Download as pdf or txt
Download as pdf or txt
You are on page 1of 41

Bessel’s Differential equation

(continued)

Aug 2, 2020 MATH C241 Prepared by MSR 1


In this lecture we study properties of
Bessel’s functions, which are solutions of
Bessel’s equation.

Aug 2, 2020 MATH C241 Prepared by MSR 2


Zeros of Bessel Functions
Fact:
(i) If 0  p < 1/2, then every solution of the
Bessel’s equation has an infinite number
of positive zeros and the distance between
successive zeros is <  and  as x .
(ii) If p = 1/2, then every solution of the
Bessel’s equation has an infinite number
of positive zeros and the distance
between successive zeros is .
Aug 2, 2020 MATH C241 Prepared by MSR 3
(iii) If p > 1/2, then every solution of the
Bessel’s equation has an infinite number
of positive zeros and the distance
between successive zeros is >  and 
as x .

Aug 2, 2020 MATH C241 Prepared by MSR 4


Problem 5 Page 357

Prove: (i) 2
J1/ 2 ( x)  sin x
x

(ii) 2
J  1/ 2 ( x)  cos x
x

Aug 2, 2020 MATH C241 Prepared by MSR 5


Proof of (i): We know
2 n 1/ 2

J1/ 2 ( x)   ( 1) n 1  x 
 
n 0 n !(n  1/ 2)!  2 
Now
 1  3  1  1
 n   !   n    n     n  
 2  2  2  2
 1  1  1
 n    n     n  
 2 2  2
...
Aug 2, 2020 MATH C241 Prepared by MSR 6
 1  1  3 1 1
 n    n    n   .........     
 2 2 2  2  2

 2n  1   2n  1   2n  3   1
    .........   
 2  2  2   2

13 ... (2n  1) (2n  1)


 n 1

2

Aug 2, 2020 MATH C241 Prepared by MSR 7


 1
Hence n ! n   !
 2

13 ... (2n  1)


12 ... n  n 1

2
2 4 ... 2n 13 ... (2n  1)
 n
 n 1

2 2
(2n  1)! 
 2 n 1
2
Aug 2, 2020 MATH C241 Prepared by MSR 8
Hence 2 n 1/ 2
2 n 1

2 n  x
J1/ 2 ( x)   ( 1)  
n 0 (2n  1)!   2 
2 n 1
2 
n x
  ( 1)
 x n 0 (2n  1)!
3 5 7
2  x x x  2
 x    ...   sin x
x  3! 5! 7!  x
The other part is similarly proved.
Aug 2, 2020 MATH C241 Prepared by MSR 9
Properties of Bessel functions of the First kind
We note the definition of Jp(x) :
2n p
(1)

1 n  x
J p ( x)  ( 1)  
n 0 n !(n  p )!  2 

(2) d p p

dx
 
x J p ( x) x J p  1 ( x)

Aug 2, 2020 MATH C241 Prepared by MSR 10


Proof : d  p 
1  x
2n p

n
LHS =  x  ( 1)   

dx  n 0 n !(n  p )!  2  
2 n 2 p
d  
n 1 x 
=   ( 1) 2n p 

dx  n 0 n !(n  p )! 2 
Differentiating term by term, we get
2 n 2 p  1
n 2n  2 p x

 
=   ( 1) 2n p 
 n 0 n !(n  p )! 2 
Aug 2, 2020 MATH C241 Prepared by MSR 11
2n p  1
p

1 n x
= x  ( 1) 2n p 1
n 0 n !(n  p  1)! 2
2n p 1
p

1
n  x
= x  ( 1)  
n 0 n !(n  p  1)!  2 

p
x J p  1 ( x) = RHS

Aug 2, 2020 MATH C241 Prepared by MSR 12


(3) d p
dx
  p

x J p ( x)  x J p 1 ( x)

Proof :
2n p
d  p n 1
 x 
LHS =  x  ( 1)   

dx  n 0 n !(n  p )!  2  
2n
d  n

1 x 
=   ( 1) 2n p 

dx  n 0 n !(n  p )! 2 
Aug 2, 2020 MATH C241 Prepared by MSR 13
Differentiating term by term, we get
2n 1
  n 2n x 
=   ( 1) 2n p 
 n 1 n !(n  p)! 2 
2n 1
  n 1 x  Replace n by n+1
=   (  1) 2n p 1 
 n 1 (n  1)!(n  p)! 2 
2 n 1
  n 1 1 x 
=   ( 1) 2 n  p 1 
 n 0 n !(n  p  1)! 2 
2 n  p 1
p

n1 x
= - x  ( 1) 2 n  p 1
n 0 n !(n  p  1)! 2
Aug 2, 2020 MATH C241 Prepared by MSR 14
2 n  p 1
 p

n 1  x
= - x  (  1)  
n 0 n !( n  p  1)!  2
 p
 x J p 1 ( x)
In particular, we note that
d
 J 0 ( x)   J1 ( x)
dx
d
 xJ1 ( x)  xJ 0 ( x)
dx
Aug 2, 2020 MATH C241 Prepared by MSR 15
Recurrence relation for Bessel functions
d p p
We have
dx
 x J p ( x)   x J p  1 ( x)
Differentiating the LHS, we get
p p 1 p
x J p ( x )  px J p ( x ) x J p  1 ( x )
Dividing throughout by xp, we get
p
J p ( x)  J p ( x) J p  1 ( x) ….(4)
x
Aug 2, 2020 MATH C241 Prepared by MSR 16
d p  p
We have
dx

x J p ( x)  x J p 1 ( x) 
Differentiating the LHS, we get
 p  p 1  p
x J p ( x)  px J p ( x)  x J p 1 ( x)
Multiplying throughout by xp, we get
p
J p ( x)  J p ( x)  J p 1 ( x ) …. (5)
x
Aug 2, 2020 MATH C241 Prepared by MSR 17
(4)+(5) gives 2 J p ( x) J p  1 ( x)  J p 1 ( x) (6)

(4)-(5) gives 2 p J ( x)  J ( x)  J ( x) (7)


p p 1 p 1
x
From Eq. (7), we get
the Recurrence relation for Jp(x), namely
p
J p 1 ( x) 2 J p ( x)  J p  1 ( x)
x
Aug 2, 2020 MATH C241 Prepared by MSR 18
For example, p = 1/2 gives
1/ 2
J 3/ 2 ( x) 2 J1/ 2 ( x )  J  1/ 2 ( x)
x
1 2 2
 sin x  cos x
x x x
For example, p = 2 gives
4 42 
J 3 ( x )  J 2 ( x )  J1 ( x )   J 1 ( x )  J 0 ( x )   J 1 ( x )
x xx 

Aug 2, 2020 MATH C241 Prepared by MSR 19


Fact: The positive zeros of Jp(x) and
Jp+1(x) alternate on the positive x-axis.
Proof: Let x1 and x2 be two successive zeros
of Jp(x). Hence Jp(x1) = Jp(x2) = 0 and Jp(x) is
not zero for any x between x1 and x2.
Hence J p ( x1 ) and J p ( x2 ) are of opposite
p
signs. Hence J p ( x)  x J p ( x)  J p 1 ( x) gives
J p 1 ( x1 ) and J p 1 ( x2 ) are of opposite signs.
Aug 2, 2020 MATH C241 Prepared by MSR 20
Or Jp+1(x) is zero at some point between x1 and
x2.
Now let x1 and x2 be two successive zeros of
Jp+1(x). Hence Jp+1(x1) = Jp+1(x2) = 0 and
Jp+1(x) is not zero for any x between x1 and x2.
Hence J p 1 ( x1 ) and J p 1 ( x2 ) are of opposite

sign. Now (4) (with p replaced by p+1) gives


p 1
J p 1 ( x)  J p 1 ( x) J p ( x)
x
Aug 2, 2020 MATH C241 Prepared by MSR 21
Hence J p ( x1 ) and J p ( x2 ) are of opposite signs.

Or Jp(x) is zero at some point between x1 and


x2.
That is between any two successive zeros of
Jp(x), there is a zero of Jp+1(x) and between
any two successive zeros of Jp+1(x), there is a
zero of Jp(x).
Thus the positive zeros of Jp(x) and Jp+1(x)
alternate on the positive x-axis.
Aug 2, 2020 MATH C241 Prepared by MSR 22
Look at the differential equation
2 2 2 2
x y  xy  (a x  p ) y 0

 

Changing the independent variable x to t by


the substitution a x = t, the above equation
becomes the Bessel’s equation
2
2 d y dy 2 2
t 2
t  (t  p ) y 0
dt dt
whose solution is y c1 J p (t )  c2Yp (t )
Aug 2, 2020 MATH C241 Prepared by MSR 23
i.e. y c1 J p (ax)  c2Yp (ax)
c1, c2 arbitrary constants.

In particular, Jp(ax) is a solution of the


differential equation
2 2 2 2
x y  xy  (a x  p ) y 0

Aug 2, 2020 MATH C241 Prepared by MSR 24


Orthogonality of Bessel functions

Let 1  2  ...  n  ...


be the positive zeros of Jp(x) arranged in
increasing order. Then
1 0 m n

 J p (m x) J p (n x) x dx  1 2
x 0  2 J p 1 (n ) m n

Aug 2, 2020 MATH C241 Prepared by MSR 25


The first part of the above result states that
Jp(mx) and Jp(nx) are orthogonal over the
interval [0, 1] with respect to the weight
function x.
Proof of the Orthogonality properties.
By the remarks made earlier, u = Jp(ax)
satisfies the differential equation
2 2 2 2
x y  xy  (a x  p ) y 0

Aug 2, 2020 MATH C241 Prepared by MSR 26


2 2 2 2
i.e. x u  xu  (a x  p )u 0

 

Similarly if v = Jp(bx), then


2 2 2 2
x v  xv  (a x  p )v 0

 

Multiplying the first equation by v, the


second by u and subtracting, we get
2 2 2 2
x (u v  uv )  x(u v  uv )  (a  b ) x uv 0

 
  
i.e. x(u v  uv)  (u v  uv)  (a 2  b 2 ) xuv 0
Aug 2, 2020 MATH C241 Prepared by MSR 27
i.e. d
x (u v  uv)  (u v  uv)  (a 2  b 2 ) xuv 0
dx

or d 2 2
 x(u v  uv )  (a  b ) xuv 0
 
dx
Integrating with respect to x from 0 to 1,we get
1
2 2 1
(a  b ) x u v dx  x(uv  u v )  x 0
0

At the lower limit x = 0, the RHS is clearly 0.


Aug 2, 2020 MATH C241 Prepared by MSR 28
u (1) J p (a), v(1) J p (b)

Hence if a = n and b = m are distinct


positive zeros of Jp(x), then u(1) = v(1) = 0.
Hence 1
2 2
(a  b ) x u v dx 0
0

1
As a  b,
x u v dx 0
0

Aug 2, 2020 MATH C241 Prepared by MSR 29


Proof of the second part
u = Jp(ax) satisfies the differential equation
2 2 2 2
x y  xy  (a x  p ) y 0

 
2 2 2 2
i.e. x u  xu  (a x  p )u 0

 
Multiplying by 2u, we get
2 2 2 2 2
x 2uu   2 x (u )  2(a x  p )uu  0
i.e. d [ x 2 (u) 2 ]  d [(a 2 x 2  p 2 )u 2 ]  2a 2 xu 2 0
dx dx
Aug 2, 2020 MATH C241 Prepared by MSR 30
Integrating with respect to x from 0 to 1,we get
1
2 2 2 2 2 2 2 2 1
2a x u dx  
 x u   ( a x  p )u 
 x 0
0

If p > 0, u (0)  J p (0) 0


If p= 0, a 2 x 2  p 2 0 if x 0
Thus the RHS is zero at the lower limit x = 0.
d
Also u  J p (ax) aJ p (ax)
dx
Aug 2, 2020 MATH C241 Prepared by MSR 31
u (1)  J p (a ) 0
if a is a positive zero of Jp(x).
Hence the RHS = a2[Jp(a)]2 at the upper
limit x = 1.
1

Thus 2a 2 x u 2 dx a 2 [ J p (a )]2
0

1
2 1 2
or  x u dx  [ J 
p ( a )]
0
2
Aug 2, 2020 MATH C241 Prepared by MSR 32
p
We know J p ( x)  J p ( x)  J p 1 ( x)
x

Hence when x = a , a positive zero of Jp(x),


J p (a)  J p 1 (a )

So we get 1
12 2
 x [ J p (n x)] dx  [ J p 1 (n )]
0
2

Aug 2, 2020 MATH C241 Prepared by MSR 33


Fourier-Bessel Series Expansion
Let 1  2  ...  n  ...
be the positive zeros of Jp(x) arranged in
increasing order. Then given a piecewise
continuous function f(x), we can expand it as a
series: f ( x) a1 J p (1 x)  a2 J p (2 x)  ....

called the Fourier-Bessel series of f(x).

Aug 2, 2020 MATH C241 Prepared by MSR 34


Multiplying both sides by x Jp(nx) and
integrating with respect to x, from 0 to 1,
we get (noting that Jp(nx) and Jp(mx) are
orthogonal if n  m)
1
an 2
 x f ( x) J p (n x) dx  [ J p 1 (n )]
0
2
n =1, 2, …
Thus the coefficients ans are all found.
Aug 2, 2020 MATH C241 Prepared by MSR 35
Problem 5 Page 364
Find the Fourier-Bessel series expansion of
the function f ( x)  x p
in terms of the functions J p (n x)
( ns are the positive zeros of J p ( x) )
f ( x) a1 J p (1 x)  a2 J p (2 x)  ....
1
where an 2
 x f ( x) J p (n x) dx  [ J p 1 (n )]
0
2
Aug 2, 2020 MATH C241 Prepared by MSR 36
1 1
Now p 1
x f ( x) J
0
p (n x) dx x
0
J p (n x) dx

Put t = nx. We thus get


 n
n
1 p 1 1 d p 1
  p 2 t J p (t ) dt  p 2  (t J p 1 (t )) dt
0
n n 0
dt
1 p 1 1 an
 p 2
n J p 1 (n )  J p 1 (n )  [ J p 1 (n )]2
n n 2
2
gives an  (n = 1,2,…)
n J p 1 (n )
Aug 2, 2020 MATH C241 Prepared by MSR 37
Thus the Fourier-Bessel series expansion of xp
is
p

2
x  J p (n x)
n 1 n J p 1 (n )

… Next slide

Aug 2, 2020 MATH C241 Prepared by MSR 38


Problem 4 Page 363
Find the Fourier-Bessel series expansion of
the function  1
1 0  x  2

1 1
f ( x)  x
2 2
 1
 0  x  1
 2

in terms of the functions J 0 (n x)


( ns are the positive zeros of J 0 ( x) )
Aug 2, 2020 MATH C241 Prepared by MSR 39
f ( x) a1 J 0 (1 x)  a2 J 0 (2 x)  ....
1
where an 2
 x f ( x) J 0 (n x) dx  [ J1 (n )]
0
2
1 1/ 2
Now
x f ( x) J
0
0 (n x) dx  x J 0 (n x) dx
0

Put t = nx. We thus get


 /2 n / 2
n
1 1 d
  2 t J 0 (t ) dt  2  (t J1 (t )) dt
0
n n 0 dt
Aug 2, 2020 MATH C241 Prepared by MSR 40
1 n n  1 J ( n ) an 2
 2 J1 ( ) 1  [ J1 (n )]
n 2 2 2n 2 2
J1 (n / 2)
gives an  2 (n = 1,2,…)
n [ J1 (n )]

Thus the Fourier-Bessel series expansion of


f(x) is 
J1 (n / 2)
f ( x)  J 0 (n x)
n 1 n [ J1 (n )]

End of Lecture
Aug 2, 2020 MATH C241 Prepared by MSR 41

You might also like