A Note on the Gamma Function
For > 0,
() =
t1 et dt.
(1)
Except in special cases this integral cannot be evaluated exactly and one
has to use approximations. However, we can integrate by parts to get the
following important result
() =
t1 et dt
t1 et
0
= 0 0 + ( 1)
t2 et dt = ( 1) ( 1) .
This relation
() = ( 1) ( 1) .
(2)
is called a recursion relation.
If = n is a positive integer, then using this recursion repeatedly we get
(n) = (n 1)(n 1) = (n 1)(n 2)(n 2) (1).
Since
(1) =
et dt = 1,
the above equation becomes
(n) = (n 1)(n 2) 1 = (n 1)!.
Let now rewrite Eq.(2) as follows:
( 1) =
1
().
1
(3)
What is neat about this is that, since the definition given by Eq.(1) is valid
for positive values of , we can use Eq.(3) to define the Gamma function for
negative, non-integral values1 . For example, from Eq.(3),
(1/2) = 2(1/2),
(3/2) = (2/3)(1/2) = (2/3)(2)(1/2)
and so on. This procedure can be used to expand the definition to complex
numbers as well!
It also so happens that one can evaluate (1/2) exactly, using the substitution, t = x2 /2 so that dt = xdx and reducing the integral to the normal
distribution integral.
(1/2) =
=
Z0
0
t1/2 et dt
1/2
x2 /2
ex
2 /2
ex
2 /2
xdx
dx
1 Z x2 /2
= 2
e
dx = 2 [1 (0)] = 2 (1/2) = .
2 0
1
Since the integral in Eq.(1) is undefined for = 0, the gamma function is undefined
for non-positive integer values.