Stellenbosch University

Mathematics 244: Analysis

Gamma and Beta Functions1

R∞ R1
The integrals 0 ex xn−1 dx and 0 xm−1 (1 − x)n−1 dx, with m ∈ R and n ∈ R, are important in
some applications. The gamma and beta functions are defined in terms of these integrals, and these
functions are convenient tools for calculating certain difficult integrals.

The Gamma Function

The gamma function, denoted by Γ, was introduced in an effort to generalize the factorial function to
non-integer real numbers. We will see that Γ (n + 1) = n! for n ∈ N.
R∞
Theorem 1. The integral I = 0 e−x xn−1 dx is convergent if n is a positive real number, and diver-
gent if n ≤ 0.
Definition 1. The gamma function Γ : (0, ∞) → R is defined for each positive real number n by
Z ∞
Γ(n) = e−x xn−1 dx.
0

Example. (1)
Z ∞
Γ(1) = e−x x1−1 dx
0
Z ∞
= e−x dx
0
Z t
= lim e−x dx
t→∞ 0
= 1.

Theorem 2. (1) Γ(n + 1) = nΓ(n) for each n ∈ (0, ∞).

(2) Γ(n + 1) = n! for each n ∈ N.
R∞
(3) 0 e−ax xn−1 dx = Γ(n) an for each n ∈ (0, ∞), and for a ∈ (0, ∞).
1
(4) Γ(n) = π − 2 2n−1 Γ n2 Γ n+1



2 for each n ∈ (0, ∞).
R ∞ −y2 2n−1
(5) Γ(n) = 2 0 e y dx for each n ∈ (0, ∞).
R1 n−1
(6) Γ(n) = 0 ln y1 dy for each n ∈ (0, ∞).

Proof. (1) We apply Definition 1 and obtain

1 These notes are based on an Afrikaans set of notes composed by a previous Mathematics 244 lecturer.

1
Stellenbosch University Mathematics 244

= nI'(n).

(2) Die resultaat word met behulp van die Beginsel van Wiskundige lnduksie bewys. Laat
S = {n EN: r(n+l) = n!}. Omdat r(l+l) = 1r(1) = 1 = l! (uit (1) en Voorbeeld
(1)) volg dat 1 E S. Gestel dat k � S vir 'n k E N. Dan geld dat f(k + 1) = kl.
Dan geld, van (1), dat I'((k + 1) + l) = (k + l)r(k + 1) = (k + l)k! = (k + 1)! V11n
die Beginsel
(2) Exercise van Wiskundige
(use mathematical lnduksie volg dat S = N, dit "lvil se, r(n + 1) = n! vir
induction).
n

(3) 1 e-=x -l dx
elke EN.
"°
(3) We split the integral at 1:

- 1
1
e-•,:xn-1 da; + loo e-•:rxn-1 dx

1
- lim
s➔o+
J.
8
e-""'x"- 1 dx + Lim
t�oo Jf'1 e-""'x"-
1
dx.
Set ·u = aa; in both integ)'als on the right hand side. Then,

r(n)

(4) Ditkan
(4) Omit bewys
proof. word.
(5) We apply Definition 1 and then split the integral at 1:

2
Stellenbosch University Mathematics 244

(5) I'(n) = 100 e -"'x"- 1 dx

t
= Jim [1 e-"'x"- 1 dx + Jim i e-"x"- 1 dx
j
.,➔O+- s £-too 1

1"°
Set x. = y2 in both integrals on the right-hand side. Then,

e-zx -l dx
n

- , l...im
o+ ,Ii
f
1 e-v' y 2
"-2 2ydy + lim
t➔oo
1·.,li
1
cu' y2"-2 2y dy

- 2 fo°" - 2n-L dy.

e 11'y

(6) Set
(6) Setye-"
= e−x in=
y at
both iil (5) above
(*)integrals andthen
at (*) and finish thethe
finish proof.
proof.

In definisie2. 1For
Definition is ndie funksie
≤ 0, the function r
op Γ(0,
is oo) gedefinieer.
defined recursively, Die van r kan
definisieversame!iJlg
as follows:
vergroot
(a) For n word, diedefine
en we u.itbreiding
1°"
Γ(n) = van+r1).na die groter definisieversameling word steeds
1
∈ (−1, 0), n Γ(n
1 1
(b) IfrΓ(n)
deur aangedu.i . Omdat
is defined for n ∈die iiltegraal
(−m, −m + 1), thene-"x•-
Γ(n) = dx vir n � 0 divergeer, kan hierd.ie
n Γ(n + 1) for n ∈ (−m − 1, −m), with
m, n ∈ N.

integraal nie
Examples. (2)gebruik word om r vir n � 0 te defrnieer
  IUe. Die uitbreiding van r na nie­
= (−2)Γwyse, .en op s6 'n manier dat f(n+ 1) =
positiewe wa.a.rdes van n geskied opΓ'n −rekursiewe
1 1
nr(n) vir sekere waardes van n steeds geld.
2 2

r word vir n � 0 rekursief soos volg gedefinieer:

(3)
Definisie

2.   
Die funksie
           
7 2 5 2 2 3 2 2 2 1
=
Γ − = − Γ − = − − Γ − = − − − Γ −
(a) Vir n E2(-1,0) 7isf(n) 2 tr(n+7 1). 5

2
 
7

5
 
3 2
 
2 2 2 1 16 1
= − − − (−2)Γ = Γ .
(b) As I'(n} gedefinieer is vir n E (-m,-m + 1) 1 clan 1s f(n) - tr(n +
7 5 3 2 105 2 1) vir
n E It(-m
Remark. can -1,-m), wa.a.r
be shown that EN.
m continuous
Γ is on (0, ∞). Moreover, Γ(n) → ∞ if n → 0+ , and
Γ(n+1) −
Γ(n) = → −∞ if n → 0 . The function Γ is not (and will not be) defined at 0 and has an
Voorbeelde.
infinite
n
discontinuity =
(2) at r0.(- ½) (-2)r (½)-
From Definition 2 we can now deduce that, for each integer k < 0, Γ(n) → ∞ if n → k + , and
(3) →r(-D
Γ(n) =
−∞ if k → (-�)r(-i)
0− . Thus, Γ has =
(-�)
infinite r( -�) (-at
(-ndiscontinuities =
i)( -J)r(
�)(all-negative - ½) The
integers. =
l�;rfunction
(½)- Γ is
not (and will not be) defined at negative integers.
n
Dit kan aangetoon word dat r kontinu is op (O,oo). Verder, I'(n) = I'( n+ l) ➔ oo as

n-+ o+, en r(n) = r(n + l) ➔ -oo as n ➔ 0-. Die funksie r word dus nie by die getal
n
0 gedefin.ieer nie. Ons lei af dat r dus ook nie vir negatiewe heeltallige waa.rdes van n
gedefinieer is nie. Die grafiek van r sien nou
3 soos volg daar u.it:

B 4
Stellenbosch University Mathematics 244

Plot created by Alessio Damato using Gnuplot.

Z ∞
Γ(5) 4! 3
Examples. (4) x4 e−2x dx = = 5 = . . . . by Theorem 2(3)
0 25 2 4

                 
9 7 7 7 5 5 7 5 3 1 1
(5) Γ = Γ = Γ = ··· = Γ . . . . by Theorem 2(1)
2 2 2 2 2 2 2 2 2 2 2

Z ∞  
−x3 1 1
(6) We will show that e dx = Γ .
0 3 3

Proof. Note that

Z ∞ Z 1 Z ∞
−x3 −x3 3
e dx = e dx + e−x dx
0 0 1
Z 1 Z t
3 3
−x
= e dx + lim e−x dx
0 t→∞ 1
Z 1 Z t
3 3
= lim+ e−x dx + lim e−x dx
s→0 s t→∞ 1

4
Stellenbosch University Mathematics 244

Now, set u = x3 in both integrals on the right-hand side. Then

Z ∞ Z 1 Z t3
−x3 1 − 2 −u 1 − 2 −u
e dx = lim u e du + lim
3 u 3 e du
0 s→0 +
s3 3 t→∞ 1 3
Z ∞
1 − 2 −u
= u 3 e du
0 3
1 ∞ 1 −1 −u
Z
= u 3 e du
3 0
 
1 1
= Γ . . . . by Definition 1
3 3

Γ(3)Γ 25 2! · 32 · 21 · Γ 12



16
(7)
= 1 = 315 .
Γ 11 9 7 5 3 1


2 2 · 2 · 2 · 2 · 2 ·Γ 2

Z ∞
5
e−2x x3 dx in terms of Γ 4


(8) Determine 5 .
0

Z t
5
Solution. Let t > 0 and set I = e−2x x3 dx. Now, set v = x5 . Then,
0

Z t5
1 3 4
I= e−2v v 5 v − 5 dv.
5 0
So,
Z ∞ Z t
−2x5 3 5
e x dx = lim e−2x x3 dx
0 t→∞ 0
Z t5
1 1
= lim e−2v v − 5 dv
t→∞ 5 0
Z t5
1 1
= lim e−2v v − 5 dv
5 t→∞ 0
1 ∞ −2v 4 −1
Z
= e v 5 dv
5 0
1 Γ 45


= · 4 .
5 25

Z ∞
2
(9) Determine 3−4x dx in terms of Γ (n) for some n ∈ (−1, 1)\{0}.
0

Solution. Write
Z ∞ Z t
−4x2 2
3 dx = lim 3−4x dx
0 t→∞ 0
Z t
2
= lim e−4x ln 3
dx.
t→∞ 0

5
Stellenbosch University Mathematics 244

Now, set v = 4x2 ln 3 and use the same method as in (9) above. Then we obtain
Z ∞
Γ 12


2
3−4x dx = 1
0 4 (ln 3) 2

The Beta Function

We will no introduce the beta function which is closely related to the gamma function, binomial
coefficients, and the beta distribution (used in probability theory).

Z 1
Theorem 3. The integral xm−1 (1 − x)n−1 dx is convergent if m > 0 and n > 0, and divergent
0

otherwise.
Definition 3. The beta function β : (0, ∞) × (0, ∞) → R is defined for positive real numbers m and
n by Z 1
β(m, n) = xm−1 (1 − x)n−1 dx.
0

Theorem 4. For all m, n ∈ (0, ∞) the following hold.

(1) β(m, n) = β(n, m).

∞
xm−1
Z
(2) β(m, n) = dx.
0 (1 + x)m+n

Z π
2
(3) β(m, n) = 2 sin2m−1 θ·cos2n−1 θ dθ.
0

m
(4) β(m+1, n) = β(m, n).
m+n

m n
(5) β(m+1, n+1) = · β(m, n).
m+n+1 m+n

6
Stellenbosch University Mathematics 244

Γ(m)Γ(n)
(6) β(m, n) = .
Γ(m + n)

√
 
1
(7) Γ = π.
2

Proof. (1) We see that

Now, set 1 - x = u. Then -d:r; = du. Therefore,

fJ(m,n) =

- fJ(n,m).

2) fJ(m, n)
((2) We observe the following.

! Ii'
1
'
2
= lim
. ,-1-0+ .s
1 1
u - (1- ur- du+ lim
m
t➔l-
um - 1 (1- ur- 1 du .
1

x l 1 dx
Now, set u = --. Then 1 - 11, = --, x = -- - 1 and d-u = ( ) 2•
l+x l+x 1 -u

1
I+x
Therefore,
1
7 xm -1 l dx
P(m,n) - Jim
s-40+ _1 _ 1 (l + x)m-1 (1 + x)n-1 (1 + z)2
+
1-•
 - fJ(n,m).
Stellenbosch University Mathematics 244

- fJ(n,m).
(2) fJ(m, n)
(2) fJ(m, n)

! Ii'
1! u
'
2
=
Ii'
1 1
um - 1 (1- ur- 1 du .

1
lim m
- (1- ur- du+ ' lim
. ,-1-0+ .s 2 t➔l
= lim
. ,-1-0+ .s
um - 1 (1- ur- 1 du+ lim
t➔l- 1
- 1
um - 1 (1- ur- 1 du .
x l 1 dx
Now, set u = --. Then 1 - 11, = --, x = -- - 1 and d-u = ( ) •
l+x
x l+x 1 -u
1 dx 2

11
l I+x
Now, set u = --. Then 1 - 11, = --, x = -- - 1 and d-u = (
Therefore, ) •
l+x l+x 1 -u I+x 2
Therefore, 1 xm -1 l
- dx
P(m,n) Jim +
s-40+ _1 _ (l x
1
1-• 1
+m -1
x)m-1 (1 +lx)n-1 +
( dxz)2
1
P(m,n) - Jim
1
s-40+ _1 _ 1 1+z 2 +
(l + x)m- (1 + x)n-1 ( )
1-•

(3) fJ(m,n)
(3) In the definition of β (m, n), we split the integral into a sum and write each as a limit, as follows:

(3) fJ(m,n) l
t
= j
lim ll xm- 1 (1-xr- 1
,
dx. + Jim xm- 1 (1-x)"- 1 dx.
lt
= j
s➔O+ s,
lim l x - (1-xr- dx
m 1 1 t-tl- ! m 1
. + Jim x - (1-x)"- 1 dx.
l
Now, set x = sin 0 (0 $ 0 $ 1r/2). Then dx = 2 sin0 · cos 0d0. Therefore,
s➔O+ s 2 t-tl- !

Now, set x = sin2 0 (0 0 $ 1r/2). Then dx = 2 sin0 · cos 0d0. Therefore,

ll
$,/f
rcsin

/J(m,n) = lim arcsin,/f sin2m-2 0 • cos2"-2 0 · 2 sin 0 · cos 0d0

/J(m,n) = .s->o+
lim a in VS sin2m-2 0 • cos2"-2 0 · 2 sin 0 · cos 0d0
a.res
.s->o+ a.resin VS
B 9
B 9

(4) We apply Definition 3 and obtain

8
Stellenbosch University Mathematics 244

9
Stellenbosch University Mathematics 244



 
  
  



  
  




=
   
!_ 5

    
Therefore, So that #"
 
 "" 
2s ! ""
 "" 4H€fYN]TX
  


 "

(5) We apply parts (1) and (4) of this theorem and obtain




 " " " (by (4))
 
"



" "

 (by (1))
„P{„Z

 
  (by (4))
!
"
 "$%&  `

 
" !"
 "
"
 (by (1))


"

(6) Omit proof (double integrals are used).

(7) It follows from 
T}FfYYfs}O<}
  &ofM†, (3) that
Œ
 
zVd
‡Bgv 



 (Ž



Therefore,
3w p ˆ * 




 p‰ pŠ 
 
?HC<sHq#   $ 

 

FROM



 

3w 


Therefore,

10
Stellenbosch University Mathematics 244

Examples. (10)
Z π Z π
!
2
3 1 2
2(2)−1 2( 12 )−1
sin θ dθ = 2 sin θ · cos θ dθ
0 2 0
 
1 1
= β 2,
2 2
1 Γ (2) Γ 21


= ·
Γ 52


2
2
= .
3

Z 1
1 1
(11) Determine √ √ dx.
0 x 1−x

Solution. We see that

Z 1 Z 1
1 1 1 −1
√ √ dx = x− 2 · (1 − x) 2 dx
0 x 1−x 0
 
1 1
=β , . . . by Theorem 3
2 2
Γ 21 Γ 12



=
Γ 21 + 21


√ √
π π
= . . . by Theorem 4(7)
1
= π.

Z a p
(12) Suppose that a > 0. Determine t2 a2 − t2 dtby using (i) a trigonometric substitution and
0

(ii) the beta function.

Solution.
a4 π
(i) Set t = a sin θ and the answer will be 16 .
(ii) We will try to write the integral in a form that can be recognized as a value of the beta
function.
v
Z a p Z a u  2 !
2 2 2
u
2t 2 t
t a − t dt = t a 1− dt
0 0 a
s  2
Z a
2 t
= at 1 − dt.
0 a

11
Stellenbosch University Mathematics 244

t 2 t2 2t dt a2 dx 2
a√


Now, set x = a = a2 . Then dx = a2 and thus dt = 2t = 2a x
dx. Therefore,

a 1√ a2
Z p Z
t2 a2 − t2 dt = aa2 x 1 − x √ dx
0 0 2a x
4 Z 1
a 1 1
= x 2 (1 − x) 2 dx
2 0
a4 1 3 −1
Z
3
= x 2 (1 − x) 2 −1 dx
2 0
a4
 
3 3
= β ,
2 2 2
a4 Γ 23 Γ 32



= ·
Γ 32 + 32


2
a4 12 Γ 12 12 Γ 12



= ·
2 Γ (3)
4 1√ 1√
a π2 π
= · 2
2 2!
a4 14 π
=
2·2
a4 π
= .
16

(13)
Z π Z π
!
2 1 2
2( 72 )−1
sin6 θ · cos5 θ dθ = 2 sin θ · cos2(3)−1 θ dθ
0 2 0
 
1 7
= β ,3
2 2
7


1 Γ 2 Γ (3)
= · 7


2 Γ 2 +3
8
= .
693

∞
√
Z
3
(14) Calculate ye−y dy if it exists.
0

Solution.
∞ t
√ √
Z Z
3 3
ye−y dy = lim ye−y dy.
0 t→∞ 0

12
Stellenbosch University Mathematics 244

Set u = y 3 . Then
∞ t3
√
Z Z
−y 3 1 1 2
ye dy = lim u 6 e−u u− 3 du
0 t→∞ 0 3
Z t3
1 1
= lim u− 2 e−u du
3 t→∞ 0
1 ∞ − 1 −u
Z
= u 2 e du
3 0
 
1 1
= Γ
3 2
√
π
= .
3

(15)
Z 1
x4 (1 − x)3 dx = β (5, 4)
0
Γ (5) Γ (4)
=
Γ (9)
4!3!
=
8!
1
= .
280

13