Supplement 2
Some Special Functions
In Supplement 2, n is a positive integer, unless otherwise specied.
2.1. Gamma-function
The gamma-function (z) is an analitic function of z everywhere, except for the points
z = 0, 1, 2, . . .
For Re z > 0,
(z) =
tz1 et dt.
0
For (n + 1) < Re z < n, where n is an integer,
(z) =
e
0
n
(1)m z1
dt.
t
m!
m=0
The gamma-function possesses the following properties:
(z + 1) = z(z),
(z)(z) =
(2z) =
,
z sin(z)
(n + 1) = n!,
(z)(1 z) =
22z1
1
(z) z +
,
2
,
sin(z)
(3z) =
n1
k=0
(z + n)
= (z)n ,
(z)
c 1995 by CRC Press, Inc.
1
1
+z
z =
,
2
2
cos(z)
33z1/2
1
2
(z) z +
z+
,
2
3
3
(nz) = (2)(1n)/2 nnz1/2
1
= ,
2
1
= 2 ,
2
(1) = (2) = 1.
k
,
z+
n
1
= n (2n 1)!!,
n+
2
2
1
2n
n
= (1)n
,
2
(2n 1)!!
(z + 1)
= Czw .
(w + 1)(z w + 1)
2.2. Bessel functions J and Y
2.2.1. Basic Formulae
The Bessel functions of the rst and second kind J and Y (function Y is also called
the Neumann function) are solutions of the Bessel equation 2.1.2.121 and are dened by
the formulae
J (x) =
(1)k (x/2)+2k
,
k! ( + k + 1)
k=0
Y (x) =
J (x) cos J (x)
.
sin
The formulae for Y (x) is valid for = 0, 1, 2, . . . ; (see below for the integral representation of Y (x), as well as equation 2.1.2.121 for = 0, 1, 2, . . . ).
Function Z (x) = C1 J (x) + C2 Y (x) is reered to as the cylindric function.
The Bessel functions possess the following properties:
2Z (x) = x[Z1 (x) + Z+1 (x)],
d
1
1
Z = [Z1 (x) Z+1 (x) = [Z (x) Z1 ],
2
x
dx
d
[x Z (x)] = x Z1 (x),
dx
d n
[x J (x)] = xn Jn (x),
x dx
Jn (x) = (1)n Jn (x),
d
[x Z (x)] = x Z+1 (x),
dx
d n
[x J (x)] = (1)n xn J+n (x),
x dx
Yn (x) = (1)n Yn (x)
2.2.2. Bessel functions for = n
2
sin x,
J1/2 (x) =
x
2
1
J3/2 (x) =
sin x cos x ,
x x
Jn+1/2 (x) =
1
2
n = 0, 1, 2, . . .
; n = 0, 1, 2, . . .
2
cos x,
x
2
1
J3/2 (x) =
cos x sin x ,
x
x
J1/2 (x) =
[n/2]
(1)k (n + 2k)!
2
n
sin x
x
2
(2k)! (n 2k)! (2x)2k
k=0
[(n1)/2]
(1)k (n + 2k + 1)!
n
,
+ cos x
2
(2k + 1)! (n 2k 1)! (2x)2k+1
k=0
[n/2]
(1)k (n + 2k)!
2
n
Jn1/2 (x) =
cos x +
x
2
(2k)! (n 2k)! (2x)2k
k=0
[(n1)/2]
(1)k (n + 2k + 1)!
n
,
sin x +
2
(2k + 1)! (n 2k 1)! (2x)2k+1
k=0
2
cos x,
x
Yn+1/2 (x) = (1)n+1 Jn1/2 (x),
Y1/2 (x) =
c 1995 by CRC Press, Inc.
2
sin x,
x
Yn1/2 (x) = (1)n Jn+1/2 (x),
Y1/2 (x) =
2.2.3. Wronskians an Similar Formulae
Notation: W (f, g) = f gx fx g.
W (J , J ) =
J (x)J+1 (x) + J (x)J1 (x) =
2
sin(),
x
2 sin()
,
x
W (J , Y ) =
2
,
x
J (x)Y+1 (x) J+1 (x)Y (x) =
2
.
x
2.2.4. Integral Representation
Functions J and Y may be expressed in terms of denite integrals (with x > 0):
J (x) =
cos(x sin ) d sin
exp(x sinh t t) dt,
0
Y (x) =
sin(x sin ) d
(et + et cos )ex sinh t dt.
0
For || <
1
2
, x > 0,
21 x
J (x) =
1/2 ( 12 )
sin(xt) dt
,
2
(t 1)+1/2
21 x
Y (x) =
1/2 ( 12 )
cos(xt) dt
.
1)+1/2
(t2
For = 0, x > 0,
2
J0 (x) =
sin(x cosh t) dt,
0
2
Y0 (x) =
cos(x cosh t) dt.
0
2.2.5. Integrals with Bessel Functions on Closed Intervals
x J (x) dx =
0
2 (
++1 ++3
x++1
x2
,
, + 1;
,
F
2
2
4
+ + 1)( + 1)
where Re( + ) > 1, F (a, b, c; x) is the hypergeometric series (see equation 2.1.2.158),
x
cos()() ++1 + + 1
++3
x2
x Y (x) dx =
F
x
, + 1,
,
2
2
4
2 ( + + 1)
0
2 () +1
+1
+3
x2
F
x
, 1 ,
,
,
+1
2
2
4
where Re > | Re | 1.
2.2.6. Asymptotic Expansion, as |x|
J (x) =
2
x
cos
1
4x 2 M
m=0
2m
(1) (, 2m)(2x)
2M
+ O(|x|
1
4x 2 M
sin
(1)m (, 2m + 1)(2x)2m1 + O(|x|2M 1 ) ,
4
m=0
c 1995 by CRC Press, Inc.
Y (x) =
1
4x 2 M
(1)m (, 2m)(2x)2m + O(|x|2M )
sin
4
m=0
1
4x 2 M
m
2m1
2M 1
+ cos
(1) (, 2m + 1)(2x)
+ O(|x|
) ,
4
m=0
2
x
where (, m) =
1
22m m!
(4 2 1)(4 2 32 ) . . . [4 2 (2m 1)2 ] =
( 12 + + m)
m! ( 12 + m)
2.3. Modied Bessel Functions I and K
2.3.1. Basic Formulae
The modied Bessel functions of the rst and second kind I and K (function K is
also called the Basset function) are solutions the modied Bessel equation 2.1.2.122 and are
dened by the formulae
I (x) =
k=0
(x/2)2k+
,
k! ( + k + 1)
K (x) =
I I
,
2 sin
(see equation 2.1.2.122 with = 0, 1, 2, . . . for the representation of K (x)).
The modied Bessel functions possess the following properties:
K (x) = K (x);
In (x) = (1)n In (x),
2I (x) = x[I1 (x) I+1 (x)],
d
1
I (x) = [I1 (x) + I+1 (x)],
2
dx
2K (x) = x[K1 (x) K+1 (x)],
d
1
K (x) = [K1 (x) + K+1 (x)],
2
dx
2.3.2. Modied Bessel Functions for = n
n = 0, 1, 2, . . .
1
2
; n = 0, 1, 2, . . .
2
2
sinh x,
cosh x,
I1/2 (x) =
I1/2 (x) =
x
x
2 1
2 1
I3/2 (x) =
sinh x + cosh x , I3/2 (x) =
cosh x + sinh x ,
x
x
x x
n
n
k
(1) (n + k)!
(n + k)!
1
n x
,
(1)
e
In+1/2 (x) =
ex
k! (n k)! (2x)k
k! (n k)! (2x)k
2x
k=0
k=0
n
n
(1)k (n + k)!
(n + k)!
1
n x
In1/2 (x) =
,
+
(1)
e
ex
k! (n k)! (2x)k
k! (n k)! (2x)k
2x
k=0
k=0
x
1 x
K1/2 (x) =
K3/2 (x) =
e ,
1+
e ,
2x
2x
x
n
(n + k)!
x
Kn+1/2 (x) = Kn1/2 (x) =
.
e
2x
k! (n k)! (2x)k
k=0
c 1995 by CRC Press, Inc.
2.3.3. Wronskians and Similar Formulae
Notation: W (f, g) = f gx fx g.
W (I , I ) =
2
sin(),
x
I (x)I+1 (x) I (x)I1 (x) =
2 sin()
,
x
W (I , K ) =
1
,
x
I (x)K+1 (x) + I+1 (x)K (x) =
1
.
x
2.3.4. Integral Representation
Functions I and K may be expressed in terms of denite integrals (with x > 0, > 12 ):
1
x
I (x) =
exp(xt)(1 t2 )1/2 dt,
1/2 2 ( + 12 ) 1
K (x) =
exp(x cosh t) cosh(t) dt.
0
2.3.5. Integrals with Modied Bessel Functions on Closed Intervals
x I (x) dx =
0
++1 ++3
x++1
x2
F
,
,
+
1;
,
2
2
4
2 ( + + 1)( + 1)
where Re( + ) > 1, F (a, b, c; x) is the hypergeometric series (see equation 2.1.2.158),
x K (x) dx =
0
21 () +1 + 1
+ 3 x2
F
x
, 1 ,
,
+1
2
2
4
21 () ++1 + + 1
+ + 3 x2
F
+
x
, 1 + ,
,
,
++1
2
2
4
where Re > | Re | 1.
2.3.6. Asymptotic Expansion, as x
M
(4 2 1)(4 2 32 ) . . . [4 2 (2m 1)2 ]
ex
M 1
I (x) =
(1)m
+
O(x
)
,
1+
m! (8x)m
2x
m=1
M
(4 2 1)(4 2 32 ) . . . [4 2 (2m 1)2 ]
x
K (x) =
1+
+ O(xM 1 ) .
e
m
2x
m! (8x)
m=1
2.4. Degenerate Hypergeometric Functions
2.4.1. Denitions
The degenerate hypergeometric functions (a, b; x) and (a, b; x) are solutions of the
degenerate hypergeometric equation 2.1.2.65.
c 1995 by CRC Press, Inc.
If b = 0, 1, 2, 3, . . . , function (a, b; x) is expressed in terms of Kummers series:
(a, b; x) = 1 +
(a)k xk
,
(b)k k!
k=1
where (a)k = a(a + 1) . . . (a + k 1), (a)0 = 1. Function (a, b; x) is dened as follows:
(a, b; x) =
(1 b)
(b 1) 1b
(a, b; x) +
x (a b + 1, 2 b; x).
(a b + 1)
(a)
Table 2.1 represents some special cases where is expressed in terms of simpler functions.
2.4.2. Basic Properties
Kummers transformation:
(a, b; x) = ex (b a, b; x),
(a, b; x) = x1b (1 + a b, 2 b; x).
Linear relations for function :
(b a)(a 1, b; x) + (2a b + x)(a, b; x) a(a + 1, b; x) = 0,
b(b 1)(a, b 1; x) b(b 1 + x)(a, b; x) + (b a)x(a, b + 1; x) = 0,
(a b + 1)(a, b; x) a(a + 1, b; x) + (b 1)(a, b 1; x) = 0,
b(a, b; x) b(a 1, b; x) x(a, b + 1; x) = 0,
b(a + x)(a, b; x) (b a)x(a, b + 1; x) ab(a + 1, b; x) = 0,
(a 1 + x)(a, b; x) + (b a)(a 1, b; x) (b 1)(a, b 1; x) = 0,
Linear relations for function :
(a 1, b; x) (2a b + x)(a, b; x) + a(a b + 1)(a + 1, b; x) = 0,
(b a 1)(a, b 1; x) (b 1 + x)(a, b; x) + x(a, b + 1; x) = 0,
(a, b; x) a(a + 1, b; x) (a, b 1; x) = 0,
(b a)(a, b; x) x(a, b + 1; x) + (a 1, b; x) = 0,
(a + x)(a, b; x) + a(b a 1)(a + 1, b; x) x(a, b + 1; x) = 0,
(a 1 + x)(a, b; x) (a 1, b; x) + (a c + 1)(a, b 1; x) = 0.
Dierentiation formulae:
d
a
(a, b; x) = (a + 1, b + 1; x),
dx
b
(a)n
dn
(a, b; x) =
(a + n, b + n; x),
n
dx
(b)n
d
(a, b; x) = a(a + 1, b + 1; x),
dx
dn
(a, b; x) = (1)n (a)n (a + n, b + n; x),
dxn
Wronskian:
W (, ) = x x =
c 1995 by CRC Press, Inc.
(b) b x
x e .
(a)
2.4.3. Integral Representation
1
(b)
ext ta1 (1 t)ba1 dt,
(if b > a > 0)
(a) (b a) 0
1
(a, b; x) =
ext ta1 (1 + t)ba1 dt,
(if a > 0, x > 0)
(a) 0
(a, b; x) =
where (a) is the gamma-function.
2.4.4. Integrals with Degenerate Hypergeometric Functions
(a, b; x) dx =
b1
(a1, b1; x)+C,
a1
xn (a, b; x) dx = n!
n+1
k=1
xn (a, b; x) dx = n!
n+1
k=1
(a, b; x) dx =
1
(a1, b1; x)+C,
1a
(1)k+1 (1 b)k xnk+1
(a k, b k; x) + C,
(1 a)k (n k + 1)!
(1)k+1 xnk+1
(a k, b k; x) + C.
(1 a)k (n k + 1)!
2.4.5. Asymptotic Expansion, as |x|
N
(b) x ab (c a)n (1 a)n n
N 1
(a, b; x) =
) ,
x > 0,
x + O(x
e x
n!
(a)
n=0
N
(b)
(a)n (a b + 1)n
a
n
N 1
(a, b; x) =
) ,
x < 0,
(x)
(x) + O(|x|
n!
(b a)
n=0
N
(a)n (a b + 1)n n
(a, b; x) = xa
(1)n
< x < +.
x + O(|x|N 1 ) ,
n!
n=0
2.5. Legendre Functions
2.5.1. Denitions
The associated Legendre functions P (z) and Q (z) of the rst and second kind are
linearly-independent solutions the Legendre equation (see 2.1.2.212):
(1 z 2 )yzz
2zyz + [( + 1) 2 (1 z 2 )1 ]y = 0,
where parameters , and variable z may be arbitrary real or complex numbers.
With |1 z| < 2, the following formulae may be used:
z + 1 /2
1
1z
F , 1 + , 1 ,
,
2
(1 ) z 1
z 1 2
z + 1 2
1z
1z
F , 1+, 1+,
F , 1+, 1,
Q (z) = A
+B
,
z+1
2
z1
2
()(1 + + )
()
A = ei
, i2 = 1,
, B = ei
2
2(1 + )
P (z) =
c 1995 by CRC Press, Inc.
where F (a, b, c; z) is the hypergeometric series (see equation 2.1.2.158).
With |z| > 1, the following formulae may be used:
P (z)
1 + 2 + 2 + 3 1
21 ( 12 ) +1 2
/2
,
,
, 2
=
(z 1)
F
z
2
2
2
z
( )
1
2 ( 2 + )
+ 1 1 2 1
+ 2
/2
+
,
,
, 2 ,
(z 1)
F
z
2
2
2
z
(1 + )
Q (z) = ei
( + + 1)
2+1 ( +
3
2
(z 1)
2
/2
2++
2
1 + + 2 + 3 1
,
,
, 2
2
2
z
,
The functions
P (z) P0 (z),
Q (z) Q0 (z)
are called the Legendre functions and are solutions of the Legendre equation 2.1.2.148.
The modied associated Legendre functions on the cut z = x, 1 < x < 1, are dened
by the formulae
1
1 1 i
e2
P (x + i0) + e 2 i P (x i0) ,
2
1
1 i 1 i
Q (x) = e
e 2
Q (x + i0) + e 2 i Q (x i0) .
2
P (x) =
2.5.2. Trigonometric Expansions
With 1 < x < 1, the modied associated Legendre functions can be expressed in terms
of trigonometric series:
( 12 + )k (1 + + )k
2+1 ( + + 1)
P (cos ) =
sin[(2k + + + 1)],
(sin
)
( + 32 )
k!( + 32 )k
k=0
Q (cos ) =
( + + 1)
( +
3
2
(sin )
( 12 + )k (1 + + )k
k=0
k!( +
3
2 )k
cos[(2k + + + 1)],
where 0 < < .
2.5.3. Some Relations
( + n + 1) n
n = 0, 1, 2, . . .
P (z)
( n + 1)
(1 + + )
Q (z) =
ei P (z)
P (z) .
2 sin()
(1 + )
P (z) = P1
(z),
Pn (z) =
For 1 < x < 1,
P+1
(x) =
2 + 1
+
xP (x)
P (x),
+1
+ 1 1
P+1
(x) = P1
(x) (2 + 1)(1 x2 )1/2 P1 (x),
d
x
+
(x).
P (x) = 2
P (x) 2
P
dx
x 1
x 1 1
c 1995 by CRC Press, Inc.
For 0 < x < 1,
P (x) = P (x) cos[( + )] 2 1 Q (x) sin[( + )],
Q (x) = Q (x) cos[( + )]
1
2
P (x) sin[( + )].
Wronskians:
1
,
1 x2
++2
++1
2
2
2
where k = 2
+1 +2 .
2
2
For n = 0, 1, 2, . . . ,
W (P , Q ) =
Pn (x) = (1)n (1 x2 )n/2
W (P , Q ) =
dn
P (x),
dxn
k
,
1 x2
Qn (x) = (1)n (1 x2 )n/2
dn
Q (x).
dxn
The Legendre polynomials Pn (x) and the Legendre function Qn (x) are given by the
formulae
dn
1
Pn (x) =
(x2 1)n ,
n! 2n dxn
n
1
1
1+x
Qn (x) = Pn (x) ln
Pm1 (x)Pnm (x).
2
1x
m
m=1
Functions Pn = Pn (x) can be conveniently calculated by the recurrence relations
P0 = 1,
P1 = x,
P2 =
1
(3x2 1),
2
...,
Pn+1 =
2n + 1
n
xPn
Pn1 .
n+1
n+1
Three leading functions Qn = Qn (x) are
Q0 =
1
1+x
ln
,
2
1x
Q1 =
x
1+x
ln
1,
2
1x
Q2 =
3x2 1
1+x
3
ln
x.
4
1x
2
2.5.4. Integral Representation
For n = 0, 1, 2, . . .
( + n + 1)
Pn (z) =
z + cos t z 2 1 cos(nt) dt, Re z > 0,
( + 1)
0
n
n ( + n + 1)
Q (z) = (1) +1
(z + cos t)n1 (sin t)2+1 dt,
(z 2 1)n/2
2
( + 1)
0
Re > 1,
Note that in the latter formulae z = x, 1 < x < 1.
2.6. The Weierstrass function
2.6.1. Denitions
The Weierstrass function = (z, g2 , g3 ) is dened implicitly by the elliptic integral
dt
z=
3
4t g2 t g3
and satises the rst order dierential equation
(z )2 = 43 g2 g3 .
c 1995 by CRC Press, Inc.
2.6.2. Some Properties
Below (z) stands for (z, g2 , g3 ).
(z) = (z),
(z1 + z2 ) = (z1 ) (z2 ) +
1
4
(z1 ) (z2 )
(z1 ) (z2 )
2
.
In the vicinity of the point z = 0, the Weierstrass function can be expanded into the
series
g22 6
g2 2
g3 4
3g2 g3 8
1
(z) = 2 +
z +
z +
z +
z + .
20
28
1200
6160
z
c 1995 by CRC Press, Inc.