 + 𝑛)(𝑘 + 4 − 𝑛)(𝑘 + 2 + 𝑛)(𝑘 + 2 − 𝑛) 0
and so on.
Substituting these values of 𝑎𝑖 ′𝑠 in (2), we get

𝑘
𝑥2
𝑦 = 𝑎0 𝑥 (1 −
(𝑘 + 2 + 𝑛)(𝑘 + 2 − 𝑛)
𝑥4
+ − ⋯)
(𝑘 + 4 + 𝑛)(𝑘 + 4 − 𝑛)(𝑘 + 2 + 𝑛)(𝑘 + 2 − 𝑛)
∴ For 𝑘 = 𝑛, we have

𝑛
𝑥2 𝑥4
𝑦 = 𝑎0 𝑥 (1 − + − ⋯)
4(1 + 𝑛) 4 ⋅ 8 (1 + 𝑛)(2 + 𝑛)
 1
If 𝑎0 = , this solution is called Bessel’s function of the first
2𝑛 Γ(𝑛+1)
kind of order n and is denoted by 𝐽𝑛 (𝑥 ), where Γ(𝑛) is Gamma
Function.
𝑥𝑛 𝑥2
∴ 𝐽𝑛 (𝑥 ) = 𝑛 [1 + (−1) 2
2 Γ(𝑛 + 1) 2 1(𝑛 + 1)
2
𝑥4
+ (−1) 4 + ⋯]
2 2! (𝑛 + 1)(𝑛 + 2)
∞
𝑥𝑛 𝑟
𝑥 2𝑟
= 𝑛 ∑ [(−1) 𝑟 ]
2 Γ(𝑛 + 1) 2 𝑟! (𝑛 + 1)(𝑛 + 2) … (𝑛 + 𝑟)
𝑟=0
∞
𝑥 𝑛+2𝑟 1
= ∑(−1 )𝑟 ( )
2 𝑟! Γ(𝑛 + 𝑟 + 1)
𝑟=0
Case II: When 𝒌 = −𝒏.
The series solution is obtained by replacing n by -n in the value of 𝐽𝑛 .
∞
𝑥 −𝑛+2𝑟 1
∴ 𝐽−𝑛 (𝑥 ) = ∑(−1 )𝑟 ( )
2 𝑟! Γ(−𝑛 + 𝑟 + 1)
𝑟=0

when n is not integer 𝐽−𝑛 (𝑥 ) is distinct from 𝐽𝑛 (𝑥 ).

Note: The solution of Bessel’s equation is called Bessel’s functions.
General solution of Bessel’s equation:
General solution of Bessel’s equation is
𝑦 = 𝐴𝐽−𝑛 (𝑥 ) + 𝐵𝐽𝑛 (𝑥 )
where 𝐴 and 𝐵 are two arbitrary constants.

Series solution of Bessel’s equation for 𝒏 = 𝟎

The Differential equation is of the form
𝑑2 𝑦 1 𝑑𝑦
+ + 𝑦 = 0 … … … … . . (1)
𝑑𝑥 2 𝑥 𝑑𝑥
is known as Bessel’s equation for 𝑛 = 0.

Let
∞

𝑦 = ∑ 𝑎𝑚 𝑥 𝑚+𝑘 , 𝑎0 ≠ 0 … … … … … (2)
𝑚=0

be the series solution of (1). Then

∞
𝑑𝑦
= ∑ 𝑎𝑚 (𝑚 + 𝑘)𝑥 𝑚+𝑘−1
𝑑𝑥
𝑚=0
∞
𝑑2 𝑦
2
= ∑ (𝑚 + 𝑘)(𝑚 + 𝑘 − 1)𝑎𝑚 𝑥 𝑚+𝑘−2
𝑑𝑥
𝑚=0

Substituting these values in (1), we get

∞ ∞

∑ (𝑚 + 𝑘)2 𝑎𝑚 𝑥 𝑚+𝑘−2 + ∑ 𝑎𝑚 𝑥 𝑚+𝑘 = 0 … … … … . . (3)

𝑚=0 𝑚=0
which is an identity.
Equating the co-efficient of lower power of 𝑥 i.e. 𝑥 𝑘−2 to zero, we get
𝑎0 𝑘 2 = 0
or, 𝑘 = 0, 0 (∵ 𝑎0 ≠ 0)
Next equating to zero, the co-efficient of next power of x i.e., 𝑥 𝑘−1 , we have
𝑎1 (𝑘 + 1)2 = 0
or, ∴ 𝑎1 = 0, 𝑓𝑜𝑟 𝑘 = 0
Again, equating to zero, the co-efficient of the general term i.e., 𝑥 𝑘+𝑚 , we get
the recurrence relation as
𝑎𝑚+2 (𝑘 + 𝑚 + 2)2 + 𝑎𝑚 = 0
𝑎𝑚
∴ 𝑎𝑚+2 = − … … … … … (4)
(𝑘 + 𝑚 + 𝑛)2
Then putting m= 1,3,5,7, … in (4) and using 𝑎1 = 0, we find
𝑎1 = 𝑎3 = 𝑎5 = ⋯ = 0
Now putting 𝑚 = 0,2,4,6, … in (4) successively, we get
1
𝑎2 = − 𝑎
(𝑘 + 2)2 0
1 1
𝑎4 = − 𝑎2 = 𝑎
(𝑘 + 4)2 (𝑘 + 4)2 (𝑘 + 2)2 0
and so on.
Substituting these values of 𝑎𝑖 ′𝑠 in (2), we get

𝑘
𝑥2 𝑥4
𝑦 = 𝑎0 𝑥 (1 − + − ⋯)
(𝑘 + 2)2 (𝑘 + 2)2 (𝑘 + 4)2
∴ For 𝑘 = 0, we have
𝑥2 𝑥4
𝑦 = 𝑎0 (1 − 2 + 2 2 − ⋯ )
2 2 ⋅4

If 𝑎0 = 1, this solution is denoted by 𝐽0 (𝑥 ).

𝑥2 𝑥4 𝑥4
∴ 𝐽0 (𝑥 ) = 1 − 2 + 2 2 − 2 2 2 + ⋯
2 2 ⋅4 2 ⋅4 ⋅6
Where 𝐽0 (𝑥 ) is called Bessel’s function of zeroth order.9

Recurrence formula:
𝒅
1) (𝒙𝒏 𝑱𝒏 ) = 𝒙𝒏 𝑱𝒏−𝟏 or, 𝒙𝑱′𝒏 (𝒙) = −𝒏𝑱𝒏 (𝒙) + 𝒙𝑱𝒏−𝟏 (𝒙)
𝒅𝒙
𝒅
2) (𝒙−𝒏 𝑱𝒏 ) = −𝒙−𝒏 𝑱𝒏+𝟏 or, 𝒙𝑱′𝒏 (𝒙) = 𝒏𝑱𝒏 (𝒙) − 𝒙𝑱𝒏+𝟏 (𝒙).
𝒅𝒙

Proof:
1) We know that
 ∞ 𝑥 𝑛+2𝑟
(−1)𝑟 ( )
𝐽𝑛 (𝑥 ) = ∑ 2
Γ(𝑟 + 1)Γ(𝑛 + 𝑟 + 1)
𝑟=0
∞
𝑛
(−1)𝑟 𝑥 2(𝑛+𝑟)
∴ 𝑥 𝐽𝑛 (𝑥 ) = ∑ 𝑛+2𝑟
2 Γ(𝑟 + 1)Γ(𝑛 + 𝑟 + 1)
𝑟=0
∞
𝑑 𝑛 (−1)𝑟 2(𝑛 + 𝑟)𝑥 2(𝑛+𝑟)−1
∴ {𝑥 𝐽𝑛 (𝑥 )} = ∑ 𝑛+2𝑟
𝑑𝑥 2 Γ(𝑟 + 1)Γ(𝑛 + 𝑟 + 1)
𝑟=0

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

𝑑 𝑛
∴ {𝑥 𝐽𝑛 (𝑥 )} = 𝑥 𝑛 𝐽𝑛−1 (𝑥)
𝑑𝑥
∴ 𝑥 𝑛 𝐽𝑛′ (𝑥 ) + 𝑛𝑥 𝑛−1 𝐽𝑛 (𝑥 ) = 𝑥 𝑛 𝐽𝑛−1 (𝑥)

∴ 𝒙𝑱′𝒏 (𝒙) = −𝒏𝑱𝒏 (𝒙) + 𝒙𝑱𝒏−𝟏 (𝒙)

2) We know that
∞ 𝑥 𝑛+2𝑟
(−1)𝑟 ( )
𝐽𝑛 (𝑥 ) = ∑ 2
Γ(𝑟 + 1)Γ(𝑛 + 𝑟 + 1)
𝑟=0
∞
−𝑛
(−1)𝑟 𝑥 2𝑟
∴𝑥 𝐽𝑛 (𝑥 ) = ∑ 𝑛+2𝑟
2 Γ(𝑟 + 1)Γ(𝑛 + 𝑟 + 1)
𝑟=0
∞
𝑑 −𝑛 (−1)𝑟 2𝑟𝑥 2𝑟−1
∴ {𝑥 𝐽𝑛 (𝑥 )} = ∑ 𝑛+2𝑟
𝑑𝑥 2 Γ(𝑟 + 1)Γ(𝑛 + 𝑟 + 1)
𝑟=0
 𝑑 −𝑛
{𝑥 𝐽𝑛 (𝑥 )}
𝑑𝑥
∞ 𝑥 {(𝑛+1)+2(𝑟−1)}
(−1)𝑟−1 ( )
= −𝑥 −𝑛 ∑ 2
Γ{(𝑟 − 1) + 1}Γ{(𝑛 + 1) + (𝑟 − 1) + 1}
𝑟=1

[Since Γ(r) = ∞ 𝑤ℎ𝑒𝑛 𝑟 = 0, so the term corresponding to 𝑟 = 0

vanishes]
∞ 𝑥 {(𝑛+1)+2𝑚}
(−1 ( ) )𝑚
= −𝑥 −𝑛 ∑ 2
Γ(m + 1)Γ{(𝑛 + 1) + 𝑚 + 1}
𝑚=0

= −𝑥 −𝑛 𝐽𝑛+1 (𝑥 )
[ By putting 𝑚 = 𝑟 − 1]
𝑑 −𝑛
∴ {𝑥 𝐽𝑛 (𝑥 )} = −𝑥 −𝑛 𝐽𝑛+1 (𝑥 )
𝑑𝑥
∴ 𝑥 −𝑛 𝐽𝑛′ (𝑥 ) − 𝑛𝑥 −𝑛−1 𝐽𝑛 (𝑥 ) = −𝑥 −𝑛 𝐽𝑛+1 (𝑥)

∴ 𝒙𝑱′𝒏 (𝒙) = 𝒏𝑱𝒏 (𝒙) − 𝒙𝑱𝒏+𝟏 (𝒙)

3) From recurrence relations (1) and (2), we have

𝑥𝐽𝑛′ (𝑥 ) = −𝑛𝐽𝑛 (𝑥 ) + 𝑥𝐽𝑛−1 (𝑥 ) … … … … (1)
𝑥𝐽𝑛′ (𝑥 ) = 𝑛𝐽𝑛 (𝑥 ) − 𝑥𝐽𝑛+1 (𝑥 ) … … … … … . (2)
Adding (1) and (2), we get
2𝑥𝐽𝑛′ (𝑥 ) = 𝑥𝐽𝑛−1 (𝑥 ) − 𝑥𝐽𝑛+1 (𝑥 )

∴ 𝟐𝑱′𝒏 (𝒙) = 𝑱𝒏−𝟏 (𝒙) − 𝑱𝒏+𝟏 (𝒙)

4) From recurrence relations (1) and (2), we have

𝑥𝐽𝑛′ (𝑥 ) = −𝑛𝐽𝑛 (𝑥 ) + 𝑥𝐽𝑛−1 (𝑥)
 𝑥𝐽𝑛′ (𝑥 ) = 𝑛𝐽𝑛 (𝑥 ) − 𝑥𝐽𝑛+1 (𝑥 )
Subtracting (2) from (1), we get
0 = 2𝑛𝐽𝑛 (𝑥 ) − 𝑥𝐽𝑛+1 (𝑥 ) − 𝑥𝐽𝑛−1 (𝑥)
2𝑛𝐽𝑛 (𝑥 ) = 𝑥[𝐽𝑛−1 (𝑥 ) + 𝐽𝑛+1 (𝑥 )]

𝟐𝒏𝑱𝒏 (𝒙) = 𝒙[𝑱𝒏−𝟏 (𝒙) + 𝑱𝒏+𝟏 (𝒙)]

Definition:
Bessel’s function of 2nd kind of order n or Neumann function is (given
by)

𝒅𝒙
𝒀𝒏 (𝒙) = 𝑱𝒏 (𝒙) ∫ .
𝒙[𝑱𝒏 (𝒙)]𝟐

Illustrative Examples:
Example 1:
Show that when 𝒏 is a positive integer

i) 𝑱−𝒏 (𝒙) = (−𝟏)𝒏 𝑱𝒏 (𝒙) and

ii) 𝑱𝒏 (−𝒙) = (−𝟏)𝒏 𝑱𝒏 (𝒙) for +𝒗𝒆 or −𝒗𝒆
integers.
Solution:
i) We have
∞
𝑥 −𝑛+2𝑟 1
𝐽−𝑛 (𝑥 ) = ∑(−1 )𝑟 ( )
2 r! Γ(−𝑛 + 𝑟 + 1)
𝑟=0

since, if 𝑝 is an integer, then Γ(−𝑝) is infinite for 𝑝 > 0. So, we get

terms in 𝐽−𝑛 equal to zero till −𝑛 + 𝑟 + 1 < 1, i.e., 𝑟 < 𝑛.
Hence, we can write
∞
(−1)𝑟 𝑥 −𝑛+2𝑟
𝐽−𝑛 (𝑥 ) = ∑ ( )
r! Γ(−𝑛 + 𝑟 + 1) 2
𝑟=0
∞
(−1)𝑛+𝑠 𝑥 𝑛+2𝑠
=∑ ( ) ,
(n + s)! Γ(𝑠 + 1) 2
𝑠=0

by putting 𝑟 = 𝑛 + 𝑠
∞
(−1)𝑠 𝑥 𝑛+2𝑠
= (−1 )𝑛 ∑ ( )
s! Γ(𝑛 + 𝑠 + 1) 2
𝑠=0

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

Case a: Let 𝑛 be a +𝑣𝑒 integer

Replacing 𝑥 by −𝑥 in (1), we have
∞
1 𝑥 𝑛+2𝑟
𝐽𝑛 (−𝑥 ) = ∑(−1 )𝑟 (− )
r! Γ(𝑛 + 𝑟 + 1) 2
𝑟=0
∞ 1 𝑥 𝑛+2𝑟
= (−1 )𝑛 ∑𝑟=0 ( )
r!Γ(𝑛+𝑟+1) 2
= (−1)𝑛 𝐽𝑛 (𝑥 ).

Case b: Let 𝑛 be a −𝑣𝑒 integer

So let 𝑛 = −𝑚, where 𝑚 is a +𝑣𝑒 integer.
𝐽𝑛 (𝑥 ) = 𝐽−𝑚 (𝑥 ) = (−1)𝑚 𝐽𝑚 (𝑥 ) by (𝑖)
Replacing 𝑥 by −𝑥 , we have
𝐽𝑛 (−𝑥 ) = (−1)𝑚 𝐽𝑚 (−𝑥 )
= (−1)𝑚 (−1)𝑚 𝐽𝑚 (𝑥 ) by case (a)
= (−1)𝑚 𝐽−𝑚 (𝑥 ) = (−1)2𝑚 (−1)−𝑚 𝐽−𝑚 (𝑥 )
 = (−1)𝑚 𝐽−𝑚 (𝑥 ) (∵ (−1)2𝑚 = 1)
= (−1)𝑛 𝐽𝑛 (𝑥 ).
Hence, 𝑱𝒏 (−𝒙) = (−𝟏)𝒏 𝑱𝒏 (𝒙) for +𝒗𝒆 or −𝒗𝒆 integers.
Example 2:

Prove that 𝑱′𝟎 = −𝑱𝟏 .

Solution: Putting 𝑛 = 0 in the recurrence formula

𝑥𝐽𝑛′ = 𝑛𝐽𝑛 − 𝑥𝐽𝑛+1
We get, 𝑥𝐽0′ = −𝑥𝐽1
∴ 𝐽0′ = −𝐽1
Example 3:
Show that

𝟐
i) 𝑱−𝟏 (𝒙) = √ 𝒄𝒐𝒔𝒙
𝟐
𝝅𝒙

𝟐
ii) 𝑱𝟏 (𝒙) = √ 𝒔𝒊𝒏𝒙
𝟐
𝝅𝒙
𝟐 𝟐
𝟐
iii) ( 𝑱𝟏 (𝒙)) + ( 𝑱−𝟏 (𝒙)) = 𝝅𝒙
𝟐 𝟐

Solution:
We know that
𝑥𝑛 𝑥2 𝑥4
𝐽𝑛 (𝑥 ) = [1 − + + ⋯]
2𝑛 Γ(𝑛+1) 2(2𝑛+2) 2⋅4(2𝑛+2)(2𝑛+4)
1
i) Putting 𝑛 = − in (1), we get
2
 1
−
𝑥 𝑥2 2 𝑥4
𝐽𝑛 (𝑥 ) = 1 [1 − + + ⋯]
− 1 2 ⋅ 1 2 ⋅ 4 ⋅ 3 ⋅ 1
2 2Γ ( )
2
2 𝑥2 𝑥4 2
= √ [1 − + + ⋯ ] = √ 𝑐𝑜𝑠𝑥
𝜋𝑥 2! 4! 𝜋𝑥
1
ii) Putting 𝑛 = in (1), we get
2
1
𝑥2 𝑥2 𝑥4
𝐽𝑛 (𝑥 ) = 1 [1 − + + ⋯]
3 2 ⋅ 3 2 ⋅ 4 ⋅ 3 ⋅ 5
22 Γ ( )
2
𝑥 1 1 𝑥3 𝑥5 2
=√ ⋅ [𝑥 − + + ⋯ ] = √ 𝑠𝑖𝑛𝑥
2 1⋅ 𝜋 𝑥 3! 5! 𝜋𝑥
2 √
iii) Squaring and adding (1) and (2) we get the result.
𝟐 𝟐
𝟐
( 𝑱𝟏 (𝒙)) + ( 𝑱 𝟏 (𝒙)) =
𝟐 −𝟐 𝝅𝒙

Example 4:
Prove that

𝟐 𝒄𝒐𝒔𝒙
i) 𝑱−𝟑 (𝒙) = √ (− − 𝒔𝒊𝒏𝒙)
𝟐
𝝅𝒙 𝒙

𝟐 𝒔𝒊𝒏𝒙
ii) 𝑱𝟑 (𝒙) = √ ( − 𝒄𝒐𝒔𝒙)
𝟐
𝝅𝒙 𝒙

Solution:
We know that
2
𝐽−1 (𝑥 ) = √ 𝑐𝑜𝑠𝑥 … … … … … . (1)
2 𝜋𝑥

2
𝐽1 (𝑥 ) = √ 𝑠𝑖𝑛𝑥 … … … … … . . (2)
2 𝜋𝑥
Also, we have the recurrence relation
2𝑛
𝐽𝑛−1 (𝑥 ) + 𝐽𝑛+1 (𝑥 ) =𝐽 (𝑥 ) … … … … . . (3)
𝑥 𝑛
1
i) Putting 𝑛 = − in (3), we get
2
1
𝐽 3 (𝑥 ) + 𝐽1 (𝑥 ) = − 𝐽−1 (𝑥 )
−2 2 𝑥 2
2 𝑐𝑜𝑠𝑥
∴ 𝐽−3 (𝑥 ) = √ (− − 𝑠𝑖𝑛𝑥), by (1), (2)
2 𝜋𝑥 𝑥

1
ii) Putting 𝑛 = in (3), we get
2
1
𝐽−1 (𝑥 ) + 𝐽3 (𝑥 ) = 𝐽1 (𝑥 )
2 2 𝑥 2
2 𝑠𝑖𝑛𝑥
∴ 𝐽3 (𝑥 ) = √ ( − 𝑐𝑜𝑠𝑥), by (1), (2)
2 𝜋𝑥 𝑥

Example 5:

Express 𝑱𝟒 (𝒙) in terms of 𝑱𝟎 and 𝑱𝟏

Solution:
From the recurrence relation
2𝑛
𝐽𝑛+1 (𝑥 ) = 𝐽 (𝑥 ) − 𝐽𝑛−1 (𝑥 )
𝑥 𝑛
we get by putting 𝑛 = 3, 2.
6
𝐽4 (𝑥 ) =
𝐽 (𝑥 ) − 𝐽2 (𝑥 )
𝑥 3
4
𝐽3 (𝑥 ) = 𝐽2 (𝑥 ) − 𝐽1 (𝑥 )
𝑥
6 4 24 6
∴ 𝐽4 (𝑥 ) = ( 𝐽2 − 𝐽1 ) − 𝐽2 = ( 2 − 1) 𝐽2 − 𝐽1
𝑥 𝑥 𝑥 𝑥
Again from (1),
2
𝐽2 (𝑥 ) =
𝐽 (𝑥 ) − 𝐽0 (𝑥 )
𝑥 1
48 8 24
∴ 𝐽4 (𝑥 ) = ( 3 − ) 𝐽1 (𝑥) − ( 2 − 1) 𝐽0
𝑥 𝑥 𝑥
Example 6:
Show that:

i) 𝒄𝒐𝒔𝒙 = 𝑱𝟎 − 𝟐𝑱𝟐 + 𝟐𝑱𝟒 − ⋯ and

ii) 𝒔𝒊𝒏𝒙 = 𝟐𝑱𝟏 − 𝟐𝑱𝟑 + 𝟐𝑱𝟓 − ⋯

Solution:
We know that
1
𝑥(𝑧− )
𝑧 1 1 1
𝑒 2 = 𝐽0 + (𝑧 − ) 𝐽1 + (𝑧 2 + 2 ) 𝐽2 + (𝑧 3 − 3 ) 𝐽3 + ⋯
𝑧 𝑧 𝑧

Putting 𝑧 = 𝑒 𝑖𝜃 , we have
(𝑒 𝑖𝜃 −𝑒 −𝑖𝜃 )
𝑒𝑥 2

= 𝐽0 + (𝑒 𝑖𝜃 − 𝑒 −𝑖𝜃 )𝐽1 + (𝑒 2𝑖𝜃 + 𝑒 −2𝑖𝜃 )𝐽2

+ (𝑒 3𝑖𝜃 − 𝑒 −3𝑖𝜃 )𝐽3 + ⋯
or,
𝑒 𝑥𝑖𝑠𝑖𝑛𝜃 = 𝐽0 + (2𝑖𝑠𝑖𝑛𝜃 )𝐽1 + (2𝑐𝑜𝑠2𝜃 )𝐽2 + (2𝑖𝑠𝑖𝑛3𝜃 )𝐽3 + ⋯
or, cos(𝑥𝑠𝑖𝑛𝜃 ) + 𝑖𝑠𝑖𝑛(𝑥𝑠𝑖𝑛𝜃 ) = (𝐽0 + 2𝑐𝑜𝑠2𝜃𝐽2 + 2𝑐𝑜𝑠4𝜃𝐽4 +
⋯ ) + 𝑖(2𝑠𝑖𝑛𝜃𝐽1 + 2𝑠𝑖𝑛3𝜃𝐽3 + 2𝑠𝑖𝑛5𝜃𝐽5 + ⋯ )
Equating the real and imaginary parts, we have
cos(𝑥𝑠𝑖𝑛𝜃 ) = 𝐽0 + 2𝑐𝑜𝑠2𝜃𝐽2 + 2𝑐𝑜𝑠4𝜃𝐽4 + ⋯
and
𝑠𝑖𝑛(𝑥𝑠𝑖𝑛𝜃 ) = 2𝑠𝑖𝑛𝜃𝐽1 + 2𝑠𝑖𝑛3𝜃𝐽3 + 2𝑠𝑖𝑛5𝜃𝐽5 + ⋯
 𝜋
Putting 𝜃 = , we have
2

𝑐𝑜𝑠𝑥 = 𝐽0 − 2𝐽2 + 2𝐽4 − ⋯

and
𝑠𝑖𝑛𝑥 = 2𝐽1 − 2𝐽3 + 2𝐽5 − ⋯
Example 7:
Show that when 𝒏 is integral,
𝝅
a) 𝝅𝑱𝒏 = ∫𝟎 𝒄𝒐𝒔(𝒏𝜽 − 𝒙𝒔𝒊𝒏𝜽)𝒅𝜽
𝝅 𝝅
b) 𝝅𝑱𝟎 = ∫𝟎 𝒄𝒐𝒔(𝒙𝒔𝒊𝒏∅)𝒅𝜽 = ∫𝟎 𝒄𝒐𝒔(𝒙𝒔𝒊𝒏∅)𝒅∅ and hence
deduce that
𝒙𝟐 𝒙𝟒 𝒙𝟔
𝑱𝟎 ( 𝒙 ) = 𝟏 − 𝟐 + 𝟐 𝟐 − 𝟐 𝟐 𝟐 + ⋯
𝟐 𝟐 ⋅𝟒 𝟐 ⋅𝟒 ⋅𝟔
∞
(−𝟏)𝒓 𝒙𝟐𝒓
=∑ .
(𝟐𝒓 𝒓!)𝟐
𝒓=𝟎

Solution:
a) We shall use the following result:
𝜋 𝜋 𝜋
𝑤ℎ𝑒𝑛 𝑚=𝑛
∫ 𝑐𝑜𝑠𝑚∅𝑐𝑜𝑠𝑛∅𝑑∅ = ∫ 𝑠𝑖𝑛𝑚∅𝑠𝑖𝑛𝑛∅𝑑∅ = {02 𝑤ℎ𝑒𝑛 𝑚=𝑛
0 0

cos(𝑥𝑠𝑖𝑛𝜃 ) = 𝐽0 + 2𝑐𝑜𝑠2𝜃𝐽2 + ⋯ + 2𝑐𝑜𝑠2𝑚𝜃𝐽2𝑚 + ⋯

…………….. (2)
and
𝑠𝑖𝑛(𝑥𝑠𝑖𝑛𝜃 )
= 2𝑠𝑖𝑛𝜃𝐽1 + 2𝑠𝑖𝑛3𝜃𝐽3 + ⋯ + 2 sin(2𝑚 + 1) 𝜃𝐽2𝑚+1 + ⋯
………………….. (3)
Now multiplying both sides of (2) by 𝑐𝑜𝑠2𝑚𝜃 and then integrating
between the limits 0 to 𝜋, we get
 𝜋
∫ 𝑐𝑜𝑠(𝑥𝑠𝑖𝑛𝜃)𝑐𝑜𝑠2𝑚𝜃𝑑𝜃
0
𝜋 𝜋
= 𝐽0 ∫ 𝑐𝑜𝑠2𝑚𝜃𝑑𝜃 + 2𝐽2 ∫ 𝑐𝑜𝑠2𝜃𝑐𝑜𝑠2𝑚𝜃𝑑𝜃 + ⋯
0 0
𝜋
+ 2𝐽2𝑚 ∫ cos2 2𝑚𝜃𝑑𝜃 + ⋯
0
𝜋
= 2 ⋅ 𝐽2𝑚 ⋅ = 𝜋𝐽2𝑚 … … … . . (4)
2
Similarly multiplying both sides of (2) by cos (2𝑚 + 1)𝜃 and then
integrating between the limits 0 to 𝜋 and using (1), we have
𝜋
∫ 𝑐𝑜𝑠(𝑥𝑠𝑖𝑛𝜃 )cos (2𝑚 + 1)𝜃𝑑𝜃 = 0 … … … … … (5)
0

Again, multiplying both sides of (3) by sin (2𝑚 + 1)𝜃 and then
integrating between the limits 0 to 𝜋, we get
𝜋
∫ 𝑠𝑖𝑛(𝑥𝑠𝑖𝑛𝜃)sin (2𝑚 + 1)𝜃𝑑𝜃
0
𝜋 𝜋
= 𝐽1 ∫ 𝑠𝑖𝑛𝜃sin (2𝑚 + 1)𝜃𝑑𝜃 + 2𝐽2 ∫ 𝑠𝑖𝑛3𝜃sin (2𝑚 + 1)𝜃𝑑𝜃 + ⋯
0 0
𝜋
+ 2𝐽𝑚+1 ∫ sin2 (2𝑚 + 1) 𝜃𝑑𝜃 + ⋯
0
𝜋
= 2 ⋅ 𝐽2𝑚+1 ⋅ = 𝜋𝐽2𝑚+1 … … … (6)
2
Similarly, we get
𝜋
∫ 𝑠𝑖𝑛(𝑥𝑠𝑖𝑛𝜃 )sin 2𝑚𝜃𝑑𝜃 = 0 … … … … … (7)
0

Therefore,
𝜋
∫ 𝑐𝑜𝑠(2𝑚𝜃 − 𝑥𝑠𝑖𝑛𝜃 )𝑑𝜃
0
 𝜋 𝜋
= ∫ 𝑐𝑜𝑠2𝑚𝜃𝑐𝑜𝑠 (𝑥𝑠𝑖𝑛𝜃 )𝑑𝜃 + ∫ 𝑠𝑖𝑛2𝑚𝜃𝑠𝑖𝑛(𝑥𝑠𝑖𝑛𝜃 )𝑑𝜃
0 0

= 𝜋𝐽2𝑚 + 0, by (4) & (7)

= 𝜋𝐽2𝑚 … … … … . . (8)
Also,
𝜋
∫ cos[(2𝑚 + 1) 𝜃 − 𝑥𝑠𝑖𝑛𝜃 ]𝑑𝜃
0
𝜋
= ∫ cos (2𝑚 + 1)𝜃𝑐𝑜𝑠(𝑥𝑠𝑖𝑛𝜃 )𝑑𝜃
0
𝜋
+ ∫ sin (2𝑚 + 1)𝜃𝑠𝑖𝑛(𝑥𝑠𝑖𝑛𝜃)𝑑𝜃
0

= 𝜋𝐽2𝑚+1 , by (5) & (6) ……………… (9)

So, in virtue of (8) and (9), for all positive integral values of 𝑛, we have
𝜋
∫ 𝑐𝑜𝑠(𝑛𝜃 − 𝑥𝑠𝑖𝑛𝜃 )𝑑𝜃 = 𝜋𝐽𝑛
0

If 𝑛 be a negative integer, say 𝑛 = −𝑚, where 𝑚 is positive, then

𝜋
∫ 𝑐𝑜𝑠(𝑛𝜃 − 𝑥𝑠𝑖𝑛𝜃 )𝑑𝜃
0
𝜋
= ∫ 𝑐𝑜𝑠(−𝑚𝜃 − 𝑥𝑠𝑖𝑛𝜃 )𝑑𝜃
0
𝜋
= − ∫ cos {−𝑚(𝜋 − 𝜑) − 𝑥𝑠𝑖𝑛(𝜋 − 𝜑)}𝑑𝜑 , 𝑝𝑢𝑡𝑡𝑖𝑛𝑔 𝜃 = 𝜋 − 𝜑
0
𝜋
= ∫ cos {𝑚𝜋 − (𝑚𝜑 − 𝑥𝑠𝑖𝑛𝜑)}𝑑 𝜑
0
𝜋
= (−1 )𝑚 ∫ cos (𝑚𝜑 − 𝑥𝑠𝑖𝑛𝜑)𝑑 𝜑
0

= (−1)𝑚 𝜋𝐽𝑚 , by (10)

 = 𝜋𝐽−𝑚 (∵ 𝐽−𝑚 (𝑥 ) = (−1)𝑚 𝐽𝑚 (𝑥 ))
= 𝜋𝐽𝑛 … … … … … (11)
Hence, in virtue of (10) and (11), for all integral values of 𝑛,
𝜋
∫ 𝑐𝑜𝑠(𝑛𝜃 − 𝑥𝑠𝑖𝑛𝜃 )𝑑𝜃 = 𝜋𝐽𝑛
0

b) From (2), we have

𝜋
∫ 𝑐𝑜𝑠(𝑥𝑠𝑖𝑛𝜑)𝑑𝜑
0
𝜋 𝜋
= 𝐽0 ∫ 𝑑𝜑 + 2𝐽2 ∫ 𝑐𝑜𝑠2𝜑𝑑𝜑 + ⋯ + ⋯
0 0
= 𝜋𝐽0 .
𝜋
Now putting 𝜃 = + 𝜑 in (2), we get
2

𝑐𝑜𝑠(𝑥𝑠𝑖𝑛𝜑) = 𝐽0 − 2𝐽2 𝑐𝑜𝑠2𝜑 + 2𝐽4 𝑐𝑜𝑠4𝜑 − ⋯

𝜋
∴ ∫ 𝑐𝑜𝑠(𝑥𝑐𝑜𝑠𝜑)𝑑𝜑
0
𝜋 𝜋
= 𝐽0 ∫ 𝑑𝜑 + 2𝐽2 ∫ 𝑐𝑜𝑠2𝜑𝑑𝜑 + ⋯ + ⋯
0 0
= 𝜋𝐽0 .
Hence,
1 𝜋
𝐽0 = ∫ 𝑐𝑜𝑠(𝑥𝑐𝑜𝑠𝜑)𝑑𝜑
𝜋 0
1 𝜋 𝑥 2 cos2 𝜑 𝑥 4 cos4 𝜑
= ∫ (1 − + − ⋯ ) 𝑑𝜑
𝜋 0 2 4
1 𝑥2 1 𝑥4 1 ⋅ 3 𝑥6 1 ⋅ 3 ⋅ 5
= [𝜋 − ⋅ 𝜋 + ⋅ 𝜋− ⋅ 𝜋 +⋯]
𝜋 2 2 4 2⋅4 6 2⋅4⋅6

𝜋
2𝑟
1 ⋅ 3 ⋅ 5 ⋅⋅⋅ (2𝑟 − 1)
∵ ∫ cos 𝜑 𝑑𝜑 = 𝜋
0 2 ⋅ 4 ⋅ 6 ⋅⋅⋅ 2𝑟
 𝑥2 𝑥4 𝑥6
=1− 2+ 2 2− 2 2 2+⋯
2 2 ⋅4 2 ⋅4 ⋅6
∞
(−1)𝑛 𝑥 2𝑟
=∑
(2𝑟 𝑟!)2
𝑟=0

Example 8:

Show that 𝑱𝒏 (𝒙) = 𝟎 has no repeated roots except at 𝒙 = 𝟎.

Solution:
If possible, suppose 𝐽𝑛 (𝑥 ) = 0 has repeated roots. Then at least two
roots must be equal, say 𝛼 i.e., 𝛼 is a double root of 𝐽𝑛 (𝑛). Then from
the theory of equations, we have
𝐽𝑛 (𝛼 ) = 0 𝑎𝑛𝑑 𝐽𝑛′ (𝛼 ) = 0 … … … . (1)
We have the recurrence relations,
𝑛
𝐽𝑛−1 (𝑥 ) = 𝐽𝑛 (𝑥 ) + 𝐽𝑛′ (𝑥 )
𝑥
𝑛
𝐽𝑛+1 (𝑥 ) = 𝐽𝑛 (𝑥 ) − 𝐽𝑛′ (𝑥 )
𝑥
Putting 𝑥 = 𝛼 in (2) and (3) and using (1), we get
𝐽𝑛+1 (𝛼 ) = 0 and 𝐽𝑛−1 (𝛼 ) = 0, except when 𝑥 = 0.
Since two different power series have distinct sum function, so
𝐽𝑛+1 (𝛼 ) = 0 = 𝐽𝑛−1 (𝛼 ) must be absurd. Hence 𝐽𝑛 (𝑥 ) = 0 has no
repeated roots except at 𝑥 = 0.