Math 456 Lecture Notes:

Bessel Functions and their Applications to

Solutions of Partial Dierential Equations
Vladimir Zakharov
June 3, 2009
1 Gamma Function
Gamma function (s) is dened as follows:
(s) =
_

0
e
t
t
s1
dt (1)
As far as:
t
s1
=
1
s

t
t
s
(2)
By plugging (2) into (1) we get
s(s) =
_

0
e
t
d
dt
t
s
dt = e
t
t
s
|

0
+
_

0
e
t
t
s
dt (3)
or
s(s) = (s + 1) (4)
Then (1) = 1 and (2) = 1.
By induction we obtain:
(n + 1) = n! (5)
Then
(s) =
1
s
(s + 1)
(s)
1
s
if s 0
(s 1) =
1
s(s 1)
(s + 1)
(s n) =
1
(s n)(s n + 1) s
(s + 1) (6)
(s) has holes at all negative integral values of s.
To nd asymptotic behavior of Gamma-function as s , we use so called Laplace
Method.
2
(s + 1) =
_

0
e
t
t
s
dt =
_

0
e
(t,s)
dt (7)
(t, s) = t s ln(t)
Function (t, s) has a minimum at t = s. Indeed:

t
= 1
s
t
= 0 if t = s (8)
Near this minimum:
=
o
(s) +
1
2

(s)
2
+ , = t s

o
(s) = s s ln(s)

(s) =
1
s
Now we will replace in (7) (t, s) to its approximate value (8) and go from integration
by t to integration by . Whithout loss of accuracy we can consider that < <
.
Then
(s + 1) e
o(s)
_

2
2s
d (9)
e
o(s)
=
_
s
l
_
s
Now we replace =

2sy and remember that

_

e
y
2
dy =

. We end up with
the following answer:
(s + 1)

2s
_
s
l
_
n
(10)
n!

2n
_
n
l
_
n
3
This is the Stirling approximation for n = 5, where n! = 120. The Stirling
approximation gives 5! 118.045. The accuracy of the Stirling approximation is
reasonable. We accept without proof:
(x)(1 x) =

sin(x)
(11)
where
2
(
1
2
) = so (
1
2
) =

.
2 Bessel Equation Appears
Let us try to solve the diusion equation
u
t
= u (12)
inside the disk of radius a in polar coordinates:
u =
1
r

r
r
u
r
+
1
r
2

2
u

2
(13)
We impose boundary conditions u(r = a) = 0 with initial data u(t = 0) = (r, ).
In polar coordinates the previous equation becomes:
u
t
=
_
1
r

r
r
u
r
+
1
r
2

2
u

2
_
(14)
Partial solutions to this equation can be found of the following form:
u(r, , t) = e
in
e
tk
2
R(r) (15)
The radial part R(r) satises the equation
1
r

r
r
R
r
+
_
k
2

n
2
r
2
_
R = 0 (16)
k
2
can take discrete values k
2
= k
2
1
, , k
2
N
,
Corresponding radial functions R
N
(r) satisfy the Dirichlet condition R
N
(a) = 0.
By change of variables z = kr we have:
4
1
z

z
z
R
z
+
_
1
s
2
z
2
_
R = 0 (17)
where we have replaced n
2
= s
2
, assuming that s is an arbitrary real number.
The previous equation is the Bessel equation. At z 0 it becomes the equipotent
equation:
1
z

z
z
R
z

s
2
z
2
R = 0 (18)
which can be solved explicitly:
R = C
1
z
s
+ C
2
z
s
(19)
One can seek a solution of (17) in the form
R =
_
z
2
_
s
F(z, s) (20)
F satises the equation:
F

+
2s + 1
z
F

+ F = 0 (21)
The solution of equation (21) can be found in the form of series:
F =

k=0
C
k
_
z
2
_
2k
(22)
After dierentiating by z, the rst term in (22) vanishes. One can see that:
F

+
2s + 1
z
F

k=1
C
k
[(2k 1)2k + 2k(2s + 1)]
4
_
z
2
_
2(k1)
(23)
Let us replace k k + 1. Now:
F

+
2s + 1
z
F

k=0
C
k+1
(k + 1)(k + s + 1)
_
z
2
_
2k
(24)
5
By substitution we will nally get:
C
k+1
(k + 1)(k + s + 1) + C
k
= 0 (25)
or
C
k+1
=
C
k
(k + 1)(k + s + 1)
(26)
This can then be solved as follows:
C
k
=
(1)
k
(k + 1)(k + s + 1)
=
(1)
k
k!(k + s + 1)
(27)
In particular for integral s = n:
C
k
=
(1)
k
k!(n + k)!
(28)
3 Bessel Function
The Bessel function J
s
(z) is dened by the series:
J
s
(z) =
_
z
2
_
s

k=0
(1)
k
k!(s + k + 1)
_
z
2
_
2k
(29)
This series converges for all z on the complex plane, thus J
s
(z) is the entire function.
If z 0, then
J
s
(z)
_
z
2
_
s
1
(s + 1)
(30)
If s
2
is not an integer, then J
s
(z) is the second solution of the Bessel equation.
Now:
J
s
(z)
_
z
2
_
s
1
(s + 1)
(31)
J
s
(z) is regular at z 0, while J
s
(z) is singular. So, let s . By the use of the
Stirling Formula we get:
J
s
(z)
1

2s
_
lz
2s
_
s
(32)
6
What happens if s n? Notice that (s + 1)(s + 1) = (s + 2). Hence (31)
can be re-written as follows:
J
s
(z)
_
z
2
_
s
s + 1
(s + 2)
(33)
Let s 1, then (33) tends to zero and the rst term in (29) vanishes. All the other
terms are nite thus:
J
1
(z) =
_
z
2
_
1

k=1
(1)
k
k!(k)
_
z
2
_
2k
(34)
By replacing k k + 1, we get
J
1
(z) =
_
z
2
_

k=0
(1)
k
(k + 1)!(k + 1)
_
z
2
_
2k
(35)
Now (k + 1)! = (k + 1)k! and (k + 1)(k + 1) = (k + 2), hence:
J
1
(z) =
z
2

k=0
(1)
k
k!(k + 2)
_
z
2
_
2k
(36)
On the right hand side of this equation we have J
1
(z), hence
J
1
(z) = J
1
(z) (37)
In a similar way, we obtain:
J
n
(z) = (1)
n
J
n
(z) (38)
For integral n, J
n
and J
n
are linearly dependent, and we must construct a sec-
ond solution of the Bessel equation by another way. Let us calculate the following
derivative:
d
dz
z
s
J
s
(z) =
1
2
s
d
dz

k=0
(1)
k
k!(s + k + 1)
_
z
2
_
2k
=
1
2
s

k=1
(1)
k
k!(s + k + 1)
2k
_
z
4
__
z
2
_
2(k1)
=
z
2
s+1

k=0
(1)
k
k!(s + k + 2)
_
z
2
_
2k
7
By multiplying by z
s
we get:
z
s
d
dz
z
s
J
s
(z) = J
s+1
(z) (39)
or
J
s+1
(z) =
s
z
J
s
(z) J

s
(z) (40)
Let us consider now the following function:
z
s
J
s
= 2
s

k=0
(1)
k
k!(s + k + 1)
_
z
2
_
2(s+k)
(41)
After dierentiating by z and multpling by z
s
, we obtain:
z
s
d
dz
(z
s
J
s
) =
_
z
2
_
s1

k=0
(1)
k
(s + k)
k!(s + k + 1)
_
z
2
_
2k
=
_
z
2
_
s1

k=0
(1)
k
k!(s + k)
_
z
2
_
2k
= J
s1
Finally,
z
s
d
dz
(z
s
J
s
(z)) = J
s1
(z) (42)
J
s1
(z) =
s
z
J
s
+ J

s
(z) (43)
By combining (40) and (43) we get
J
s+1
(z) + J
s1
(z) =
2s
z
J
s
(z) (44)
4 Bessel Functions of Half-integral Index
Let us introduce the function g dened as follows:
8
J
s
=
1
z
1
2
g =

1
z
g (45)
After plugging into the Bessel equation (17), one realizes that g satises the equation:
g

+
_
1
s
2

1
4
z
2
_
g = 0 (46)
Let s =
1
z
. Then,
g

+ g = 0 (47)
as far as J
s
is regular at z
g = c sin(z) J
s
cz
1
2
To nd c, we remember the asymptotics of the Bessel functions at z 0.
J1
2
(z)
_
z
2
_1
2 1
(
3
2
)

_
3
2
_
=
_
1
2
+ 1
_
=
1
2

_
1
2
_
=

2
J1
2
(z)

2z

Finally,
J1
2
(z) =

2
z
sin(z) (48)
According to (40), all J
n+
1
2
(z) are expressed through a combination of power and
trigonometric functions. In particular,
J3
2
(z) = z
1
2
d
dz
(z
1
2
J1
2
(z)) =

z
1
2
d
dz
sin(z)
z
=

_
1
z
3
2
sin(z)
1
z
1
2
cos(z)
_
9
or
J3
2
(z) =

2
z
_
sin(z)
z
cos(z)
_
Separating this procedure we get:
J5
2
(z) =

2
z
__
3
z
2
1
_
sin(z)
3
z
cos(z)
_
Now, let z in equation (40). One can put approximately:
J
s+1
(z) J

s
(z)
Then, we have:
J3
2

2
z
cos(z) J5
2

sin(z)
J7
2

2
z
cos(z)
One can see that J
n+
1
2
(z) has an innite amount of zeros on the real axis. The same
statement is correct for all Bessel functions.
5 Integral Representation
Let us study the integral:
A
n
(z) =
1
2
_

e
iz sin()in
d (49)
To evaluate this integral, we use the Taylor expansion of the exponent:
e
iz sin()
=

p=0
1
p!
(iz sin())
p
=

p=0
1
p!
_
z
2
_
p
(e
i
e
i
)
p
(50)
Now, notice that the integral:
10
I
p,
=
1
2
_

(e
i
e
i
)
p
e
in
d = 0 if p < 0 (51)
Then, we denote p = n + q. The integrand in (51) can be presented in the form:
1
2
(e
i
e
i
)
n+q
e
in
= (1 e
2i
)
n
(e
i
e
i
)
q
Suppose that q is odd (q = 2k +1). All terms in the rst parentheses are even powers
of e
i
, while all terms in the second parentheses are odd powers (positive or negative)
on e
i
. As a result, the integrand is a linear combination of odd powers of e
i
. Thus
the integral is zero, and we can put q = 2k. We obtain the following intermediate
result:
A
n
(z) =
_
z
2
_
n

n=0
1
(n + 2k)!
_
z
2
_
k
I
k,n
(52)
Where
I
k,n
=
1
2
_

(e
i
e
i
)
n+2k
e
in
d (53)
To calculate I
k,n
, we use the binomial expansion in the parentheses. In this expansion,
we are interested only in the single term proportional to e
in
. All other terms after
multiplication to (53) and integration over are cancelled. Hence,
(e
i
e
i
)
n+2k

(n + 2k)!
k!(n + k)!
(e
i
)
n+k
(e
i
)
k
=
(1)
k
(n + 2k)!
k!(n + k)!
e
in
and
I
n,k
=
(1)
k
(n + 2k)!
k!(n + k)!
(54)
By plugging (54) into (52), we get nally:
A
n
(z) =
_
z
2
_
n

k=0
(1)
k
k!(n + k)!
_
z
2
_
k
= J
n
(z)
We obtained the integral representation for J
n
(z):
11
J
n
(z) =
1
2
_

e
iz sin()in
d (55)
This result is correct for positive n. Let us notice that,
J
n
(z) = (1)
n
J
n
(z) (56)
Bessel functions of even order are even functions on z, while functions of odd order
are odd. Now, we can nd A
n
(z) at negative n. Let us change simultaneously the
signs on z and n.
A
n
(z) =
1
2
_

e
iz sin()+in
d
Now by replacing , we restore the previous result. Hence,
A
n
(z) = A
n
(z) = J
n
(z)
A
n
(z) = J
n
(z) = (1)
n
J
n
(z) (57)
Finally, for all integrals on
A
n
(z) = (1)
n
J
n
(z)
notice also that J
n
is real. Then (55) can be rewritten as follows:
J
n
(z) =
1
2
_

cos(z sin n)d (58)

Now, look at e
iz sin()
. This is a periodic function which can be expanded in the Fourier
series. Apparently,
e
iz sin()
=

n=
J
n
(z)e
in
= J
0
(z) +

n=1
J
n
(z)(e
in
+ (1)
n
e
in
) (59)
The separating of imaginary and real parts in (59) gives us:
cos(z sin()) = J
0
(z) + 2

k=1
J
2k
(z) cos(2k) (60)
sin(z sin()) = 2

k=
J
2k+1
(z) sin((2k + 1))
12
By introducing t = e
i
, one can transform (59) to the following expansion:
e
z
2
(t
1
t
)
=

n=
J
n
(z)t
n
(61)
This means that F(z, t) = e
z
2
(t
1
t
)
is a generating function for the entire community
of Bessel functions of integral orders.
6 Asymptotic behavior at z
To nd the asymptotic behavior of the Bessel functions at z , we will use the
device similar to the one used for the derivation of the Stirling formula. We present
integral (55) in the form:
J
n
=
1
2
_

e
i(z,)
d (62)
(z, ) = z sin() n (63)
If z , the integrand is the fast oscillation function everywhere except the two
points where
d
d
= 0. These points are dened by the equation:
z cos() = n
at z
cos() 0

2
The contributions of points

2
give complex conjugated results. Hence, it is
enough to study the neighbourhood of the point =

2
. Let us introduce =

2
+ .
For small ,
(z, ) z
n
2

1
2
z
2
(64)
Integral (62) can be replaced approximately by the following integral:
J
n
(z) =
1

e
i(z
n
2

4
)
_

e
iz
2

2
d, Real part
Let us make the change of varibles:
13
=

2
iz
y,
1

i
= e
i
4
Then,
J
n
(z) =

z
e
i(z
n
2

4
)
_ i
4

i
4

e
y
2
dy (65)
Figure 1: Contour of Integration.
Integration is going in the complex plane along the straight line turned by 45

with
respect to the real axis. This is demonstrated in Fig. 1.
However, the contour of integration can be turned back and returned to the real
axis (To justify this fact, we need to use some elemetns of complex analysis. But,
this is true.) In other words, the integral in (65) can be replaced by the integral
_

e
y
2
dy
=

. We end up with the following result:

J
n
(z)

2
z
cos
_
z
n
2


4
_
(66)
We derived this expression only for integral n. In fact, this is correct for all s. To
prove this, we have to use a more sophisticated integral respresenation for J
s
(z) which
is valid not only for integrals. In general,
J
s
(z)

2
z
cos
_
z
s
2


4
_
(67)
In particular,
J1
2

2
z
cos
_
z

2
_

2
z
sin(z)
This is the unique Bessel function coinciding with its own asymptotic behavior.
14
7 Zeros of Bessel function
It is clear from (67) that the Bessel function J
s
(z) has an innite amount of zeros for
the half axis 0 < z < . Let us denote these zeros as a
s
N
, where N = 1, 2, .... From
(67), one can conclude that the distance between two neighboring zeros tends to .
a
s
N+1
a
s
N
as N (68)
The rst ve Bessel functions of integral order are plotted on Figure 1. The rst ve
of each are presented in Table 1. Notice that:
a
5
5
a
5
4
= 3.2377
While:
a
0
5
a
0
4
= 3.1394
Both values are close to . The derivatives of Bessel functions have the following
asymptotic behavior:
J

s
(z)

2
z
sin
_
z
s
2


4
_
(69)
The derivatives of J

s
(z) also have an innite amount of zeros b
s
N
. Again:
b
s
N+1
b
s
N
if N (70)
15
Table 1: ROOTS of the FUNCTION J
n
(x) are given in the following table.
zero J
0
(x) J
1
(x) J
2
(x) J
3
(x) J
4
(x) J
5
(x)
1 2.4048 3.8317 5.1336 6.3802 7.5883 8.7715
2 5.5201 7.0156 8.4172 9.7610 11.0647 12.3386
3 8.6537 10.1735 11.6198 13.0152 14.3725 15.7002
4 11.7915 13.3237 14.7960 16.2235 17.6160 18.9801
5 14.9309 16.4706 17.9598 19.4094 20.8269 22.2178
Table 2: The ROOTS of its DERIVATIVES are given in the following table.
zero J

0
(x) J

1
(x) J

2
(x) J

3
(x) J

4
(x) J

5
(x)
1 3.8317 1.8412 3.0542 4.2012 5.3175 6.4156
2 7.0156 5.3314 6.7061 8.0152 9.2824 10.5199
3 10.1735 8.5363 9.9695 11.3459 12.6819 13.9872
4 13.3237 11.7060 13.1704 14.5858 15.9641 17.3128
5 16.4706 14.8636 16.3475 17.7887 19.1960 20.5755
Zeros of the rst ve J

n
(z) are represented in Table 2.
16
2 4 6 8 10
-0.4
-0.2
0
0.2
0.4
0.6
x
Bessel Function of the First Kind
J
J
J
J
J
1
2
3
4
5
Figure 1
16a
Let us rewrite the Bessel equation as follows:
d
dz
zJ

s
+ zJ
s

s
2
z
J
s
= 0 (71)
By multiplying by 2zJ

, we get
d
dz
(z
2
J
2
s
s
2
J
2
s
) + 2z
2
JJ

= 0
2z
2
JJ

= z
2
d
dz
J
2
=
d
dz
z
2
J
2
2zJ
2
Finally
2zJ
2
s
=
d
dz
_
z
2
J
2
s
+ (z
2
s
2
)J
2
s
_
(72)
Integrating (72) with respect to z from 0 to a
N
, we get:
_
a
N
0
zJ
2
s
(z)dz =
1
2
a
2
N
J
2
s
(a
N
) =
1
2
a
2
N
J
2
s1
(a
N
) (73)
The last part of equation (72) follows from equations (43) and (44). In virtue of (43),
J

s
(a
N
) = J
s1
(a
N
). In virtue of (44), J
s+1
(a
N
) = J
s1
(a
N
), thus:
J
2
s+1
(a
N
) = J
2
s1
(a
N
) = J
2
s
(a
N
) (74)
From Table 1, one can see that the rst zero a
n
0
grows with n. The following statement
is correct: The number of zeros of J
s
(z) on the interval
0 < z <
_
m +
s
z
+
1
4
_
(75)
is exactly m. Putting m = 1 into (75) we get:
a
s
1
<
_
3
4
+
s
2
_
(76)
For s = 5, we get a
5
1
< 10.35. In reality a
5
1
= 8.7715. We see that this estimate is
rather accurate.
17
8 Orthogonality and Fourier-Bessel series
Let J
s
(z) be the Bessel function of order (or index) s. Let a
s
N
be its zeros such that
J
s
(a
s
N
) = 0. Suppose that 0 < r < a is an interval on the real azis. We consider now
the set of the function R
(s)
N
(r) = J
s
(
r
R
a
s
N
). This is the set of functions against the
weight r. In other words
_
a
0
R
(s)
N
(r)R
(s)
M
(r) rdr = 0 if N = M (77)
To prove this fact, we rst mention that
R
s
N
(r) = J
s
(a
s
N
) = 0 (78)
Then it is easy to check that these functions satisfy the equations
1
r

r
r
R
s
N
r
+
_
k
N

s
2
r
2
_
R
s
N
= 0, k
N
=
a
N
a
(79)
1
r

r
r
R
s
M
r
+
_
k
M

s
2
r
2
_
R
s
M
= 0. k
M
=
a
M
a
(80)
By multiplying these equations by rR
M
and rR
N
respectively and subtracting the
results, we get
R
s
M

r
r
R
s
N
r
R
s
N

r
r
R
s
M
r
= (k
2
M
k
2
M
)rR
N
R
M
(81)
The left hand side can be rewritten as the following:

r
r[R
M
, R
N
] = (k
2
M
k
2
M
)rR
N
R
M
(82)
[R
M
, R
N
] = R
M
R
M
r
R
N
R
N
r
(83)
[R
M
, R
N
]|
r=a
= 0 (84)
Then if k
2
M
= k
2
M
, integration from zero to a leads to the condition of (77). Notice that
we could replace functions R
s
N
(r) by

R
s
n
(r) = J
s
(
r
a
b
s
N
). They satisfy the condition

n
(a) = 0. In this case, equation (8.8) is again satised. The Wronskian [R
M
, R
N
] is
zero at r = a. Hence function

R
s
N
(r) satises the orthogonality condition (77).
Suppose that f(r), 0 < r < a, is some real or complex function dened on the
interval (0, r). We can represent this function as a linear combination of R
s
N
(r).
18
Let
f(r) =

N=1
f
N
R
s
N
(r) (85)
By multiplying this to rR
s
M
(r) and integrating, we get
f
N
=
1

s2
N
_
a
0
f(r)rR
s
N
(r)dr =
1

s2
N
_
a
0
f(r)rJ
s
N
_
r
a
a
N
_
dr
Here

2
N
=
_
a
0
rR
5
N
(r)dr =
1
2
a
2
a
2
N
_
a
N
0
zJ
2
z
(z)dz =
1
2
a
2
J
2
s1
(a
N
) (86)
Now one important remark, all functions R
s
N
(r) (
r
2a
a
N
)
s
at r 0. It means that
the series (78) reasonably converges if the function f(r) behaves at r as
f(r) cr
s
(87)
If the asymptotics of (80) holds, the conditions for the convergence of the series are
very similar to corresponding conditions for the standard Fourier series. In particular,
if f(a) = 0 |f

(r)| < C, where C is some arbitrary constant, this series converges

absolutely and uniformly on 0 < r < a.
A function f(r, ) dened in the disk 0 < r < a can be expanded in this disk in
the Bessel - Fourier series. First, present f(r, ) as a Fourier series in angles.
f(r, ) =

n=
f
n
(r)e
in
(88)
f
n
(r) =
1
2
_
2
0
f(r, )e
in
d (89)
What is asymptotic of f
n
(r) if r 0? Let us return to the Cartesian coordinates
(x = r cos , y = r sin ). Let f
n
() be presented as follows:
f
n
= f
0
(r) + rf
1
() +
1
2
r
2
f
2
() + +
1
r
R
n1
f
n1
() (90)
f
1
() = f
x
cos + f
y
sin
All other f
n
() are trigonometric polynomials of order n 1. Apparently:
19
_
2
0
f
k
()e
in
d = 0 if k < n
Hence f
n
(r) P
n
r
n
as r 0, with P
n
some constant, and functions f
n
(r) are god
for expansion in series of Fourier function of order n.
Finally
f(r, ) =

n=

N=1
f
nN
e
in
J
n
_
ra
N
a
_
(91)
f
nN
=
1
2
2
nM
_
2
0
e
in
d
_
a
0
rJ
n
_
ra
N
a
_
f(r, )dr (92)
In particular, if f(x, y) = (x x
0
)(y y
0
) = r
0
(
0
)(r r
0
),
f
nN
=
1
2
2
nM
r
0
e
in
0
J
N
_
r
0
a
a
N
_
(93)
Series (84) are especially good and fast converging if f(r, ) satises the Dirichlet
condition f(a, ) = 0. If this function satises the Neumann condition (f
r
(a, ) = 0),
one can use the better following set of orthogonal functions:
1, J
n
_
r
a
b
1
_
, J
n
_
r
a
b
n
_
,
9 Application of the Fourier - Bessel series to so-
lutions of PDEs in circular domains
In this chapter, we apply Bessel function to solution of boundary problems for some
basic equations of mathematical physics. We will solve equations inside the circle of
radius a, 0 < r < a, with zero boundary conditions on the circle.
First, we will start with the Poisson equation
U =
1
r

r
r
u
r
+
1
r
2

2
= f(r, ) (94)
U|
r=a
= 0
20
We expand f(r, ) in the Bessel - Fourier series
f(r, ) =

N=1

n=
f
nN
e
in
J
n
_
ra
n
N
a
_
(95)
f
nN
=
1
2
2
nM
_
2
0
e
in
d
_
a
0
rJ
n
_
ra
n
N
a
_
f(r, )dr (96)
A solution of the Poisson equation can be found in the form of a similar series
U(r, ) =

N=1

n=
U
nN
e
in
J
n
_
ra
n
N
a
_
(97)
The coeecients U
nN
and f
nN
are connected by a simple relation
U
nN
=
a
2
(a
n
N
)
2
f
nN
(98)
Suppose that f(x, y) (x x
0
)(y y
0
) = r
0
(r r
0
)(
0
) and x
0
= r
0
cos
0
,
y
0
= r
0
sin
0
. Here we used the relation
dxdy = rdrd (99)
f
n
N =
1
2
2
nN
e
in
r
0
J
n
_
r
0
a
a
n
N
_
(100)
Now,
f(r, ) = G(r, r
0
,
0
) =
r
0
2

n=

N=1
e
in(
0
)
J
n
(
ra
n
N
a
)

2
nN
J
n
_
r
0
a
n
N
a
_
G(r, r
0
,
0
) is the Green function for the Poisson equation. In other words:
U(r, ) =
_
2
0
d
0
_
a
0
G(r, r
0
,
0
)f(r
0
,
0
)dr
0
(101)
Now we will solve the diusion equation
u
r
=
_
1
r

r
r
u
r
+
1
r
2

2
_
21
inside of the circle 0 < r < a with zero boundary condition u|
r=a
= 0 and initial value
u|
t=0
= f(r, ). Again, we must perform Fourier - Bessel expansion on (95) and (96).
The solution is given by the expression:
U(r, , t) =

N=1

n=
f
nN
e
in
J
n
_
r
a
a
n
N
_
e
(
a
N
a
)
2
t
(102)
The Green function of the the diusion equation on the circle can be written as
follows:
G(r, r
0
,
0
, ) =
r
0
2

N=1

n=
1

2
nN
e
i(
0
)
J
n
_
r
a
a
n
N
_
J
n
_
r
0
a
a
n
N
_
1

4k
e
(
a
N
a
)
2

(103)
A solution of the uniform diusion equation inside the circle
U
t
= U + G(r, , ) (104)
U|
t=0
= f(r, ), U|
r=a
= 0
is expressed through this Green function by the use of the standard formula
U(r, , t) =
_
2
0
d
0
_
a
0
f(r
0
,
0
)G(r, r
0
,
0
, t)dr
0
(105)
+
_
t
0
d
_
2
0
d
0
_
a
0
g(r
0
,
0
, )G(r, r
0
,
0
, t )dr
0
The solution of the forced wave equation

2
U
t
2
= c
2
U + g(r, , t)
U|
t=0
= A(r, ) U
t
|
t=0
= B(r, )
is presented through the Green function:
22
G(r, r
0
,
0
, ) =
r
0
2

N=1

n=
1

2
nN
w
nN
e
i(
0
)
J
n
_
r
a
a
n
N
_
J
_
r
0
a
a
N
_
sin w
nN
(106)
by the standard formula again:
U(r, , t) =
_
2
0
d
0
_
a
0
A(r
0
,
0
)G
t
(r, r
0
,
0
, t)dr
0
+ (107)
+
_
2
0
d
0
_
a
0
B(r
0
,
0
)G(r, r
0
,
0
, t)dr
0
+
_
t
0
d
_
2
0
d
0
_
a
0
g(r
0
,
0
, )G(r, r
0
,
0
, t )dr
0
Where w
nN
are eigenfrequencies (w
nN
=
c
a
a
n
N
). Using Table 1, one can order the
eigenfrequencies as follows.
w
q+1
> w
q
, q
1
= 1, ,
w
1
= w
0
a
0
1
= 2.4048w
0
, w
0
=
c
a
w
2
= w
0
a
1
1
= 3.8317w
0
w
3
= w
0
a
2
1
= 5.1336w
0
w
4
= w
0
a
0
2
= 5.5201w
0
w
5
= w
0
a
0
3
= 6.3802w
0
If the Dirichlet boundary condition is replaced by the Neumann boundary condition.
In this case instead of J
n
(
r
a
a
N
) there should be J
n
(
r
a
b
n
N
). We must now correct the
value for
2
nm
:

2
nm
=
_
a
0
J
2
n
_
r
a
b
n
N
_
rdr =
a
2
b
2
N
_
b
N
0
J
2
n
(z)zdz (108)
23
10 Bessel Functions of Second and Third Kinds
Bessel functions of the second kind, also called Neumann or Webers Functions, are
dened as solutions of the Bessel equation with the following asymptotics:
N
s
(z)

2
z
sin
_
z
s
2


4
_
(109)
If s is not an integer, one can present N
s
(z) as a linear combination of J
s
and J
s
N
s
(z) = AJ
s
+ BJ
s
(110)
To nd A, B we have to solve the following equation:
sin() = Acos() + B cos( + s)
= z
s
2


4
As far as
cos( + s) = cos() cos(s) sin() sin(s)
we get
A + B cos(s) = 0
1 = B sin(s)
Finally,
N
s
=
cos(s)J
s
J
s
sin(s)
(111)
What is going on if s n? As far as cos(n) = (1)
n
, J
n
= (1)
n
J
n
, both the
numerator and denominator in (111) tend to zero and we should use Lhopitals Rule.
As a result:
N
s
(z) =
cos(n)

s
J
s

s
J
s
|
s=n
cos(n)
=
1

_

s
J
s
(1)
n

s
J
s
|
s=n
_
(112)
24
To study the behavior of N
n
(z) at z 0 one can use the series representation of (29)
J
s
(z) =
_
z
2
_
s
n

k=0
(1)
k
k!(s + k + 1)
_
z
2
_
k
Now,
_
z
2
_
= e
sln
z
2
d
ds
_
z
2
_
s
= ln
_
z
2
_
e
sln
z
2
= ln
_
z
2
__
z
2
_
s
In the same way,
Line is Cut O in Notes
Thus, N
n
(z) can be presented as follows:
N
n
(z) = 2 ln
_
z
2
_
J
n
(z)

N
n
(z) (113)
In (113),

N
n
(z) = N
(1)
n
(z) + N
(2)
n
(z) appears as a result of dierentiating
1
(s+k+1)
by s. Finally, we put s = n, then

N
n
(z) is expanision only over integer powers of n,
both positive and negative.
When s n, the rst n terms in the expansion (29) tend to zero. However,
their derivatives by s are not zero. They can easily be calculated. The last term in J
s
vanishes as s n, which corresponds to k = n 1. This term gives the following
contribution to J
s
(z)
1
(n 1)!(s + n)
_
z
2
_
n2
=
s + n
(n 1)!(s + n + 1)
_
z
2
_
n2
Dierentiating the above expression by s at s = n gives the following contribution to
N
(1)
n
(z):
1
(n 1)!
_
z
2
_
n2
25
Collecting all similar terms (k n1) together, we end up with the following explicit
expression for N
(1)
n
(z)
N
(1)
n
(z) =
n1

k=0
(n k 1)!
k!
_
z
2
_
2kn
(114)
N
(1)
n
(z) has singularities as z . The most singular term is:
N
(1)
n
(z) (n 1)!
_
2
z
_
n
(115)
N
(2)
n
(z) =

k=0
a
k
(1)
k
_
z
2
_
2k+n
(116)
is the fast converging power series in (116)
a
k
=
1
k!(n + k)!
_
(k + 1) + (k + n + 1)
_
(117)
Here, (s) is a new special function
(s) =
d
ds
ln((s)) =

(s)
(s)

(s) =
_

0
e
t
ln(t)t
s1
dt (118)
Through the use of integration by parts, one can nd the explicit value of (s) in all
integral points can be [Line Cut O Notes]. Remembering that e
t
=
d
dt
e
t
, we get
from (118)

(s) = (s 1)

(s 1) + (s 1)
By dividing by (s) and replacing s s = 1, we end up with the dierence equation
for
(s + 1) = (s) +
1
s
(1) =
_

0
e
t
ln(t)dt = c (c 0.5772) (119)
26
c is the so-called Euler constant. Finally, we get for any integral point.
(k + 1) = c + 1 + +
1
k
(120)
To nish with the Neumann functions, we present the asymptotic behavior of the rst
two at z 0
N
0
(z) 2
_
ln
z
2
+ c
_
+ 0(z
2
)
N
1
(z)
2
z
+ 0(z) ln(z) (121)
Functions N
n
are even if n is even and odd if N
n
is odd.
N
n
(z) = (1)
n
N
n
(z) (122)
The Bessel functions of third kind are also know as the Hankel functions, which are
dened as follow:
H
(1)
s
(z) = J
s
(z) + iN
s
(z)
H
(2)
s
(z) = J
s
(z) iN
s
(z) (123)
At z they have the following amymptotics
H
(1)
s
(z)

2
z
e
i(z
s
2

4
)
H
(2)
s
(z)

2
z
e
i(z
s
2

4
)
(124)
apparently, H
(2)
s
(z) =

H
(1)
s
(z). The Neumann functions are plotted in Figure 2.
11 Modied Bessel Functions
Modied Bessel functions are solutions of the modied Bessel equation
1
z
d
dz
z
dR
dz

_
1
s
2
z
2
_
R = 0 (125)
27
This equation appears if we perform the transformation z iz. In other words, the
modied Bessel functions are Bessel functions of imaginary argument. However, one
muse be careful performing the transform z iz because we have to observe not
only asymptotics at z 0, but also asmyptotics at z . One solution Y
s
(z) of
equation (125) is dened by the series
I
s
(z) =
_
z
2
_
s

k=0
1
k!(k + s + 1)
_
z
2
_
2k
(126)
At z 0
I
s
(z)
z
s
z
s
(s + 1)
(127)
At z
I
s
(z)

2
z
e
z
(128)
I
s
is the real function of real argument. They are connected with Bessel functions of
the rst kind by the relation:
I
s
(z) = e

2
si
J
s
(iz) (129)
In particular,
I
n
(z) = i
n
J
n
(iz) (130)
Modied Bessel functions of second kind are dened by the relation
k
s
(z) =
i
2
e

2
si
H
(1)
s
(iz) (131)
They have asymptotics at
k
s
(z)
_

2z
e
z
, z (132)
Both I
s
(z) and k
s
(z).
Modied Bessel functions of the First and Second kind are plotted on Figures 3
and 4.
12 Applications of the Modied Bessel Function
The modied Bessel functions are commonly used for solutions to many dierent
applied problems. Let us consider the cylinder of radius R and length 2a
0 < r < R a < z < a
28
2 4 6 8 10
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
Bessel Function of the Second Kind
Figure 2
28a
A Bessel function of the second kind Nn(x) is a solution to the BESSEL
DIFFERENTIAL EQUATION which is singular at the origin. Bessel func-
tions of the second kind are also called NEUMANN FUNCTIONS or
WEBER FUNCTIONS. The above plot shows Nn(x) for n = 1 , 2 , ... , 5
1 2 3 4 5
0
0.5
1.0
1.5
2.0
2.5
3.0
I
I
I
I
I
1
2
3
4
5
x
Modied Bessel Function of the First Kind
Figure 3
28b
A function In(x) which is one of the solutions to the MODIFIED BESSEL
DIFFERENTIAL EQUATION and is closely related to the BESSEL FUNC-
TION OF THE FIRST KIND Jn(x). The above plot In(x) for n = 1, 2, ... , 5.
1 2 3 4 5
0
2
4
6
8
10
Modied Bessel Function of the Second Kind
Figure 4
28c
The function Kn(x) which is one of the solutions to the MODIFIED BESSEL
DIFFERENTIAL EQUATION. The above plot shows Kn(x) for n = 1, 2, ... , 5.
Let us solve the Laplace equation:
u =
1
r

r
r
u
r
+
1
r
2

2
u

2
+

2
u
z
2
= 0 (133)
with the lateral boundary conditions:
u|
z=a
= 0, u|
z=a
= 0
u|
r=R
= f(z, ) a < z < a
f(z, + 2) = f(z, )
To solve equation (133), we will use separation of variables and look for solutions in
the form:
u = e
in
cos

m
2a
z

R
n,m
(r) (134)
R
n,m
satises the following equation:
1
r

r
r
R
r

k
2
m
+
n
2
z
2

R = 0, k
2
m
=
m
2a
(135)
Equation (135) has the following solution satisfying the condition:
R|
r=R
= 1, R =
I
n
(k
m
z)
I
n
(k
m
R)
(136)
The solution of the Laplace equation is:
u =

n=

m=1
f
n,m
e
in
cos

m
2a
z

I
n
(k
m
z)
I
n
(k
m
R)
(137)
f
n,m
- coecient of double Fourier series to be found from the condition:
f(z, ) =

n=

m=1
f
n,m
e
in
cos

m
2a
z

(138)
They are:
f
n,m
=
1
2a

2
0
d

a
a
f(, z)e
in
cos

m
2a
z

dz (139)
29
Let us now solve the Laplace equation (133) outside the innite cylinder (0 < r < R
and < z < ). Again, the boundary condition is:
u|
r=R
= f(z, ) < z <
f(z, + 2) = f(z, )
To solve the problem, we must specify the boundary condition in innity:
u
r
0 at r
Now, we should perform the Fourier transformations in the z-direction:
u(k, , r) =

u(z, , r)e
ikz
dz
f(k, ) =

f(z, )e
ikz
dz (140)
and realize the expansion in the Fourier series in angles, thus:
f(k, ) =

f
n
(k)e
in
f
n
(k) =
1
2

2
0
f(k, )e
in
d (141)
The radial part of the solution satises the modied Bessel equation:
1
r

r
r
R
r

k
2
+
n
r

R = 0 (142)
It should be taken as follows:
R(k, n) =
K
n
(kr)
K
n
(kR)
The solution is given by the inverse Fourier Transform:
u(z, r, ) =
1
2

dt

n=
f
n
(k)
K
n
(kr)
K
n
(kR)
e
ikz+in
(143)
30
13 Heavy Chain
The future Bessel function appeared in mathematics in 1732 when Danie Bernoully
solved the problem on oscillations of the hung, heavy chain. Let the chain of length
l and linear density be hung such that it can move only in one direction. Let the
coordinate x be taken such that x = 0 at the free end of the chain. Then the deviation
from equilibrium state u = u(x, t) satises the equation:

2
u
t
2
= g

2
u
x
2
+
u
x

(144)
u|
x=l
= 0
Use separation of variables:
u = X(x)T(t)
where T(t) = sin(t + ) leads to the equation:
xX

+ X

(x) +

2
g
X(x) = 0 (145)
with the boundary condition:
X(l) = 0, X(0) <
By introducing the new variable:
y = 2

x
g
We transform (145) to the Bessel equation:
d
2
X
dy
2
+
1
y
dX
dy
+ X = 0 (146)
This is the equation for Bessel functions of zero order. Thus, the solution is:
31
u = AJ
0

x
g

(147)
The characteristic frequency
k
can take consequence of discrete values (
k
, k =
1, 2, , ). They can be found from the boundary condition:
u(l) = 0 J
0

2
k

l
y

= 0
Hence,
2
k

l
g
= a
0
k
J
0
(a
0
k
) = 0

k
=
1
2

g
l
a
0
k
a
0
n
- zeros of the Bessel function J
0
.
32