Gamma and Beta Functions

May 17, 2023

Gamma function, Γ(p)

We are going to learn Gamma function Γ(p) which helps us to express special functions in terms of
it. This function is found in various branches of physics e. g., statistical mechanics, condensed matter
physics. We going to start with an interesting function namely, Factorial Function (optional).

Factorial Function
Consider the following integral
Z ∞
1 ∞ 1 1
e−αx dx = − e−αx 0 = − e−α∞ − e0 =

(1)
0 α α α
Now differentiate both sides of equation (1) with respect to α repeatedly
Z ∞  
d −αx d 1
e dx =
dα dα α
Z ∞ 0
d  −αx  1
or, e dx = − 2
0 dα α
Z ∞
1
or, −x e−αx dx = − 2
0 α

Z ∞ Z ∞
d −αx 1
Finally, e dx = x e−αx dx = (2)
dα 0 0 α2
The repeated differentiation of equation (1) with respect to α gives
Z ∞
d2 1.2
2
→ x2 e−αx dx = 3
dα 0 α
Z ∞
d3 1.2.3
→ x3 e−αx dx =
dα3 0 α4

Z ∞
n!
or in general, xn e−αx dx = (3)
0 αn+1
putting α = 1, we get
Z ∞
xn e−x dx = n! (4)
0

where n = 0, 1, 2, 3, · · · Thus we have a definite integral whose value is n! for positive integer n. We
can use equation (3) to give a meaning 0!. Putting n = 0 in (4) we get
Z ∞
∞
e−x dx = −e−x 0 = 1

0! =
0

1
 Definition of Gamma function: Γ(p)
Normally, n! denotes the factorial of positive integer , and for factorial of non integral number we use
gamma (Γ) function. We define, for any p > 0,
Z ∞
Γ(p) = xp−1 e−x dx p>0 (5)
0

For 0 < p < 1, this is an improper integral because xp−1 becomes infinite at the lower limit(do
by yourself). However, it is a convergent integral for p > 0. For, p ≤ 0 the integral diverges and so
cannot be used to define Γ(p). For, positive integer (n > 0) with (4), (5) becomes
Z ∞
Γ(n) = xn−1 e−x dx = (n − 1)! (6)
0

replacing n by n + 1 in (6)gives
Z ∞
Γ(n + 1) = xn e−x dx = n! (7)
0

Gamma function of any number(positive) gives the factorial of its previous one. From (7), we can
define
Z ∞
Γ(1) = e−x dx = 0! = 1
0

Recursion relation for the Γ(p) function

Z ∞
Replacing p by (p + 1) in Γ(p) = xp−1 e−x dx, we have
0
Z ∞
Γ(p + 1) = xp e−x dx = p!, p > −1 (8)
0

Notice that for Γ(p) condition was p > 0, and now for Γ(p + 1) condition was p > −1; such that, we
can have Γ(1) = 0! = 1. Z Z 
Let us integrate (8) by parts, udv = uv − vdu , calling u = xp , dv = e−x dx; we get du =
p xp−1 dx, v = −e−x . substituting in (8)

Z ∞
∞
−xp e−x 0 (−e−x ) p xp−1 dx

Γ(p + 1) = −
0

The integrated term is zero at the both limits [ xp = 0 for 0 and e−x = 0 for ∞], and the integral
term is nothing but the Gamma function of number p with the multiplication of p, i. e., Γ(p), so, the
expression for the Γ(p + 1) becomes
Z ∞
Γ(p + 1) = 0 + p xp−1 e−x dx = p Γ(p) (9)
0

The equation (9) is called the recursion relation [which relates the gamma functions of two successive
numbers differ by 1] for the Γ function. Repeated use of (9)

Γ(p + 1) = pΓ(p) = p(p − 1)Γ(p − 1)

Γ(p + 1) = p(p − 1)(p − 2)(p − 3) · · · Γ(p − p + 1) = p!

If we replace by p by n, Γ(p + 1) becomes Γ(n + 1) = n!

2
The Gamma function of Negative numbers
For p ≤ 0, Γ(p) is defined by the recursion relation, Γ(p + 1) = pΓ(p), solved for Γ(p)
1
Γ(p) = Γ(p + 1) (10)
p

defines Γ(p) for p < 0.

1
Example: Γ(−0.3) = Γ(0.7) and so on. Since Γ(1) = 0! = 1, we see that Γ(p) = (1/p) Γ(p + 1)
−0.3
tends to infinity as p goes to zero. Which gives Γ(−1) = Γ(−2) = Γ(−3) = · · · = 0, i. e. Γ(p) is
zero at p = 0 and all negative integers. In the intervals between negative integers, it is alternatively
positive and negative, negative from 0 to −1 and positive from −1 to −2.

 p 

Figure 1: Gamma function

Important formulas involving Gamma function

First we evaluate Γ(1/2). By definition
Z ∞
1
Γ(1/2) = √ e−t dt (11)
0 t

where we use dummy variable t, which does not change the value of integral. Put t = y 2 in (11); then
dt = 2y dy, and (11) becomes
Z ∞ Z ∞
1 −y2 2
Γ(1/2) = e 2y dy = 2 e−y dy (12)
0 y 0

or, with x in (12) gives

Z ∞ Z ∞
2 2
Γ(1/2) = 2 e−x dx = e−x dx (13)
0 −∞
Z ∞
2 2
The integral e−x dx is the well-known Gaussian integral of Gaussian function f (x) = e−x over
−∞
the entire real line, also known as the Euler–Poisson integral. Named after the German mathematician
Carl Friedrich Gauss.

3
 Figure 2: Integral of Gauss function

Multiply these two integrals for Γ(1/2) together and write the result as a double integrals
Z ∞Z ∞
2 2
2
[Γ(1/2)] = 4 e−(x +y ) dx dy
0 0

this integral need to be calculated over the first quadrant of xy−plane since x, y > 0 ; it can be
evaluated more easily in polar coordinate writing x2 + y 2 = r2 and dx dy = r dr dθ:
Z ∞ Z π/2
2
[Γ(1/2)]2 = 4 e−r rdr dθ
0 0

Now, let r2 = u, then du = 2r dr, rdr = du/2 and integrating for θ

π ∞ −u du
Z Z ∞
∞
[Γ(1/2)]2 = 4 e =π e−u du = π (−e−u )0
2 0 2 0

the integral gives zero at upper limit and 1 at lower limit, hence
√
[Γ(1/2)]2 = π ⇒ Γ(1/2) = π (14)
Z ∞
2
Another formula: Γ(p) = 2 x2p−1 e−x dx.
0
Let us start writing the formula for gamma function we know
Z ∞
Γ(p) = xp−1 e−x dx (15)
0

let x = t2 , hence dx = 2t dt, the limits are: x = 0, t = 0, x = ∞, t = ∞. Substituting in (15)

Z ∞ Z ∞
2(p−1) −t2 2
Γ(p) = t e 2t dt = 2 t2p−1 e−t dt
0 0

with variable x last equation becomes

Z ∞
2
Γ(p) = 2 x2p−1 e−x dx (16)
0

Problems:

Γ(p + 1) √
1. Find Γ(− 21 ). Hint: Use the recursion relation for negative number, Γ(p) = . Ans:−2 π
p
Z ∞
√ √ √
2. Evaluate: I = 4
x e x dx. Hint: Let x = t or t2 = x substituting in our given equation
0 Z
∞ 5
3√
with limits gives Γ(p) = 2 t 2 −1 e−t dt = π (using recursion relation).
0 2

4
 ∞
xa
Z
3. Evaluate: I = x
dx. Hint: Put ax = et , ln ax = ln et , x lna = t, dx = dt/lna; substituting
0 a
in our given problem gives I = Γ(a + 1) / ln(a + 1)

4. Prove
 that  : For n being integer,
1 1.3.5 · · · (2n − 1) √ (2n − 1)!! √ (2n)! √
(i) Γ n + = n
π= n
π= n π
2 2 2 4 n!
2n √ 2n √ 4n n! √

1
(ii) Γ − n = (−1)n π = (−1)n π(−1)n = π
2    · · · (2n − 1)
1.3.5 (2n − 1)!! (2n)!
1 1 π
(iii) Γ n + Γ −n = = (−1)n π
2 2 cos nπ
Solution:
(i) We know, the recursion relation of gamma function for n > 0

Γ(p + 1) = pΓ(p)

Now,
     
1 1 1
Γ n+ = n−1+ Γ n−1+
2 2 2

[use of recursion relation]

      
1 1 1 1
Γ n+ = n− n−2+ Γ n−2+
2 2 2 2

[again use of recursion relation] The recursion relation goes upto Γ(1/2)
 which
 occurs when we subtract
1
n in the argument of gamma function i. e., Γ n − n + 12 . So, Γ n +


becomes
2
       
1 2n − 1 2n − 3 2n − 5 1
Γ n+ = ···Γ n − n +
2 2 2 2 2

       
1 2n − 1 2n − 3 2n − 5 1 1
⇒Γ n+ = ··· Γ
2 2 2 2 2 2

[We write previous number with Γ(1/2), so that, we can write (1/2)Γ(1/2) = Γ(3/2)]. Therefore,
       
1 2n − 1 2n − 3 2n − 5 3 3
Γ n+ = ··· Γ
2 2 2 2 2 2
       
1 2n − 1 2n − 3 2n − 5 5 5
Γ n+ = ··· Γ
2 2 2 2 2 2

After rearranging we can finally write

       
1 2n − 1 2n − 3 2n − 5 5 5
Γ n+ = ··· Γ
2 2 2 2 2 2
 
1 1 3 5 2n − 5 2n − 3 2n − 1 1
Γ n+ = · · ··· · · · Γ( )
2 2 2 2 2 2 2 2
√ (2n − 1)!! √
 
1 1 · 3 · 5 · · · (2n − 5)(2n − 3)(2n − 1)
Γ n+ = · π = π
2 2n 2n
√
[2 comes in denominator n times, Γ(1/2) = π, and the product in numerator written as the double
factorial of last number, the difference between single and double factorial notation are: 6! =
6 · 5 · 4 · 3 · 2 · 1 = 720; 6!! = 6 · 4 · 2 = 48]

5
 We proved
 the first and second part of (i), now multiply both numerator and denominator by 2n n!,
1
Γ n+ is
2
1 · 3 · 5 · · · (2n − 3)(2n − 1) (2n )n! √
 
1
Γ n+ = · n · π
2 2n (2 )n!
1 · 3 · 5 · · · (2n − 3)(2n − 1) 2n · 2(n − 1) · · · (2 · 2)(2 · 1) √
 
1
Γ n+ = · · π
2 2n (2n )n!
[In numerator we used of definition n! and each term is multiplied by 2 since 2n and n! both have n
terms] Rearranging
1 · 2 · 3 · 4 · · · (2n − 3)(2n − 2)(2n − 1)(2n) √
 
1
Γ n+ = · π
2 2n 2n n!
(2n)! √
 
1
Γ n+ = n · π
2 4 n!
Γ(p + 1)
(ii) For this proof, use recursion relation for n < 0 i.e., Γ(p) = [Do by yourself].
p
(iii) multiply Zthe results from (i) and (ii).
∞ Z ∞
5 −x2 2
5. Evaluate: x e dx, Hint: compare with Γ(p) = 2 x2p−1 e−x dx, Ans: 1.
Z0 ∞ 0
−x3 3 Γ(2/3)
6. Evaluate: xe dx, Hint: let x = u, Ans: .
Z0 ∞ 3
n Γ(m + 1/n)
7. Evaluate: xm e−x dx, Hint: let xn = u, Ans: .
0 n
1 p−1
Z 1   
8. Evaluate: ln dx, Hint: let x = e−u , Ans: Γ(p).
x
Z0 1   3
1 1
9. Evaluate: x2 ln dx, Hint: let x = e−u , Ans: Γ(2).
0 Z x 3
∞
1
10. Prove that: xn e−ax dx = n+1 Γ(n + 1).
0 Z a∞
2 s!
11. Show that, for integer s, (i) x2s+1 e−ax dx = s+1
,
0 2 ar
Γ(s + 21 ) Γ(s + 21 )
Z ∞
2 (2s − 1)!! π
(ii) x2s e−ax dx = 1 = s+1/2
= s+1 s
0 s+
2a 2 2a 2 a a
Z 1√
3
11. Evaluate: ln xdx
0

Beta Function B(p, q)

We are going to know Beta B(p, q), which can be written in terms of gamma function, and also helps
us to calculate various integral very easily.
The beta function is defined by a definite integral
Z 1
B(p, q) = xp−1 (1 − x)q−1 dx p > 0, q > 0 (17)
0

There are a number of simple transformations of (1) which are useful to know

B(p, q) = B(q, p) (18)

The range of integration in (1) can be changed by putting x = y/a; then x = 1 corresponds to y = a;
and (1) becomes
Z a  Z a
y p−1  y q−1 dy 1
B(p, q) = 1− = p+q−1 y p−1 (1 − y)q−1 dy (19)
0 a a a a 0

6
To obtain trigonometric(most used) form of the beta function put x = sin2 θ, then dx = 2 sin θ cos θ dθ,
1 − x = 1 − sin2 θ = cos2 θ; x = 1 corresponding to π/2. With these substitutions, (1) becomes
Z π/2
B(p, q) = (sin2 θ)p−1 (cos2 θ)q−1 2 sin θ cos θ dθ
0

Z π/2
⇒ B(p, q) = 2 sin2p−1 θ cos2q−1 θ dθ (20)
0

Finally, let x = y/y + 1 in (1); then we get

1 p−1  q−1
(1 + y) dy − y dy
Z
y y
B(p, q) = 1−
0 1+y 1+y (1 + y)2
1
y p−1
Z
B(p, q) = dy (21)
0 (1 + y)p+q

Beta function in terms of gamma function

The relation between beta and gamma functions is

Γ(p)Γ(q)
B(p, q) = (22)
Γ(p + q)

Thus we can evaluate a B function in terms of Γ functions.

To prove (22), we start with a formula of gamma function
Z ∞
2
Γ(p) = 2 y 2p−1 e−y dy
0

Similarly,
Z ∞
2
Γ(q) = 2 x2q−1 e−x dx
0

Multiply these two equations together and change to polar coordinates:

Z ∞Z ∞
2 2
Γ(p)Γ(q) = 4 x2q−1 y 2p−1 e−(x +y ) dx dy
0 0

Z ∞ Z π/2
2
Γ(p)Γ(q) = 4 (r cos θ)2q−1 (r sin θ)2p−1 e−r rdr dθ
0 0

Z ∞ Z π/2
2(p+q)−1 −r2
Γ(p)Γ(q) = 4 r e dr (cos θ)2q−1 (sin θ)2p−1 dθ
0 0

1
[Factored out of all terms involving r and wrote x2 + y 2 = r2 ]. The r integral is Γ(p + q) and
2
1
the θ integral is B(p, q). So, the expression for Γ(p) Γ(q) becomes
2
1 1
Γ(p) Γ(q) = 4 Γ(p + q) B(p, q)
2 2
Γ(p) Γ(q)
B(p, q) = (23)
Γ(p + q)

7
 Z π/2
Example: Consider the integral dθ
0
We know the result will be simply π/2. now calculate this by (7). Writing our given integral as
Z π/2 Z π/2
dθ = sin0 θ cos0 θdθ
0 0

1 1 1
which gives us p = 1/2 and q = 1/2, since 2p − 1 = 0, 2q − 1 = 0; and the integral will be B( , ),
2 2 2
Hence
Γ( 12 ) (Γ( 21 )
Z π/2
1 1 1
dθ = B( , ) =
0 2 2 2 Γ( 12 + 21 )
Z π/2 √ √
1 1 1 1 π π π
dθ = B( , ) = =
0 2 2 2 2 Γ(1) 2
√
[Γ(1/2) = π, Γ(1) = 1], Although our integral was simple but, intrinsically, it contains beta function.

Problems:
π/2 π/2    
1−n
Z Z
1 n+1
1. Evaluate: tann θ dθ, · · · Hint: sinn θ cos−n θ dθ, · · · Ans: Γ Γ
2 2 2
Z0 π/2 Z0 π/2 √ n+1
π Γ( 2 )
2. Evaluate: sinn θ dθ, · · · Hint: sinn θ cos0 θ dθ, · · · Ans:
0 0 2 Γ( n+2
2 )
Z π/2 Z π/2 √ n+1
π Γ( 2 )
3. Evaluate: cosn θ dθ, · · · Hint: sin0 θ cosn θ dθ, · · · Ans:
0 0 2 Γ( n+22 )
Z π/2 Z π/2 Z π/2 p
dθ dθ
4. Evaluate: (i) √ · · · (ii) √ · · · (iii) sin3 x cos x dx
Z 1 0 4 cos θ 0 sin θ 0
x 2 3π
5. Evaluate: √ dx, · · · Hint: put x = u, · · · Ans:
√ 16
2
Z0 1 1 − x
dx π Γ(1/3)
6. Evaluate: √ , · · · Hint: put x3 = u, · · · Ans:
3 3 Γ(5/6)
Z 01 1 − x
Γ(p) Γ(q)
7. Evaluate: (1 + x)p−1 (1 − x)q−1 dx, · · · Hint: put x = cos 2θ, · · · Ans: 2p+q−1 = 2p+q−1 B(p, q)
Γ(p + q)
Z−11    
y dy 3 = u, · · · Ans: 1 Γ 2 Γ 4
8. Evaluate: , · · · Hint: Put y
(1 + y 3 )2 3  3 3
Z0 1 √
3 4
9. Evaluate: ln xdx, · · · Hint: Put x = e−u , · · · Ans: −Γ
0Z 3
1 1 π (2n − 1)!!
10. Evaluate: (1 − x2 ) 2 x2n dx, · · · Hint: Put x = cos θ, · · · Ans: when n = 0; π
−1 2 (2n)!
when n = 1, 2, 3, · ·Z·
1 1
11. Evaluate: (1 − x2 )− 2 x2n dx, · · · Hint: Put x = cos θ
Z−1
1
12. Evaluate: (1 − x2 )n dx, · · · Hint: Put x = cos θ
−1
√
   
1 1
13. Prove that: B(n, n) = B n, /22n−1 and hence Γ(2n) = 22n−1 Γ(n) Γ n + / π (Leg-
2 2
endre’s duplication formula).
Solution: We know,
Z π/2
B(p, q) = 2 (sin θ)2p−1 (cos θ)2q−1 dθ (24)
0

8
 22n−1
Consider p = q = n, and multiply eq. (24) by , Eq. (24) takes the form
22n−1
Z π/2
1
B(n, n) = 2 (2 sin θ cos θ)2n−1 dθ
0 22n−1
Z π/2
2
⇒ B(n, n) = (sin 2θ)2n−1 dθ
22n−1 0

Let 2θ = ϕ; dθ = dϕ/2; θ = 0, ϕ = 0; θ = π/2, ϕ = π. So,

Z π
2 dϕ
B(n, n) = 2n−1 (sin ϕ)2n−1
2 0 2
Z π/2
2
⇒ B(n, n) = 2n−1
2 (sin ϕ)2n−1 cos0 ϕ dϕ
22 0
 
2 1 1
⇒ B(n, n) = 2n−1 B n,
2 2 2
 
1 1
⇒ B(n, n) = 2n−1 B n,
2 2

Which is the first part of of our proof. Where we used the trigonometric form of beta function and
for the second part we will use the relation between gamma and beta function, Hence

1 Γ(n) Γ 12


Γ(n) Γ(n)
= 2n−1
Γ n + 12


Γ(2n) 2
1


2n−1 Γ(n) Γ n
+ 2
⇒ Γ(2n) = 2
Γ 21
1


2n−1 Γ(n) Γ n + 2
⇒ Γ(2n) = 2 √
π

FURTHER READINGS:
1. Mary L. Boas, Mathematical Methods in the Physical Sciences, 3rd ed. John Wiley Sons, Inc.
2006.
2. W.W. Bell, Special Functions for Scientists and Engineers, Dover Books on Mathematics.
3. Arfken, Weber and Harris, Mathematical Methods for Physicists, 7th ed. Academic Press.
4. S. Chowdhury, Lecture Notes on Physics, 2nd ed., American Academic Press.