Scribd
Upload
81 views4 pages

Notes 1 GammaFunction

The document discusses properties of the gamma function Γ(t). Some key points: - Γ(t) is defined by an integral involving exponential and power functions. Various changes of variables preserve the form of the integral. - Applying integration by parts to the integral definition establishes the recursive property Γ(t) = (t - 1)Γ(t - 1). - For integer values of t, this recursion gives Γ(t) = (t - 1)!, and Γ(t) can be generalized to non-integer values. - The gamma function arises in probability via the gamma distribution with density proportional to x^(α-1)e^-x

Uploaded by

Atonu Tanvir Hossain
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
81 views4 pages

Notes 1 GammaFunction

The document discusses properties of the gamma function Γ(t). Some key points: - Γ(t) is defined by an integral involving exponential and power functions. Various changes of variables preserve the form of the integral. - Applying integration by parts to the integral definition establishes the recursive property Γ(t) = (t - 1)Γ(t - 1). - For integer values of t, this recursion gives Γ(t) = (t - 1)!, and Γ(t) can be generalized to non-integer values. - The gamma function arises in probability via the gamma distribution with density proportional to x^(α-1)e^-x

Uploaded by

Atonu Tanvir Hossain
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

FACTS ABOUT THE GAMMA INTEGRAL

Page gs 2011 1
A great number of tricks revolve around the integral

(t) = x e dx
t x

z
1
0


(Note as usual that x is only a dummy symbol.) This integral is called the gamma
function. Lets consider only t>0. If you make the change of variable x=
u

, then
dx =
du

, and the integral is



(t) =
1
0
u
t
t
u
e du



which is merely another form. Yet another change, working from the original, is x =v ;
then dx = dv, and the integral is

(t) =
1
0
t t v
v e dv

=
1
0
t t x
x e dx



Lets work with the form at the top of the page. You should be able to show that
(1) =1. Now well do an integration by parts, using the mnemonic

udv uv vdu =
z z


By the way, when this mnemonic is applied to a definite integral, we write
it as udv uv vdu
a
b b
a a
b
z z
= a f . The limits refer to the variable of
integration. If this variable is x, we would write this as
u x dv x u x v x v x du x
a
b x b
x a a
b
( ) ( ) ( ) ( ) ( ) ( )
z z
=
=
=
a f .

Let
u(x) =x
t-1
and dv(x) = e
- x
dx

This creates x e dx
t x

z
1
0
= u x dv x ( ) ( )
0

z
.
FACTS ABOUT THE GAMMA INTEGRAL

Page gs 2011 2

Then uv vdu
z
becomes

x e e t x dx
t x
x
x
x t
=
=


z
1
0
2
0
1
c h d i c h
( ) = +

=
=

z
x e t x e dx
t x
x
x
t x 1
0
2
0
1
c h
( )

The value of x
t-1
e
- x
at x =0 is clearly 0. The value at x = requires a
limit process, possibly as lHospitals rule. This would have the following
form:

LHospitals rule says that
( )
( )
( )
( )
lim lim
x x
f x f x
g x g x

. (In using
this rule, x can approach any value, not just . Then
lim lim lim
x
t x
x
t
x
x
d
dx
t
d
dx
x
x e
x
e
x
e

= =
1
1 1
= lim
( )
x
t
x
t x
e

1
2
= ( )lim t
x
e
x
t
x

1
2
. This final form still looks like an

ratio; however, the logic


can be applied again repeatedly until the exponent of x is below 1.
At that point, the limiting ratio is
0

, which is certainly 0.

= 0 + ( ) t x e dx
t x

z
1
2
0
= (t - 1) (t - 1)


Weve established that (1) =1 and that (t) =(t - 1) (t - 1). Therefore,

(2) =1 (1) =1
(3) =2 (2) =2
(4) =3 (3) =6
(5) =4 (4) =24


If r is an integer, (r) =(r - 1)!

It also happens that, for example, (4.2) =3.2 2.2 1.2 0.2 (0.2). A curious
result, to be shown soon, is that
1
2
a f = . A use of this is
Then
du(x) =(t - 1) x
t-2
dx v(x) =- e
- x


v(x) is simply the integral of dv(x).
FACTS ABOUT THE GAMMA INTEGRAL

Page gs 2011 3

(3.5) =2.5 1.5 0.5

Recalling that

(t) =
t t x
x e dx

z
1
0


we can certainly write

() =

x e dx
x

z
1
0


If follows that we could use as a probability density the function

f(x) =
1
0
1
( )
( )



x e x
x
> I

We will call this the gamma probability law with parameters and . There are other
notational schemes; it is more common to see this with the roles of and exchanged.



Now about that last detail,
1
2
a f = . Start with

1
2
0
1
2
a f =

z
x e dx
x


and make change x = y
2
. Then dx =2y dy and

1
2
2
0
1
2
2
2 a f c h
=

z
y e ydy
y
= 2
2
0
e dy
y

z



Now make the change to polar coordinates. This change is

Lets write the square of this, once with x as the variable and once with y as the variable:

2
1
2
0 0
4
2 2
a f =
R
S
T
U
V
W
R
S
T
U
V
W

z z
e dy e dx
y x
= 4
2 2
0 0
e dxdy
x y +

z z
d i

FACTS ABOUT THE GAMMA INTEGRAL

Page gs 2011 4

x r
y r
=
=
R
S
T
cos
sin



with, of course, dx dy = rdrd. The set { (x, y)0 <x <, 0 <y < } is mapped into
the set { (r, )0 <r <, 0 < <
2

}.

Our integral is now

( )
2 2
2 2
2
1
2
0 0 0 0
4 4
r r
e r dr d d r e dr



= =



The integral is certainly
2

. As for the r integral, use the change of variable s =r


2
, so
that ds =2rdr, leading to

r e dr e ds e
r s s
s
s

=
=

= = =
z z
2
1
2
1
2
0 0 0
1
2
c h



Thus, the whole calculation comes to

2
1
2
4
2
1
2
a f = =



The desired conclusion follows immediately.

You might also like