ECE 6382

Fall 2020
David R. Jackson

Notes 20
Bessel Functions

Notes are from D. R. Wilton, Dept. of ECE

1
 Cylindrical Wave Functions
Helmholtz equation:

 k  0
2 2

In cylindrical coordinates:

  2 1  1  2  2  2
 2  2  2 k   0
      z
2


Separation of variables:

let    ,  , z   R     ( ) Z ( z )

Substitute into previous equation and divide by .

2
Cylindrical Wave Functions (cont.)
  2 1  1  2  2  2
 2  2  2 k   0
      z
2


   ,  , z   R     ( ) Z ( z )

 2R 1 R 1  2  2Z  2
 2 Z  Z  2 RZ  2 R   k RZ  0
      z
2


Divide by 
let
 R 1 R 1   Z  
 
 R  R 2     k 2
0
 Z 
3
Cylindrical Wave Functions (cont.)
 R 1 R 1  Z  
 
 R  R 2     k 2
0 (1)
 Z 

Therefore

Z  R 1 R 1 
 k 
2
  2
Z R  R  

f ( z) g ( , )

Hence,
f  z   constant  k z2
4
Cylindrical Wave Functions (cont.)
Hence
Z 
  k z2
Z

Z ( z )  h(k z z )   e  ikz z ,sin(k z z ),cos( k z z )

Next, to isolate the  -dependent term, multiply Eq. (1) by  2 :

 R 1 R 1  2
  
2
 2  kz   k 2  2  0
 R  R   

5
Cylindrical Wave Functions (cont.)

Hence

 2 1 R R 
   k  k z 
2 2
  (2)
   R R 

f ( ) g ( )
Note:
If  is allowed to
Hence change by 2,

 constant   2 then  must be
 an integer.

so   h ( )   e  i ,sin( ),cos( )

6
 Cylindrical Wave Functions (cont.)

From Eq. (2) we then have

 2 1 R R 
     k  k z 
2 2 2
 
  R R 

The next goal is to solve this equation for R().

First, multiply by R and collect terms:

 R   R    k  kz  R  R  0
2
  2 2 2 2

7
Cylindrical Wave Functions (cont.)
Define k 2  k 2  k z 2

 R   R   k      2  R  0

2 2
Then,
 

x  k 
Next, define
R(  )  y ( x)

dR dy dx
Note that R(  )    y ( x )k 
d  dx d 

and R( p )  y ( x ) k 2

8
Cylindrical Wave Functions (cont.)

Then we have 
2

x y  xy   x    y  0
 2 2

Bessel equation of order 

Two independent solutions: J ( x), Y ( x)

Hence y ( x)  AJ ( x)  BY ( x)

Therefore R(  )   J ( k   ), Y ( k   )

9
Cylindrical Wave Functions (cont.)
Summary

 k  0
2 2

   ,  , z   R    ( ) Z ( z )

Z ( z )   e  ik z z , sin(k z z ), cos(k z z )

   e  i , sin( ), cos( )

R(  )   J (k   ), Y ( k   )

k 2  k z2  k 2
10
 References for Bessel Functions

 M. R. Spiegel, Schaum’s Outline Mathematical Handbook, McGraw-Hill, 4th

Edition, 2012.

 M. Abramowitz and I. E. Stegun, Handbook of Mathematical Functions with

Formulas, Graphs, and Mathematical Tables, National Bureau of Standards,
Government Printing Office, Tenth Printing, 1972.

 NIST online Digital Library of Mathematical Functions (http://dlmf.nist.gov/).

 N. N. Lebedev, Special Functions & Their Applications, Dover Publications, New

York, 1972.

 G. N. Watson, “A Treatise on the Theory of Bessel Functions” (2nd Ed.),

Cambridge University Press, 1995.

11
 Properties of Bessel Functions
1
1

0.8 n=0 J (0) is finite

0.6 n=1
0
0.4 n=2
J0( x)

Jn (x)
J1( x) 0.2

Jn( 2  x)

0.2

0.4

 0.403 0.6
0 1 2 3 4 5 6 7 8 9 10
0 x 10

x
Note : J 0  0   0.
12
 Bessel Functions (cont.)
1
0.521

0
n=0
n=1
1
n=2
2

Yn (x)
Y0( x)

Y1( x)

Yn( 2  x)
3
Y (0) is infinite
4

 6.206 7
0 1 2 3 4 5 6 7 8 9 10
0 x 10

x
Note : Y0  0  tends to infinity slower than the rest.
13
 Bessel Functions (cont.)
Frobenius solution*:

 1
 2k
k

x
J ( x )    
k  0 k !   k  !  2 

z !    z  1
This series always converges.

Note : J  z  is analytic (a Taylor series) for   n (integer ).

* The Frobenius series is a generalized Taylor series that has a non-integer set of powers.
14
 Bessel Functions (cont.)

Non-integer order: n

y ( x)   J ( x), J  ( x) There are two valid solution.

Bessel equation is unchanged by   

Note:
J  ( x) is a always a valid solution

These are linearly independent when  is not an integer:

J ( x )  A1 x , J  ( x )  A2 x  as x0
15
 Bessel Functions (cont.)
Bessel function of second kind:

J ( x) cos( )  J  ( x)
Y ( x) 
sin( )
  …-2, -1, 0, 1, 2 …

(This definition gives a “nice” asymptotic behavior as x  .)

Yn ( x )  lim Y ( x )
 n

16
 Bessel Functions (cont.)
From the Frobenius solution we have:
Notice the branch point at z = 0.

  x   1 n 1  n  k  1 !  x 
2k n
2
Yn ( x)  J n  x  ln         
   2    k 0 k !  2
 2k n
1 1  x
   1   k     n  k  
k
 
 k 0 k ! n  k  !  2 

(Schaum’s Outline Mathematical Handbook, Eq. (24.9))

where   0.577216
1 1 1 (Euler’s constant)
  p  1    ( p  0),   0   0
2 3 p
17
 Bessel Functions (cont.)

Integer order:  n

Symmetry property:

J  n ( x)  (1) n J n ( x)
(They are no longer linearly independent.)
Y n ( x)  (1) Yn ( x )
n

18
 Bessel Functions (cont.)
Small-Argument Properties (x  0):

J n ( x)  Ax n , n  0,1, 2

J ( x)  Ax ,   1, 2, 3,... J n ( x )  (  1) n
J n ( x) 

Y0 ( x)  C ln  x  Examples:

Yn ( x)  Dx  n , n  1, 2,3 J 0 ( x)  1


J1  x   x / 2
Y ( x)  Bx , 0

For order zero, the Bessel function of the second kind behaves logarithmically
rather than algebraically.

19
 Bessel Functions (cont.)

Asymptotic Formulas:

x

2    
J ( x ) ~ cos  x   
x  2 4

2    
Y ( x ) ~ sin  x   
x  2 4

20
 Hankel Functions
 1
H ( x )  J ( x)  iY ( x)
 2
H ( x)  J ( x)  iY ( x)

As x
 
 1 2  i ( x   )
Incoming wave
H ( x ) ~ e 2 4
x
 
 2 2  i ( x   )
H ( x) ~ e 2 4 Outgoing wave
x

These are valid for arbitrary order .

21
 Hankel Functions (cont.)

Useful identity:

H n ( z )   1
 2
H n  ( z )
n 1 1

Im  z   0

This is a symmetry property of the Hankel function.

N. N. Lebedev, Special Functions & Their Applications, Dover Publications, New

York, 1972.

22
 Generating Function
The integer order Bessel function of the first kind can also be
defined through a generating function g(x,t):
x  1 
t 
g  x, t   e 2 t 
 
n 
Jn  x tn

The generating function definition leads to a number of useful

identities and representations:

ikx
  i  J n  k   e in
n
e  (plane wave expansion, a.k.a.Jacobi - Anger expansion)
n 

1
 J m      cos   sin   m  d (integral representation of Bessel function)
 0

(Please see next slides.)

23
 Generating Function (cont.)
Plane-wave expansion:

k  iei ie i x  t 1 
 e ikx  e
 i  k   cos
  e2
  e 
2  t 


x k  ,
t iei

Hence, from the generating function identity we have:


e ikx   J  x t
n 
n
n
x k  ,
t  iei
so


ikx
  i  J n  k   e in
n
e 
n 

24
 Generating Function (cont.)

Integral representation of Bessel function:


ik  cos
  i  J n  k   e in
n
Start with: e 
n 

 Set k  1 in above result, multiply both sides by e  im

 Integrate over  0, 2  , use orthogonality of ein

2
m
 e  i  cos im
e d  2  i  Jm   
0

2
im
Jm      e  i  cos im
e d
2 0

25
 Generating Function (cont.)
Integral representation of Bessel function (cont.):

2
im
Jm      e  i  cos im
e d
2 0

 Use       / 2, symmetry,     :

2  /2
im i  cos   /2 im   /2 
Jm      e e d 
2  /2
2  /2
im  i  sin   im 
  i 
m
 e e d 
2  /2
2
1  i sin   im 

2 e
0
e d  Periodic integrand

26
 Generating Function (cont.)
Integral representation of Bessel function (cont.):
2
1  i  sin   im 
Jm     e e d 
2 0

1  i  sin    im 

2 

e e d 


1

2  cos   sin    m   i sin   sin    m  d 


1
2 
 cos   sin    m  d 
odd function

1
  cos   sin    m  d
 0

1
J m      cos   sin   m  d
 0
27
 Recurrence Relations
Many recurrence relations can be derived from the generating function.
x 1 
t 
g  x, t   e 2 t 
 J
n 
n ( x) t n
x 1
 2  t  t  

t
e   nJ
n 
n ( x ) t n 1

x 1
x 1  2  t  t 
 1  2  e   n J n ( x ) t n 1
2 t  n 

x 1  
  1  2   J n ( x ) t   n J n ( x ) t n 1
n

2  t  n  n 

x  
  J n ( x )  t  t    n J n ( x ) t n 1
n n 2

2 n  n 

On LHS use: n  n  1 & n  n  1

x  

 n1
2 n 
 J ( x )  J n 1 ( x )  t n 1
  n
n 
n J ( x ) t n 1

28
 Recurrence Relations (cont.)

x  

  J n1( x )  J n 1( x )  t   n J n ( x ) t n1

n 1

2 n  n  

Equating like powers of t yields:

2n
J n 1 ( x )  J n 1 ( x )  J n ( x)
x

29
 Recurrence Relations (cont.)

2n
J n 1 ( x )  J n 1 ( x )  J n ( x)
x

This can be used to generate other useful recurrence relations:

2  n  1
n  n  1: J n ( x )  J n 1 ( x )  J n 2 ( x ) ("upward recursion")
x
2  n  1
n  n  1: J n ( x )  J n 1 ( x )  J n 2 ( x ) ("downward recursion")
x

30
 Recurrence Relations (cont.)
Another recurrence relation for the derivative of the Bessel function:
x 1 
t 
g  x, t   e 2 t 
 J
n 
n ( x) t n
x 1
t  x 1
 e 2 t 
1  1  2  t  t  1  1  
 g  x, t    t  e   t    J n ( x) t n
x x 2 t  2  t  n 

1  

   J n ( x) t   J n ( x) t n 1 
n 1

2  n  n  
1  
n

1
   J n 1 ( x) t   J n 1 ( x) t     J n 1 ( x)  J n 1 ( x)  t n
n

2  n  n   n  2

Also, we have
 
g  x, t    J n ( x ) t n
x n 

1
Equating like powers of t yields: J n ( x )   J n1 ( x )  J n1 ( x )
2
31
 Recurrence Relations (cont.)
1
J n ( x )   J n 1 ( x)  J n 1 ( x )
2

Then use the previous identity:

2n
J n 1 ( x )  J n 1 ( x )  J n ( x)
x

This can be used to replace Jn+1 or Jn-1.

This yields:

n
J n ( x )  J n 1 ( x )  J n ( x)
x
n
J n ( x )   J n 1 ( x )  J n ( x )
x

32
 Recurrence Relations (cont.)
The same recurrence formulas actually apply to all Bessel
functions of all orders.

If Z (x) denotes any Bessel, Neumann, or Hankel function of order

, then we have:

2
Z 1 ( x )  Z 1 ( x )  Z ( x )
x

2    1
Z ( x )  J 1 ( x )  J 2 ( x )
x
2    1
Z ( x )  J 1 ( x )  J  2 ( x )
x

 
Z  Z 1  Z Z   Z 1  Z
x x

33
 Recurrence Relations (cont.)
Integral identities also follow from the recurrence identities.

Example of integral identity:


Z  Z 1  Z  xZ   Z  xZ 1 (multiplying both sides by x)
x
Multiply by x 1  x Z   x 1Z  x Z 1

Hence,
d 
x Z   x 1Z   x Z ( x )   x Z 1 ( x ) 
 
 x Z  1 ( x ) dx  x Z ( x )
dx

  1
 
x Z ( x ) dx  x Z ( x )

Similarly, we have

  1
 
x Z ( x ) dx   x Z ( x )

34
 Recurrence Relations (cont.)

Examples:

 J  x  x dx  x J  x 
0 1 (First one,  = 1)

 J  x  dx   J  x 
1 0 (Second one,  = 0)

35
 Wronskians
From the Sturm-Liouville properties, the Wronskians for the Bessel
differential equation are found to have the following form:
x Recall :
dx x
 x
 x
 ln  
a
 ln   W (a ) a C   p ( x) dx
W ( x)  W (a ) e a
 W (a ) e a
 W (a ) e  x
  W ( x)  W ( a ) e a

x x p1  x 
p( x) 
p0  x 

The constant C can be found using the small-argument x 2 y   xy   x 2   2  y  0

approximations for the Bessel functions (keep k = 0 term p0  x   x 2

in the Frobenius series). p1  x   x

2sin
W  x   W  J , J    J   x  J   x   J  x  J   x   
x
(Please see next slide.)

Note: For   n, the Wronskian is not identically zero (in fact, it is not zero
anywhere), and hence the two functions are linearly independent.
36
 Wronskians (cont.)
 1
 2k
 k
 
x
J ( x )    
k 0 k !    k  !  2 

 
1 x 1 x
J ( x )    , J  ( x ) 
! 2     !  2 

1     x 
 1  1
1  x
J ( x )    , J  ( x )   
2 ! 2  2    !  2 

  2
W  x   J   x  J   x   J   x  J   x     
 !    ! x  !    ! x  !    ! x
2sin  Recall :
C  
 ( z ) (1  z ) 
sin  z

 ( ) (1   ) 
2sin  sin 
W  x       1 !    ! 

x 
sin 
    !    ! 
sin  37
 Wronskians (cont.)

Similarly, we have

2
W  J , Y   J   x  Y  x   J   x  Y  x  
x
2i
W  J , H(1,2)   J  x  H(1,2) x   J   x  H(1,2) x   
x

 for H(1)
 for H(2)

38
 Fourier-Bessel Series
Fourier-Bessel series:

Note:

 
f (  )   c n J  p n  , 0 a The order  and
the length a are
n 1  a arbitrary here.

where p n is the n th zero of J ( x ) : J ( p n )  0.

The coefficients are given by

a
2  
c n 
a 2  J 1  p n   2 
0
f (  ) J  p n   d 
 a

(Please see next slide.)

39
 Fourier-Bessel Series
To derive the Fourier-Bessel expansion, start with:

 
f (  )   c n J   p n  , 0 a
n 1  a

 
Multiply both sides by    m  
J p and integrate from 0 to a:
 a

    
a a
   
 f (  ) J  p m   d      c n J  p n  J   p m    d 
0  a 0  n 1  a  a 
 a
   
   c n J  p n  J  p m   d 
n 1 0  a  a
a2
 c m  J 1  p m   2 (using orthogonality +
2 result from next slide)

Note:
See Notes 18 for a derivation of the orthogonality when m  n.
40
 Fourier-Bessel Series (cont.)
Derivation of the orthogonality formula

Start with this integral identity (derivation omitted):

x2 x2  2 
J  px  xdx   J   px     1  2    
2 2 2
J px
v
2 
2   px  

Choose p  p m / a :
 
a 2   2
2  p m  a2   p m   a 2  2    p m 
0 v  a
J x  xdx   J 
 2  
a  
a 
1 J
2   
2   p m     a
a 

  a 
  a  
a2
  J  p m  
2

2
2 Recall :
a2   
   J 1 ( p m )  J ( p m )  
2  p m  J ( x )   J 1 ( x )  J ( x)
x
a2
   J 1 ( p m ) 
2

2
41
 Addition Theorems
y
  Addition theorems allow cylindrical
y harmonics in one coordinate system to
  be expanded in terms of those of a
0 x
 0 shifted coordinate system.
0
x

Shifting from global origin to local origin:


i n m   0
J n (k  ) e in
 J n  m ( k  0 )e J m (k  ) e im 
m 

 

i  n m   0 im 
 H (2)
n m ( k  0 ) e J m ( k  ) e ,   0
 m 
H n(2) ( k  ) ein  


i nm   0
J ( k  ) e H m(2) (k  ) eim  ,    0
 m  n m 0

42
 Addition Theorems (cont.)
y
 
y
 
0 x
 0
0
x

Shifting from local origin to global origin:



i  n  m    0 
J n ( k  ) e in 
 
m 
J n  m ( k  0 )e J m (k  ) e im

  i  n  m     0 
 
 m 
H (2)
nm ( k  0 ) e J m ( k  ) e im
,   0
H n(2) (k  ) ein   
 i  n  m     0 

 m 
J nm ( k  0 ) e H m(2) (k  ) eim ,    0

43
 Addition Theorems (cont.)
y
 
y
Recall :
 
0 x J  m  x    1 J m  x 
m

 0
0
H  2m  x    1 H m 2  x 
m
x
e  im   1
m

Shifting from local origin to global origin:

Special case (n = 0):


J 0 (k  )  J
m 
m (k  0 ) J m (k  ) e
im  0 

 
 
 m
H (2)
m ( k  0 ) J m ( k  ) e
im  0 
,   0
H 0(2) (k  )   

 m
im   0 
J m ( k  0 ) H (2)
m ( k  ) e ,   0

44