1

Chapter – 02

TRANSFORMATION TECHNIQUE

To find the probability function or probability density function of the function of a random
variable for a known probability function or probability density function of a random variable.
Transformation changes the shape of a distribution or relationship.

Let X be a random variable with p.m.f or p.d.f f (x ; θ) and let y=h(x) is a function of a random
variable X and it is desired to find the p.m.f or p.d.f of Y. we use the technique of
transformation.

X f (x) Y=h(x) Transform f ( y )=?

Where y = h(x) is deterministic.

Following are the important transformation techniques.

1. One to one transformation technique

2. Moment generating function transformation technique
3. Cumulative probability function transformation technique.

One to One Transformation Technique

The one to one transformation technique is also called change of variable technique. Let X be a
random variable with p.m.f or p.d.f f (x). Where X ∈ S 1. Let Y =h(x) now to find the p.m.f or
p.d.f of Y, whereY ∈ S2. X ∈ S 1 which maps on Y ∈ S2.

The step by step procedure:

i. Find the range of Y using X.

ii. Find the inverse function of X
−1
x=h ( y)
iii. Find the Jacobin of transformation

Let x=h−1 ( y) is the inverse function of X. then the Jacobin of transformation is defined as:

|J|= |dyd h ( y)|

−1

Or
 2

|J|= | |dx
dy

iv. Use the formula

The p.d.f of Y can be obtained by using one to one transformation as:

f ( y )=f x ( y ) ×|J|

Examples on Univariate cases

Example 1: Let X be a random variable with p.df.

f ( x )=2 x 0 ≤ x ≤ 1

Determine the p.d.f of y=8 x 3, using change of variable transformation technique.

Solution: The p.d.f of X is given by:

f ( x )=2 x

y=8 x

When x = 0, y = 0 and when x = 1, y = 8. That’s x ∈ ( 0 , 1 ) which maps on y ∈(0 , 8)

3 1
x= y
8

( )y
3
3 1
x=
2
1
1
x= y 3
2

| |
−2
dx 1 1
|J|= = × y 3
dy 2 3

f ( y )=f x ( y ) ×|J|

( )
1 −2
1 3 1
f ( y )=2 y × y 3
2 6
−1
1 3
f ( x )= y
6
 3

Example 2: The p.d.f of a continuous random variable X is given by:

f ( x )=4 x 3 0 ≤ x ≤ 1

Determine the the p.d.f of y=−2 ln x 4

Solution: The p.d.f X is given by:

3
f ( x )=4 x 0≤ x≤1
4
y=−2 ln x

When x = 0, y=∞ and when x = 1, y=0 that’s x ∈ ( 0 , 1 )which maps on y ∈ ( 0 , ∞ )

4
y=−2 ln x
−8
y=ln x
−8 y
x =e
−1
y
8
x=e

| || |
−1 −1
dx −1 y 1 y
|J|= = e 8
= e 8
dy 8 8

f ( y )=f x ( y ) ×|J|

( ) × 18 e
−1 3 −1
y y
8 8
f ( y )=4 e

−3 −1
8
y 1 8
y
f ( y )=4 e e
8
−1
1 2
y
f ( y )= e
2

Example 3: Let X have a logistic p.d.f

−x
e
f ( x )= 2
−∞ < x< ∞
( 1+ e−x )
Determine the p.d.f of y=e−x
 4

Solution: The p.d.f of X

−x
e
f ( x )= 2
( 1+e− x )
−x
y=e

When x=−∞, y=∞ and when x=∞ , y=0 that’s x ∈ (−∞ , ∞ ) which maps on y ∈ ( 0 , ∞ ).

ln y =¿−x ¿

x=−lny

|J|= |dxdy|=|−lny|= 1y
f ( y )=f x ( y ) ×|J|

y 1
f ( y )= ×
( 1+ y ) y
2

1
f ( y )=
( 1+ y )2

Example 4: The p.d.f of a continuous random variable is given by:

1 m −1 n−1
f ( x )= x ( 1−x ) 0≤ x≤1
β ( m, n )

Determine the p.d.f of y=−lnx and identify the p.d.f for n = 1

Solution: The p.d.f X is given by:

1 m −1 n−1
f ( x )= x ( 1−x ) 0≤ x≤1
β ( m, n )

y=−lnx

When x = 0, y=∞ and when x =1 y = 0, that’s x ∈(0 ,1) which maps on y ∈ ( 0 , ∞ )

−y
x=e

|J|=|dxdy|=|−e −y
|=e− y

f ( y )=f x ( y ) ×|J|
 5

1 m−1 n−1
f ( y )= ( e− y ) ( 1−e− y ) × e− y
β ( m ,n )

1 − y n−1
e ( 1−e )
−my+ y − y
f ( y )= e
β ( m ,n )

1 − y n−1
e ( 1−e )
−my
f ( y )=
β ( m ,n )

For n = 1

1 −my
f ( y )= e
β ( m ,1 )

Γ m+1 −my
f ( y )= e
Γm
−my
f ( y )=m e

Which is the p.d.f of an exponential distribution with parameter m.

Example 5: If Xis N ( μ , σ 2 ), then find the p.d.f of y=e x

Solution: The p.d.f of X is

( )
2
−1 x− μ
1 2 σ
f ( x )= e −∞ ≤ x ≤ ∞
√2 π σ
When x=−∞, y = 0 and when x=∞ , y = ∞ , that’s x ∈ (−∞ , ∞ )which maps on y ∈ ( 0 , ∞ )

x=ln y

|J|= |dxdy|=|lny|= 1y
f ( y )=f x ( y ) ×|J|

( )
2
−1 lny −μ
1 2 σ
f ( y )= e ×1/ y
σ √2 π

( )
2
−1 lny−μ
1 2 σ
f ( y )= e
σy √ 2 π

The resultant p.d.f is the p.d.f of lognormal distribution.

 6

Example 6: If X~N (0, 1). Find the p.d.f of y=x 2

Solution: If X~N (0, 1) then the p.d.f of X is given by:

−1 2
1 2
x
f ( x )= e −∞ < x <∞
√2 π
2
y=x

When x=−∞, y=∞ and when x=∞ , then y=∞, so the sample space for y is from 0 to ∞ .

x ∈ (−∞ , ∞ ) Which maps on y ∈ ( 0 , ∞ ) .

1
2
x=± y

| |
−1
dx 1
|J|= = y 2
dy 2

f ( y )=[ f x (−√ y )+ f x ( √ y ) ] ×|J|

[ ]
−1 2 −1 2 −1
1 y 1 y 1
f ( y )= e 2
+ e 2
× y 2

√2 π √2 π 2

[ ]
−1 2 −1
2 y 1
f ( y )= e 2
× y 2

√2 π 2

[ ]
−1 2 −1
1 y
f ( y )= e 2
×y 2

√2 π
1 −1 2
1 −1 y
f ( y )= y2 e 2

√ π √2
1 −1 2
1 −1 y
f ( y )= 1
y2 e 2

2
2 Γ 1/2

The derived p.d.f is the pd.f of chi square distribution with 1 d.f.

Example 7: If X follow exponential distribution with p.d.f

−λx
f ( x )= λ e 0< x <∞

Determine the p.d.f of y=e−x .

Solution: The p.d.f of X is

 7

− λx
f (x)=λ e
−x
y=e

When x = 0, y = 1, and when x=∞ , then y = ∞ .

x ∈ ( 0 , ∞ )which maps on y ∈ ( 1 , ∞ ).

x=−ln y

|J|= |dxdy|= 1y
f ( y )=f x ( √ y ) ×|J|

− λlog y 1
f ( y )=λ e ×
y

log y−λ 1
f ( y )=λ e
y
− λ−1
f ( y )=λ y
−(λ+1)
f ( y )=λ y

The derived p.d.f is the p.d.f of Pareto distribution.

Example 8: The p.d.f of X is given by

2

f ( x )=2 x e−x 0≤ x≤∞

Determine the p.d.f of y=x 2

Solution: The p.d.f of X is given by

2

f ( x )=2 x e−x 0≤ x≤∞

2
y=x

When x = 0, then y = 0 and when x=∞ , then y=∞

1
2
x= y

| |
−1
dx 1
|J|= = y 2
dy 2
 8

f ( y )=f x ( √ y ) ×|J|
1 −1
1 2 −y 2
f ( y )=2 y e × y
2
−y
f ( y )=e

The resultant p.d.f of of y is follow exponential distribution with λ=1

Transformation Technique for Bivariate Distribution

Let f (x 1 , x 2) be the joint p.d.f of x 1 and x 2 having two dimensional space S1. Now we are
interested to determine the joint p.d.f y 1 , y 2 or marginal p.d.f of y 1 , or y 2. Where y 1=h 1 ( x 1 , x 2 )
and y 2=h 2 ( x 1 , x 2 ). The invers function of of x 1 and x 2 is given by:

x 1=h1 ( y 1 , y 2 ) and x 2=h2 ( y 1 , y 2 ) . The Jacobin of transformation is defined as:

| |
∂ x1 ∂ x1
∂y ∂ y2
|J|= 1
∂ x2 ∂ x2
∂ y1 ∂ y2

The joint p.d.f of y 1 , y 2can be obtained as:

f ( y 1 , y 2 )=f x , x [ h1 ( y1 , y 2 ) ,h 2 ( y 1 , y 2 ) ] ×|J|
1 2

y 1 , y 2 ∈ S2 .

If it is desired to find the marginal p.d.f of y 1 or y 2, then

❑
f ( y 1) =∫ f ( y 1 , y 2 ) d y 2
y2

Or
❑
f ( y 2) =∫ f ( y 1 , y 2 ) d y 1
y1
 9

Example 1: Let x 1 and x 2 are independent random variables follow exponential distribution.
x1
Determine the p.d.f y 1= and y 2=x 1 + x 2.
x 1+ x 2

Solution: The p.d.f of x 1 and x 2 are:

− λx 1
f ( x 1 ) =λ e 0< x 1< ∞

− λ x2
f ( x 2 ) =λ e 0< x 2< ∞

As x 1 and x 2 are independent, the joint p.d.f will be equal to the product of marginal p.d.f’s.

f ( x 1 , x 2 ) =f ( x 1 ) ×f ( x 2 )

− λ x1 −λ x 2
f ( x 1 , x 2 ) =λ e ×λ e
−λ ( x 1+ x2 )
f ( x 1 , x 2 ) =λ2 e

x1 y 2=x 1 + x 2 → 2
y 1= →1
x 1+ x 2 x 2= y 2−x 1
x 1= y 1 ( x1 + x 2 ) Using equation 3
From equation 2: x 2= y 2− y 1 y 2
x 1= y 1 y 2 → 3 ∂ x2
=− y 2
∂ x1 ∂ y1
= y2
∂ y1
∂ x1 ∂ x2
= y1 =1− y 1
∂ y2 ∂ y1

| ||
∂ x1 ∂ x1
∂y
|J|= 1
∂ x2
∂ y2 y
= 2
y1
∂ x2 − y 2 1− y 1 |
= y 2− y 1 y 2 + y 1 y 2= y 2

∂ y1 ∂ y2

x1 y 2=x 1 + x 2
y 1=
x 1+ x 2 When x 1=0 and x 2=0
1
y 1= y 2=0
x2
1+ When x 1=∞ and x 2=∞
x1
y 2=∞
When x 1=0 and x 2=0

y 1=1
 10

When x 1=∞ and x 2=∞

1
y 1= =0
∞
1+
∞

f ( y 1 , y 2 )=f x , x [ h1 ( y1 , y 2 ) ,h 2 ( y 1 , y 2 ) ] ×|J|
1 2

−λ ( y 1 y 2+ y 2− y 1 y 2)
f ( y 1 , y 2 )=λ 2 e × y2
2 − λ y2
f ( y 1 , y 2 )=λ y 2 e

To determine the p.d.f of y 1

❑
f ( y 1) =∫ f ( y 1 , y 2 ) d y 2
y2

∞
f ( y 1) =λ
2
∫ y 2 e−λ y d y 2 2

Let

λ y 2=u

du
dy 2=
λ
∞

∫( )
1 −u
2 u du
f ( y 1) =λ e
0 λ λ
∞
f ( y 1) =∫ u
2−1 −u
e du
0

f ( y 1) =Γ 2=1

In the same way determine the p.d.f y 2

❑
f ( y 2) =∫ f ( y 1 , y 2 ) d y 2
y1

1
2
f ( y 2) =λ y 2 e
−λ y 2
∫ d y1
0
 11

2 −λ y 2
f ( y 2) =λ y 2 e

The p.d.f can be written as:

2
λ 2−1 −λ y
f ( y 2) = ( y2 ) e 2

Γ2

The resultant p.d.f is the p.d.f of gamma distribution with α =2.

Example 2: The p.d.f of x 1 and x 2 is given by:

1
x 1 x2 e (
m−1 n−1 − x + x )
f ( x1 , x2)= 0< x 1 , x 2 < ∞
1 2

Γ mΓn

x1
Determine the p.d.f of y 1= and y 2=x 1 + x 2
x 1+ x 2

Solution:

x1 y 2=x 1 + x 2 → 2
y 1= →1
x 1+ x 2 x 2= y 2−x 1
x 1= y 1 ( x1 + x 2 ) Using equation 3
From equation 2: x 2= y 2− y 1 y 2
x 1= y 1 y 2 → 3 ∂ x2
=− y 2
∂ x1 ∂ y1
= y2
∂ y1
∂ x1 ∂ x2
= y1 =1− y 1
∂ y2 ∂ y1

| ||
∂ x1 ∂ x1
∂y
|J|= 1
∂ x2
∂ y2 y
= 2
y1
∂ x2 − y 2 1− y 1 |
= y 2− y 1 y 2 + y 1 y 2= y 2

∂ y1 ∂ y2

x1 y 2=x 1 + x 2
y 1=
x 1+ x 2 When x 1=0 and x 2=0
1
y 1= y 2=0
x2
1+ When x 1=∞ and x 2=∞
x1
When x 1=0 and x 2=0 y 2=∞

y 1=1
 12

When x 1=∞ and x 2=∞

1
y 1= =0
∞
1+
∞

f ( y 1 , y 2 )=f x , x [ h1 ( y1 , y 2 ) ,h 2 ( y 1 , y 2 ) ] ×|J|
1 2

1
y 1 y 2) ( y 2 − y 1 y 2 ) e (
m−1 n−1 − y − y y 2+ y1 y 2 )
f ( y 1 , y 2 )= ( 2 1
× y2
ΓmΓ n

1 m−1 m−1 n−1 n−1 − y

f ( y 1 , y 2 )= ( y 1 ) ( y 2 ) ( y 2 ) ( 1− y 1 ) e × y 2 2

ΓmΓ n

1 m−1 m+ n−1 n−1 − y

f ( y 1 , y 2 )= ( y1) ( y2) ( 1− y 1) e 2

ΓmΓ n

To determine the pd.f of y 1:

❑
f ( y 1) =∫ f ( y 1 , y 2 ) d y 2
y2

∞
1
y 1 ) ( 1− y 1 ) ∫ ( y 2 )
m−1 n−1 m+n−1 − y
f ( y 1) = ( e d y2 2

Γ mΓn 0

1 m−1 n−1
f ( y 1) = ( y 1 ) ( 1− y 1 ) Γ ( m+n )
Γ mΓn

1
f ( y 1) = ( y 1 )m −1 ( 1− y1 ) n−1
ΓmΓn
Γ ( m+n )

1 m−1 n−1
f ( y 1) = ( y 1 ) ( 1− y 1 )
β ( m , n)

The resultant pdf is the pdf of beta distribution of the first kind.

In the same way determine the p.d.f y 2

❑
f ( y 2) =∫ f ( y 1 , y 2 ) d y 2
y1
 13

1
1
e ∫ ( y 1 ) ( 1− y 1 ) d y 2
m+ n−1 − y m−1 n−1
f ( y 2) = ( y2) 2

Γ mΓn 0

1 m+ n−1 − y Γ m Γ n
f ( y 2) = (y ) e 2

Γ mΓn 2 Γ ( m+n )

1 m +n−1 − y
f ( y 2) = ( y2) e 2

Γ ( m+ n )

The resultant pdf is the pdf of gamma distribution with (m, n).

Example 3: If x 1 and x 2 follow independent standard normal distribution. Determine the p.d.f
x1
y 1= and y 2=x 2. Also determine the pdf of y 1.
x2

Solution: The p.d.f of x 1 and x 2 are:

−1 2
1 x
2 1
f ( x1 )= e −∞ < x 1 <∞
√2 π
−1 2
1 x
2 2
f ( x2 )= e −∞< x 2 <∞
√2 π
As x 1 and x 2 are independent, the joint p.d.f will be equal to the product of marginal p.d.f’s.

f ( x 1 , x 2 ) =f ( x 1 ) ×f ( x 2 )

−1 2 2
1 (x + x )
2 1 2
f ( x1 , x2)= e
2π

x1 y 2=x 2
y 1=
x2
x 1= y 1 y 2 ∂ x2
=0
∂ x1 ∂ y1
= y2 ∂ x2
∂ y1 =1
∂ x1 ∂ y2
= y1
∂ y2
 14

| |
∂ x1 ∂ x1
∂y
|J|= 1
∂ x2
∂ y2 y2
∂ x2
=
0 | y1
1 |
= y2

∂ y1 ∂ y2

x1 y 2=x 2
y 1=
x2
1 x 2=−∞ → y 2=−∞
y 1=
x2
x 2=∞ → y 2 =∞
x1

x 1=−∞ , x2 =−∞
y 1=0
x 1=∞ , x2=∞

y 1=∞

f ( y 1 , y 2 )=f x , x [ h1 ( y1 , y 2 ) ,h 2 ( y 1 , y 2 ) ] ×|J|
1 2

−1
1 2 (( y y2 ) + y 2 )
2 2
1
f ( y 1 , y 2 )= e × y2
2π
2
− y2 2
1 2
( y 1+ 1 )
f ( y 1 , y 2 )= e × y2
2π

To obtain the marginal pdf of y 1

2
∞ − y2 2
1 ( y 1+1 )
f ( y 1) = ∫y e
2 π −∞ 2
2
d y2

Let
2
y2 2
u= ( y 1+ 1 )
2

2 2u
y 2=
( y 21 +1 )
 15

d y2 2
2 y2 =
du ( y12+1 )

d y2 1
=
du ( y 21 +1 ) y 2

du
d y 2=
(y 2
1 +1 ) y 2
∞
1 du
f ( y 1) = ∫ y2 e
−u
2 π −∞ ( y 1 +1 ) y 2
2

∞
2
f ( y 1) = ∫ u 1−1 e−u du
2 π ( y 1 +1 ) 0
2

1
f ( y 1) =
π ( y12+1 )

The resultant pdf follow Cauchy distribution.

Example 4: If x 1 and x 2 follow gamma distributions. Determine the joint pdf of y 1 , y 2 and also
determine the marginal pdf of y 1.

Where

y 1=x 1 + x 2

x1
y 2=
x2

Solution: The pdf of x 1 and x 2 is given by:

α1
λ α −1 − λ x
f ( x1)= x e 1 1

Γ α1 1
α2
λ α −1 − λ x
f ( x2)= x e 2 2

Γ α2 2

Assume x 1 and x 2 are independent

f ( x 1 , x 2 ) =f ( x 1 ) ×f ( x 2 )
 16

α 1+α 2
λ α −1 α −1 − λ(x + x )
f ( x1 , x2)= x x2 e 1 2 1 2

Γ α1 Γ α2 1

y 1=x 1 + x 2 → 1 x1
y 2=
Substitute from eq.2 in eq. 1 x2
y 1=x 2 y 2+ x 2
x 1=x 2 y 2 →2
y 1=x 2 (1+ y 2 ) Substitute value of x 2 in eq. 2
y1
y1 x 1= y
x 2= (1+ y 2 ) 2
(1+ y 2 ) x 1= y 1 y 2
∂ x2 1 ∂ x1
= = y2
∂ y 1 (1+ y 2) ∂ y1
∂ x1
∂ x2 y2 = y1
= ∂ y2
∂ y2 ¿ ¿

| |
∂ x1 ∂ x1
∂y ∂ y2
|J|= 1 =¿
∂ x2 ∂ x2
∂ y1 ∂ y2

y1 y2 y1
|J|= − =
(1+ y 2 ) 1+ y 2 ¿ ¿ ¿
2

f ( y 1 , y 2 )=f x , x [ h1 ( y1 , y 2 ) ,h 2 ( y 1 , y 2 ) ] ×|J|
1 2

[ y 1 y2 y1
] ×|J|
( ) ( )
α +α α 1−1 α 2−1 − λ
λ 1
y1 y2
2
y1 +
(1+ y 2) (1+ y2 )
f ( y 1 , y 2 )= e
Γ α 1 Γ α 2 (1+ y 2) (1+ y 2 )

α +α y1
λ
1 2
α +α −1−1 α −1 1 −λ
(1+ y )
[ y +1 ] y 1 2

f ( y 1 , y 2 )= y1 y2 1 2
e 1
× 2

Γ α1 Γ α2 ( 1+ y 2 )
α +α −1−1 1 ¿¿2

α1 +α 2 y1
λ α +α −1 α −1 1 −λ
(1+ y )
[ y +1 ] 2

f ( y 1 , y 2 )= y1 y2 1 2
α +α
e 1 2

Γ α1 Γ α2 ( 1+ y 2 ) 1 2

Now to obtain the marginal pdf of y 1

 17

α +α ∞ α1−1
λ 1 2
y2
e ∫
α +α −1 − λ y
f ( y 1) = y1 1 2
α +α
dy 21

Γ α1 Γ α2 0 ( 1+ y 2 )
1 2

Let

1
=1+ y 2
u

1 1−u
y 2= −1=
u u

1
u=
1+ y 2

When y 2=0 , u = 1 and when y 2=∞ , u = 0

d y2 −1
= 2
du u

( )
α 1−1
1−u
α 1+ α2 1
λ u du
e ∫
α +α −1 − λ y
f ( y 1) = y1 1 2 1

( 1u )
α 1+α 2 2
Γ α1 Γ α2 0 u

α 1+ α2 1
λ
e ∫u
α +α −1 − λ y α −1 α −1
f ( y 1) = y1 1 2
( 1−u ) du
1 2 1

Γ α1 Γ α2 0

λ
α +α
1 2
α +α −1 − λ y Γ α1 Γ α 2
f ( y 1) = y1 e × 1 2 1

Γ α1 Γ α2 Γ ( α 1 +α 2 )
α1 +α 2
λ α + α −1 −λ y
f ( y 1) = y1 e 1 2 1

Γ ( α1 + α2 )

The derived pdf follow gamma distribution.

Exercise Questions

Q 38: If X 1 , X 2 are independent rectangular variates on [ 0 , 1 ]. Find the distribution of

X1
i.
X2
ii. X1 X2
 18

iii. X1+ X2
iv. X 1 −X 2

Solution: The pdf of rectangular distribution on [ 0 , 1 ] is given by;

f ( x )=1

The pdf of X 1 , X 2 will be given below:

f ( x 1 ) =1

f ( x 2 ) =1

As X 1 , X 2 are independent, then the joint pfd is

f ( x 1 , x 2 ) =f ( x 1 ) ×f ( x 2 )

f ( x 1 , x 2 ) =1

X1
i. Using transformation to find the pdf of
X2
X1
Let u= and v=x 2
X2

X 1 =u X 2

X 1 =uv

∂ X1 ∂ X1 ∂ X2 ∂ X2
=v , =u , =0 , =1
∂u ∂v ∂u ∂v

| ||
∂ X1 ∂ X1
|J|= ∂ u
∂ X2
∂ v = v u =v
∂ X2 0 1 |
∂u ∂v

When X 1 =0 →uv=0→ u=0 , v=0 When X 2 =0 → v=0

When X 1 =1→ uv=1
 19

u=
1 When X 2 =1→ v =1
v
v (0 ,1)
1
u= =∞
0 1
v (0 , )
u

f ( u , v )=f X 1 , X2 (u , v )|J|

f ( u , v )=1× v =v

X1
Now to find the pdf of u= on (0, 1)
X2
❑
f ( u )=∫ f ( u , v ) dv
v

1
1
f ( u )=∫ vdv =
0 2

X1
Now to find the pdf of u= on (0, 1/u)
X2

[ ]
1/ u 2
v u
f ( u )=∫ vdv=
0 2 ⋮
0

1
f ( u )= 2
2u

X1
The distribution of u=
X2

(
1
0 ≤u ≤ 1
2
f ( u )=
1
2
0 ≤u ≤ ∞
2u

ii. Now to find the pdf of X 1 X 2

Let
u=X 1 X 2
v=X 2
 20

u u
X1= =
X2 v

When X 1 =0 ,→ u=0 When X 2 =0 ,→ v=0

u When X 2 =1, → v=1
When X 1 =1, → =1
v u=v
u=v=1

∂ X 1 1 ∂ X 1 −u ∂ X 2 ∂ X2
= , = 2 , =0 , =1
∂u v ∂v v ∂u ∂v

| ||
∂ X1 ∂ X1

|
1 −u
1
|J|= ∂ u ∂v =
v v =
2
∂ X2 ∂ X2 v
0 1
∂u ∂v

f ( u , v )=f X 1 , X2 (u , v )|J|

f ( u , v )=1× v =v

X1
Now to find the pdf of u= on (0, 1)
X2
❑
f ( u )=∫ f ( u , v ) dv
v

u
1
f ( u )=∫ dv
0 v

u
1
f ( u )=∫ dv
0 v

u
f ( u )=log v ⋮
0

f ( u )=log 1−log u=log u

iii. Now to find the pdf of X 1 + X 2

 21

X1
Let u=X 1 + X 2 and v=
X2

X 1 + X 2=u →1
X1
=v → 2
X2
From eq. 2 X 1 =v X 2
u
Substitute in eq. 1: v X 2+ X 2=u → X 2=
1+ v
vu
X1=
1+ v

∂ X1 v ∂ X1 u vu vu +u−vu u
= , = − 2
= 2
=
∂ u 1+ v ∂ v 1+ v ( 1+ v ) ( 1+v ) ( 1+v )2

∂ X2 1 ∂ X2 −u
= , =
∂ u (1+v ) ∂ v ( 1+ v )2

| || ||
∂ X1 ∂ X1 v u
|J|= ∂ u
∂ X2
∂v
∂ X2
=
1+ v
1
( 1+ v )2
−u
=
−vu
3
−
u
( 1+v ) ( 1+v )
3
=
| u
( 1+v )
2

∂u ∂v (1+ v) ( 1+ v ) 2

f ( u , v )=f X 1 , X2 (u , v )|J|

1 u
f ( u , v )=
4 ( 1+ v )2

When X 1 =0∧ X 2=0 and X 1 =1, X 2 =1

X1
v= =0 , v=1
X2
❑
f ( u )=∫ f ( u , v ) dv
v

1 1
u 1 u
f ( u )= ∫ du= [ log (1+ v) ] ⋮
4 0 ( 1+ v )2 4
0
 22

Q 58: If X be a normal variates with mean m and standard deviation σ , then ( X −m

σ )
is gamma

variate with parameter ½ .

Solution: The p.d.f of X is

( )
2
−1 X−m
1 2 σ
f ( x )= e
σ √2 π

X−m
Let =z
σ

When X =−∞ → z=−∞ and when X =∞ ,→ z=∞

−1 2
1 2
z
f ( z )= e
√2 π
−1 2
2 2
z
f ( x )= e
√2 π
2
1
z = y → z= √ 2 y
2
1
z=√ 2 y 2

−1
dz 1
=√ 2× y 2
dY 2

|J|= |dydz|=|2√√2y |
f ( y )=f z ( y )|J |

2 −y √2
f ( y )= e
√2 π 2 √ y
−1
1
f ( y )= y 2
e− y
√π
1
1 −1
f ( y )= y 2 e− y
Γ 1 /2

The pdf of gamma distribution with

 23

α
λ α−1 −λy
f ( y )= y e
(
Γ α )

Which follow gamma distribution with n = ½ .

Q 61: If X and Y are independent gamma variates with parameters u and v respectively. Find the
distribution of the following variables.

i. U = X+Y is a γ ( u+v ) variates.

Solution: The pdf of gamma distribution with λ=1

1 α−1 − y
f ( y )= y e
Γ ( α)

The pdf of gamma distribution with u and v are

1 u−1 −x
f ( x )= x e
Γ ( u)

1 v−1 − y
f ( y )= y e
Γ (v)

The joint pdf of X and Y

1 u−1 v−1 −(x+ y)

f ( x , y )= x y e
Γ ( u) Γ ( v )

x
Given that u=x+ y → 1 and let v= →2
y

From eq. 2 x=vy substitute in eq. 1

u
u=vy + y → y=
1+ v

uv
x=vy=
1+ v

dx v dx u uv u+ uv−uv u
= , = − = =
du 1+ v dv 1+ v ( 1+ v )2 (1+ v ) 2
( 1+ v )2

dy 1 dy −u
= , =
du 1+ v dv ( 1+ v )2
 24

| || ||
dx dx v u
|J|= du
dy
du
dv 1+v
dy
=
1
dv 1+v
( 1+v )2
−u
=
−uv
3
−
u
( 1+ v ) ( 1+ v )
3
=
−u ( v +1 )
( 1+v )
3
|| |
( 1+v )2

u
|J|=
(1+ v )2

f ( u , v )=f X ,Y ( u , v ) ×|J|

( ) ( )
u−1 v−1
1 uv u −u u
f ( u , v )= e ×
Γ (u ) Γ ( v ) 1+ v 1+ v ( 1+ v )2
( u+ v ) −1 u−1
1 u v −u
f ( u , v )= u+ v
e
Γ (u ) Γ ( v ) ( 1+ v )

Now it is desired to find the pdf ofu=x+ y , the range of v is required.

x
We know that v=
y

When x = 0 and y = 0, then v=0

When x=∞ and y=∞, then v=∞

❑
f ( u )=∫ f ( u , v ) dv
v

( u+ v ) −1 ∞ u−1
u v
e ∫
−u
f ( u )= u+v
dv
Γ (u ) Γ ( v ) 0 ( 1+ v )

Let

1 1
1+ v= → v=1−
y y

y−1
v=
y

dv −1 −dy
= → dv= 2
dy y 2 y
 25

1 1
When v=0, then y=1 when v=∞ , then y= = =0
v+1 ∞

( )
u−1
y−1
( u+ v ) −1 1
u y dy
e ∫
−u
( )
f u=
Γ (u ) Γ ( v )
( 1y )
u+ v 2
0 y

1 u−1
u
( u+ v ) −1
y u +v ( y−1 )
e ∫
−u
f ( u )= dy
Γ (u ) Γ ( v ) 0 ( y )u+1
( u+ v ) −1 1
u
e ∫ y ( y −1 ) dy
−u v−1 u−1
f ( u )=
Γ (u ) Γ ( v ) 0

u (u+ v )−1 −u
f ( u )= e β (u , v )
Γ (u ) Γ ( v )
( u+ v ) −1
u −u Γ u Γ v
f ( u )= e
Γ (u ) Γ ( v ) Γ ( u+ v )

u(u +v )−1 −u
f ( u )= e
Γ (u+ v )

Which is also the pdf of gamma variates withu+ v .

Q 61: If X and Y are independent gamma variates with parameters u and v respectively. Find the
distribution of the following variables.

X
ii. V= is a β 1 ( u , v ) variates.
X +Y

Solution: The pdf of X and Y is

1 u−1 v−1 −(x+ y)

f ( x , y )= x y e
Γ ( u) Γ ( v )

x
Given that v= → 1 let u=x+ y → 2
x+ y

Eq. 1 becomes by substituting u=x+ y

x
v= → x=uv
u

Putting in eq. 2
 26

u=uv+ y → y=uv−u

dx dx dy dy
=v , =u , =v−1 , =u
du dv du dv

| ||
dx dx
|J|= du
dy
dv
=
v u
dy v−1 u
=uv −uv+ u=u |
du dv

f ( u , v )=f x , y (u , v)|J |

1
f ( u , v )= ( uv )u−1 ( uv−u )v−1 e−(x+ y) u
Γ (u ) Γ ( v )

( u )u+ v−1
f ( u , v )= ( v )u −1 ( v −1 )v−1 e−u
Γ (u ) Γ ( v )

To find the pdf of u, we needed to find the range of u as:

u=x+ y

When x = 0 and y = 0, then u = 1 and x= y =∞, then u=∞

❑
f ( v )=∫ f (u , v ) du
u

u−1 v−1 ∞
(v ) ( v−1 )
f ( v )=
Γ (u ) Γ ( v )
∫ ( u )u+ v−1 e−u du
0

( v )u−1 ( v−1 )v−1

f ( v )= Γ ( u+v )
Γ (u ) Γ ( v )

1
f ( v )= ( v )u−1 ( v−1 )v−1
Γ (u ) Γ ( v )
Γ ( u+ v )

1
f ( v )= ( v )u−1 ( v−1 ) v−1
β (u , v )

Which the pdf of beta 1 distribution.

Q 61: If X and Y are independent gamma variates with parameters u and v respectively. Find the
distribution of the following variables.
 27

X
iii. w= is β 2 (u , v )
y

x
Given that w= → 1 and let u=x+ y → 2
y

From eq. 1 x=w y substitute in eq. 2

u
u=w y + y → y =
1+ w

uw
x=w y=
1+ w

dx w dx u uw u+uw−uw u
= , = − = =
du 1+ w dw 1+ w ( 1+ w ) 2
( 1+ w ) 2
( 1+ w )2

dy 1 dy −u
= , =
du 1+ w dw ( 1+ w )2

| || ||
dx dx w u
|J|= du
dy
du
dv 1+w
dy
=
1
dv 1+w
( 1+ w )2
−u
=
−uw
3
−
u
( 1+w ) ( 1+w )
3
=
||
−u ( w+1 )
( 1+w )
3
|
( 1+ w )2

u
|J|=
(1+ w )2

f ( w , u )=f X , Y ( w ,u ) ×|J|

( ) ( )
u−1 v−1
1 uw u −u u
f ( w , u )= e ×
Γ ( u ) Γ ( w ) 1+ w 1+ w ( 1+ w )2
( u+w ) −1 u−1
1 u w −u
f ( w , u )= u+ v
e
Γ ( u ) Γ ( w ) ( 1+w )

Now it is desired to find the pdf ofw=x / y , the range of u is required.

u=x+ y
u−1 ∞
1 w
f ( w )= ∫ uu +w−1 e−u du
Γ ( u ) Γ ( w ) (1+ w )u +v 0
 28

u−1
1 w
f ( w )= Γ (w+u)
( ) ( )
Γ u Γ w (1+ w )u +v
u−1
1 w
f ( w )=
Γ ( u ) Γ ( w ) ( 1+ w )u +v
Γ ( w+u)
u−1
1 w
f ( w )=
β 2 (w . u) ( 1+ w )u +v

Q 61: If X and Y are independent gamma variates with parameters u and v respectively. Find the
distribution of the following variables.

x
iv. u=x+ y and v= are independent.
x+ y

Solution:

x=v ( x + y )=vu

y=u−x=u−vu

dx dx dy dy
=v , =u , =1−v , =−u
du dv du dv

| ||
dx dx
|J|= du
dy
dv
=
v u
dy 1−v −u
=|uv−uv−u|=u |
du dv

The pdf of x and y

1 u−1 v−1 −(x+ y)

f ( x , y )= x y e
Γ ( u) Γ ( v )

The joint pdf of u and v

f ( u , v )=f X ,Y ( u , v ) ×|J|

1
f ( u , v )= ( uv )u−1 ( u−vu )v−1 e−(uv +u−vu) ×u
Γ (u ) Γ ( v )
u−1 v−1
( ) u u u −1 v−1 −u
f u, v = v ( 1−v ) e ×u
Γ (u ) Γ ( v )
 29

u+ v−1
u u −1 v−1 −u
f ( u , v )= v ( 1−v ) e
( ) (
Γ u Γ v )

The range of u & v

x 1
v= =
u=x+ y x+ y y
1+
x

When x = 0, y = 0, then u =0 and v =1.

When x=∞ , y=∞, then u=∞ and v=0

❑
f ( u )=∫ f ( u , v ) dv
v

u+ v−1 1
u
e ∫ v (1−v ) dv
−u u−1 v−1
f ( u )=
Γ (u ) Γ ( v ) 0

u+ v−1
u −u
f ( u )= e β (u , v )
Γ (u ) Γ ( v )
u+ v−1
u −u Γ u Γ v
f ( u )= e
Γ (u ) Γ ( v ) Γ (u+v )
u+ v−1
u −u
f ( u )= e
Γ (u+ v)
❑
f ( v )=∫ f (u , v ) d u
v

u−1 v−1 ∞
v ( 1−v )
f ( v )=
Γ (u ) Γ ( v )
∫ uu+ v−1 e−u d u
0

u−1
v ( 1−v )v−1
f ( v )=
Γ (u ) Γ ( v )
Γ (u+ v )
u−1 v−1
u u+v−1 −u v ( 1−v )
f (u )× f (v )= e ×
Γ (u+ v) Γ (u ) Γ ( v )
Γ (u+ v )
u+ v−1 −u
u e u−1 v−1
f (u )× f (v )= × v ( 1−v )
( )
Γ u Γ v( )
 30

f ( u ) × f ( v ) =f (u , v)

Hence u and v are independent.

Q 61: If X and Y are independent gamma variates with parameters u and v respectively. Find the
distribution of the following variables.

x
v. u=x+ y and w= are independent.
y
x− y
vi. u=x+ y and z= are independent
x+ y

Q 62: If X and Y are the joint density function

−(x+ y)
e 3 4
f ( x , y )= x y
Γ4Γ5

x ≥ 0 , y ≥0 and 0 elswhere.

x
Find the pdf of u= , also determine E(u) and V(u)
x+ y

Solution: given that

x
u= → 1 ,let v=x+ y → 2
x+ y

From eq. 1: x=u ( x + y )=uv

From eq. 2: y=v−x=v−uv

dx dx dy dy
=v , =u , =−u , =1−u
du dv du dv

| ||
dx dx
|J|= du
dy
dv = v u =|v−uv +uv|=v
dy −u 1−u |
du dv

The joint pdf of x and y is

−(x+ y)
e 3 4
f ( x , y )= x y
Γ4Γ5
 31

The joint pdf of u and v is

f ( u , v )=f X ,Y ( u , v ) ×|J|
−(uv +v−uv)
e 3
f ( u , v )= ( uv ) (v−uv) 4 × v
Γ 4Γ5
−v
e 3 3 4 4
f ( u , v )= u v v (1−u) v
Γ 4Γ5

Now to determine the pdf of u

3 −v 8 4
u e v
f ( u , v )= × (1−u)
Γ4 Γ5
3 4
u (1−u) 9−1 −v
f ( u , v )= v e
Γ 4Γ5

f ( u , v )= ( u 4−1 (1−u)5−1 9−1 −v

Γ4Γ5
(v e ) )
f ( u , v )=β 1 ( 4 , 5 ) γ ( 9 )
∞
u 4−1 (1−u)5−1
f ( u )=
Γ 4Γ5
∫ 9−1 −v
v e dv
0

4−1 5−1
u (1−u)
f ( u )= Γ9
Γ 4Γ5
4−1 5−1
u (1−u)
f ( u )=
Γ 4Γ5
Γ9

1 4 −1 5 −1
f ( u )= u (1−u)
β(Γ 4 ,Γ 5)

Now to determine E(u) and V(u)

1
1
E ( u )= ∫ u 5−1 (1−u)5−1
β ( Γ 4 , Γ 5) 0

1
E ( u )= β (Γ 5,Γ 5)
β ( Γ 4 , Γ 5)
 32

Γ9 Γ 5,Γ 5 8! 4! 4
E ( u )= × = × =
Γ 4, Γ 5 Γ 10 3! 9! 9
1
1
E (u )=
2
∫
β ( Γ 4 , Γ 5) 0
6−1
u (1−u)
5−1

1
E (u )=
2
β ( Γ 6 , Γ 5)
β ( Γ 4 , Γ 5)

Γ9 Γ 6 , Γ 5 8 ! 5 ! 20 2
E (u )=
2
× = × = =
Γ 4, Γ 5 Γ 11 3 ! 10 ! 90 9

Var ( u )=E ( u 2) −( E(u) )

2

()
2
2 4 2
Var ( u )= − =
9 9 81

Q 63: If X and Y be respectively independent β 1 (u , v ) and γ ( λ , u+v ) variates, then find a pdf for
XY and identify its distribution.

Solution: The pdf of β 1 (u , v )

1 u−1 v−1
f ( x )= x ( 1−x )
β (u , v )

The pdf of γ ( λ , u+v )

u +v
( ) λ ( y )u +v−1 e−λy
f y=
Γ (u+ v)

Given that u=xy and let x=v

u
y=
v

dx dx dy 1 dy −u
=0 , =1 , = , = 2
du dv du v dv v

| || |
dx dx
0 1
1
|J|= du dv
=1 −u =
dy dy v
v v
2
du dv

The pdf of X and Y

 33

f ( x , y )=f (x)× f ( y )
u+ v
1 u−1 v−1 λ ( y )u +v−1 e− λy
f ( x , y )= x ( 1−x ) ×
β (u , v ) Γ (u+v )
u +v u−1
1 λ (1−x )v−1 ( y )u +v−1 e− λy
f ( x , y )= × x
β ( u , v ) Γ (u+ v)

f ( u , v )=f xy ( u , v ) ×| j|

u+v−1 −λ u
()
f ( u , v )=
1
×
λu+ v
β ( u , v ) Γ (u+v )
u−1
× ( v ) ( 1−v )
v−1 u

v () e
v
×
1
v

− λ( )
u
1 λu+ v
f ( u , v )= × × (v )
u−1−u−v+1−1
( 1−v )v−1 ( u )u +v−1 e v
β ( u , v ) Γ (u+v )

−λ ( )
u
1 λu+ v −(v+1)
f ( u , v )= × × (v ) ( 1−v )v−1 ( u )u +v−1 e v
β ( u , v ) Γ (u+v )

Now to find the pdf of u = xy.

The range of v is (0, 1)

❑
f ( u )=∫ f ( u , v ) dv
v

− λ( )
1 v−1 u
1 λu +v ( 1−v )
f ( u )= × ( u )u +v−1∫ e
v
dv
β ( u , v ) Γ (u+ v) 0 (v)(v+ 1)

( )
v−1
v−1 1
1 (v) −1
1 λ
u +v
v −λ ( uv )
( )
f u= ×
β ( u , v ) Γ (u+ v)
( u )u +v−1
∫ ( )(v +1) e dv
0 v

Let

1 1 1
−1=z → =1+ z → v= →
v v 1+Z

1 z +1−1 z
1−v=1− = =
z +1 z +1 z +1

dz
dv =
( 1+ z )2
 34

Rang of Z is

1 1
When v = 0, then z= −1=∞when v = 1, then z= −1=0
v v

( )
v−1 v−1
1
( ) vdz
u
u +v ∞ ( z) −λ
1
1 λ z +1
f ( u )= × ( u )u +v−1 ∫ e
z+1

β ( u , v ) Γ (u+ v)
( z +11 )
(v+1)
0

u +v ∞ v+1 v−1
1 λ ( z +1 ) ( z ) dz
f ( u )= ×
β ( u , v ) Γ (u+ v)
(u )
u +v−1
∫ ( z +1 )v−1
e
− λ ( z+1 ) u

( 1+ z )2
0

u +v ∞
1 λ
e ∫ ( z ) e dz
u +v−1 −λu v−1 − λuz
f ( u )= × (u )
β ( u , v ) Γ (u+ v) 0

Let

t dt
λuz=t → z= → dz =
λu λu
∞

( )
u +v v−1
1 λ t dt
e ∫
u +v−1 −λu −t
f ( u )= × (u ) e
β ( u , v ) Γ (u+ v) 0 λu λu

u +v ∞
1 λ u +v−1 −λu 1
v∫
v−1
f ( u )= × (u ) e ( t ) e−t dt
β ( u , v ) Γ (u+ v) ( λu ) 0
u
1 λ ( u )u −1 e− λu Γ v
f ( u )= ×
β ( u , v ) Γ (u+ v)

Γ (u+ v) λ
u
f ( u )= × ( u )u−1 e− λu Γ v
Γ u Γ v Γ (u+ v)
u
λ ( )u−1 −λu
f ( u )= u e
Γu

Hence u=xy is γ ( λ , u )

x
Q 64: If X and Y are respectively independent γ ( λ , u ) and γ ( λ , v ) variates. Show that is a
y
β 1 ( u , v ) variates.

The pdf of X:
 35

u
λ u−1 −λx
f ( x )= x e
Γu

The pdf of Y:
v
λ v−1 −λy
f ( y )= y e
Γv

X and Y are independent

f ( x , y )=f (x)× f ( y )
u v
λ u−1 − λx λ v−1 −λy
f ( x , y )= x e × y e
Γu Γv
u+ v
( ) λ u−1 v−1 − λ(x+ y)
f x, y = x y e
Γ uΓ v

Given that

x
u=
y

Let v = y

x=uy=uv

dx dx dy dy
=v , =u , =0 =1
du dv du dv

| ||
dx dx
|J|= du
dy
dv = v u =v
dy 0 1 |
du dv

f ( u , v )=f xy ( u , v ) ×| j|
u+v
λ ( uv )u−1 v v−1 e− λ(uv+v) × v
f ( u , v )=
ΓuΓ v
u+v
λ u−1 u+ v−1 − λv(v+ 1)
f ( u , v )= u v e
ΓuΓ v
❑
f ( u )=∫ f ( u , v ) dv
v
 36

u+v ∞
λ
u ∫v
u−1 u+ v−1 − λv(u+1)
f ( u )= e dv
ΓuΓ v 0

Let

z
λv ( u+1 )=z → v=
λ ( u+1 )

dz
dv =
λ ( u+ 1 )
∞

( )
u+v u +v−1 −z
λ z dz
u ∫
u−1
f ( u )= e
ΓuΓ v 0 λ ( u+1 ) λ (u+ 1 )

u+v ∞ u +v−1 −z
λ ( z) dz
u ∫ u+v−1
u−1
f ( u )= e
ΓuΓ v 0 λ ( u+1 )u+ v−1 λ ( u+1 )

u−1 ∞
1 u
u+ v ∫
u +v−1 −z
f ( u )= (z ) e dz
Γ u Γ v ( u+1 ) 0

u−1
1 u
f ( u )= Γ (u+ v )
Γ u Γ v ( u+1 )u+ v
u−1
1 u
f ( u )=
Γ u Γ v ( u+1 )u +v
Γ ( u+ v )
u−1
1 u
f ( u )=
β ( u , v ) ( u+1 )u+ v

U follow beta distribution of second kind with (u, v)

1− X
Q 68: If X is a beta variate of the first kind with l , m. Then Y = is a beta variates of second
X
kind with m ,l .

Solution: The beta first kind pdf

1 l−1 m −1
f ( x )= x ( 1−x )
β (l , m )

Given that

1−x 1
y= → 1−x=xy → x+ xy =1→ x=
x 1+ y
 37

Range of y

1
y= −1
x

When x = 0, then y=∞ and when x = 1, then y=0

|J|= |dxdy|= ( 1+1y ) 2

f ( y )=f (x)×|J|

( ) ( )
l−1 m−1
1 1 1 1
f ( y )= 1− ×
(
β l ,m ) 1+ y 1+ y (1+ y )2

( ) ( )
l−1 m−1
1 1 y 1
f ( y )= ×
β ( l ,m ) 1+ y 1+ y ( 1+ y )2
m−1
1 y
f ( y )=
β ( l ,m ) ( 1+ y )l+m

Moment Generating Function Technique

In moment generating function technique of transformation determine the m.g.f of the function
of a random variable and recognize the m.g.f for some known distribution and derive the pdf for
the function of the new random variable. This method is limited application because we only
recognize only a few m.g.fs.

Let X be a random variable with pdf f (x) and let y=h(x) . The mgf of y (if exist) can be defined
as:

M y ( t )=E ( ety )

M y ( t )=E ( e th(x) )

In case of discrete random variable

M y ( t )=∑ e
th ( x )
f (x )

In case of continuous random variable

M y ( t )=∫ eth ( x ) f ( x ) dx
 38

On the basis of m.g.f of Y determine the pdf of Y.

Example 1: The pdf of a continuous random variable is given by:

2

f ( x )=2 x e−x 0< x≪∞

Determine the pdf of y=x 2 using m.g.f technique.

Solution: The m.g.f of y=x 2

M y ( t )=E ( ety )

M y ( t )=E ( e )
2
tx

∞
M y ( t )=∫ e 2 x e
2 2
tx −x
dx
0

∞
M y ( t )=2∫ xe e
2 2
tx −x
dx
0

∞
M y ( t )=2∫ x e
2
−x (1−t )
dx
0

Let x 2 ( 1−t ) =u

2 u
x=
( 1−t )

du
dx=
2 x ( 1−t )
∞
du
M y ( t )=2∫ x e
−u

0 2 x ( 1−t )
∞
1
M y ( t )= ∫
( 1−t ) 0
1−1 −u
u e du

1
M y ( t )=
( 1−t )

λ
Compare the derive m.g.f of y with the m.g.f of gamma distribution which is . In our case
λ−t
λ=1.The pdf of gamma distribution is
 39

α
λ α −1 − λx
f ( x )= x e
Γα

For λ=1

1 α −1 − x
f ( x )= x e
Γα

Thus, the pdf of y=x 2 follow standard gamma distribution.

Example 2: If X is N(0, 1). Determine the pdf of y=x 2 using m.g.f technique.

Solution: The pdf of X is

−1 2
1 2
x
f ( x )= e −∞ < x <∞
√2 π
The m.g.f of y=x 2

M y ( t )=E ( ety )

M y ( t )=E ( e )
2
tx

∞ −1 2
1 x
∫
2
tx 2
M y ( t )= e e dx
√ 2 π −∞
2
∞ −x
1 (1−2t )
M y ( t )= ∫e
√ 2 π −∞
2
dx

Let

2 2 z
z =x ( 1−2t ) → x=
√ 1−2 t
2 zdz=2 x (1−2 t)dx

zdz=x (1−2t )dx

zdz=x (1−2t )dx

zdz
dx=
x (1−2 t)
 40

zdz
dx=
z
(1−2 t)
√ 1−2 t
dz
dx=
√ 1−2 t
∞ −1 2
1 z
M y ( t )=
√ 1−2
∫e
t √ 2 π −∞
2
dz

∞ −1 2
2 z
M y ( t )= ∫e
√ 1−2 t √ 2 π 0
2
dz

Let
2
z
u=
2
2
z =2 u

2 zdz=2 du

du du
dz= → dz=
z √2 u
∞
2 −u du
M y ( t )= ∫ e
√ 1−2 t √ 2 π 0 √ 2u
∞
1 −u du
M y ( t )= ∫ e
√ 1−2 t √ π 0 √ u
∞ 1
1 −1
M y ( t )= ∫
√ 1−2 t √ π 0
−u
u 2 e du

1 1
M y ( t )= Γ
√ 1−2 t √ π 2
−1
1 2
M y ( t )= = (1−2 t )
√ 1−2 t
−n
The m.g.f of chi square distribution is ( 1−2 t ) 2 .if n =1 then it gives us our derived m.g.f.

So y follows chi square distribution with 1 degree of freedom. The pdf of y will be:
 41

1 −y
1 2
−1
2
f ( y )= 1
y e
2
2 Γ
1
2 ()
Example 3: If X follow uniform distribution (0, 1). Find the pdf of y=3 x +1 using m.g.f
technique.

Solution: The pdf and m.g.f of U (a, b) are

1
f ( x )=
b−a
tb ta
e −e
M x ( t )=
b−a

The m.g.f of y=3 x +1

M y ( t )=E ( ety )

M y ( t )=E ( e t (3 x+1) )
1
1
M y ( t )=∫ e
t ( 3 x +1)
dx
0 1−0

t ( 3 x +1) 1
e
M y ( t )= Ι
t (3 x+1)
0
t ( 3+1 ) t ( 0 +1)
e −e
M y ( t )=
t ( 3+ 1 )−t (1)
4t t
e −e
M y ( t )=
t ( 4 )−t(1)

The m.g.f of y also follow uniform distribution with a=1 , b=4. The pdf of y is

1
f ( y )=
4−1

Example 4: If X is N ( μ , σ 2). Find the pdf of y=ax+ b using m.g.f technique.

Solution: The pdf and m.g.f of X are

( )
2
−1 x− μ
1 2 σ
f ( x )= e
σ √2 π
 42

1 2 2
μt + σ t
2
M x ( t )=e

The m.g.f of y=ax+ b

M y ( t )=E ( ety )

M y ( t )=E ( e t (ax+b ) )

( )
2
∞ −1 x−μ
1
M y ( t )= ∫
σ √ 2 π −∞
et (ax+b ) e 2 σ
dx

( )
2
tb ∞ −1 x−μ
e
M y ( t )=
σ √ 2 π −∞
∫e tax
e
2 σ
dx

Let

x−μ
z=
σ

x=μ+ σz → dx=σdz
tb ∞ −1 2
e z
M y ( t )=
σ √ 2 π −∞
∫e ta ( μ+σz )
e 2
σdz

tb+taμ ∞ −1 2
e z
M y ( t )= ∫
σ √ 2 π −∞
taσz
e e 2
σdz

t (aμ+b) ∞ −1 2
e ( z −2 aσtz )
M y ( t )=
√2 π
∫e 2
dz
−∞

t (aμ+b) ∞ −1 2
e ( z −2 aσz+ ( aσt ) 2− ( aσt ) 2)
M y ( t )=
√2 π
∫e 2
dz
−∞

1 2 2 2
t (aμ+b)+ t a σ ∞ −1 2
2 ( z −2 aσz+ ( aσt ) 2)
e
M y ( t )=
√2 π
∫e 2
dz
−∞

1 2 2 2
t (aμ+b)+ t a σ ∞ −1
2
e
2
( z−aσt )
M y ( t )=
√2 π
∫e 2
dz
−∞

Let
 43

u=z−aσt

du=dz
1 2 2 2
t (aμ+b)+ t a σ ∞ −1 2
2
e u
M y ( t )=
√2 π
∫e 2
dz
−∞

1 2 2 2
t (aμ +b)+ t a σ ∞ −1 2
2
e u
M y ( t )=2
√2 π
∫e 2
dz
0

Let
2
u
V=
2

u =2 V →u=√ 2V
2

2 udu=2 dV

dV dV
du= → du=
u √2 V
1 2 2 2
t (aμ +b)+ t a σ ∞
2
e dV
M y ( t )=2
√2 π
∫ e−V √ 2V
0

1 2 2 2
t (aμ+b)+ t a σ ∞
2
e dV
M y ( t )=
√π
∫ e−V √ V
0

1 2 2 2
t (aμ+b)+ t a σ ∞ 1
2
e −1
M y ( t )=
√π
∫V 2 −V
e dV
0

1 2 2 2
t (aμ+b)+ t a σ
2
e 1
M y ( t )= Γ
√π 2

1 2 2 2
t (aμ+ b)+ t a σ
2
M y ( t )=e

The m.g.f of y is also follow normal distribution with meanaμ+ b and variancea 2 σ 2. The pdf of y
is
 44

[ ]
2
−1 y− ( aμ +b )
1 2 aσ
f ( y )= e
aσ √ 2 π

Example 5: If x 1 and x 2 are independent standard normal variates. Determine the pdf of
2 2
y=x 1 + x 2 using m.g.f technique.

Solution: The pdfs of x 1 and x 2 are

−1 2
1 x
2 1
f ( x1 )= e
√2 π
−1 2
1 x
2 2
f ( x2 )= e
√2 π
The m.g.f of y

M y ( t )=E ( ety )

M y ( t )=E ( e )
t (x + x )
2 2
1 2

M y ( t )=E ( e ) E ( et x )
2 2
t x1 2

∞ −1 2 ∞ −1 2
1 x 1 x
∫ ∫
2 2
tx 2 1 tx 2 2
M y ( t )= e e 1
d x1 × e e 2
d x2
√ 2 π −∞ √2 π −∞
∞ −1 2 ∞ −1 2
1 x (1−2t ) x (1−2t )
M y ( t )= ∫e
2 π −∞
2 1
d x1 ×∫ e 2 2
d x2
−∞

∞ −1 2 ∞ −1 2
4 x (1−2 t) x (1−2 t )
M y ( t )= ∫e
2π 0
2 1
d x1 ×∫ e 2 2
d x2
0

∞ −1 2 ∞ −1 2
2 x (1−2 t ) x (1−2 t )
M y ( t )= ∫e
π 0
2 1
d x 1 ×∫ e 2 2
d x2
0

Let

1 2 2 2u
u= x 1 ( 1−2 t ) → x1=
2 1−2t

2du
2 x1 d x1 =
1−2t
 45

du
d x 1=
( 1−2t ) x1

du du
d x 1= → d x 1=
( 1−2t ) √ √1−2 t √2 u
2u
√ 1−2 t
1 2 2 2v
v= x 2 ( 1−2 t ) → x 2=
2 1−2t

2 dv
2 x 2 d x 2=
1−2t

du
d x 2=
( 1−2t ) x2

dv dv
d x 2= → d x 2=
( 1−2t ) √ √1−2t √2 v
2v
√ 1−2 t

∞ ∞
2 du dv
M y ( t )= ∫ e ×∫ e
−u −v
π 0 √1−2t √2 u 0 √ 1−2 t √ 2 v
∞ ∞
2 1 −u du − v dv
M y ( t )= ∫
(1−2 t) π 0
e ×∫ e
√ 2u 0 √2 v
∞ ∞
1 1 −u du −v dv
M y ( t )= ∫
(1−2 t) π 0
e ×∫ e
√u 0 √v
∞ 1 ∞ 1
1 1 −1 −1
M y ( t )= ∫
(1−2 t) π 0
u 2 e du ×∫ v 2 e dv
−u −v

1 1
M y ( t )= ×π
(1−2 t) π

1 −2/ 2
M y ( t )= =(1−2t )
(1−2 t)

The derived m.g.f is the m.g.f of chi squares distribution with n= 2. The pdf of the concerned chi
square is
 46

2 −y
1 2
−1
2
f ( y )= 2
y e
2
2 Γ
1
2 ()
Example 6: If X 1 , X 2 ,⋯ , X n follow Poisson distribution with parameters μ1 , μ 2 , ⋯ , μ n. Find the
pdf y= X 1+ X 2 +⋯+ X n using m.g.f technique.

Solution: The pdf and m.g.f of Poisson distribution are:

μ x
e μ
( )
f x= x=1 , 2 , 3 ,…
x!
t

M x ( t )=e μ (e −1)

The m.g.f of y= X 1+ X 2 +⋯+ X n

M y ( t )=E ( ety )

M y ( t )=E ( e )
t ( X 1+ X 2+⋯+ X n)

M y ( t )=E ( e t X e t X ⋯ et X )
1 2 n

M y ( t )=E ( e t X ) E ( e t X ) ⋯ E ( et X )
1 2 n

n
M y ( t )=∏ E ( et x ) i

i=1

n
M y ( t )=∏ M x (t )
i=1

n
M y ( t )=∏ e μ (e −1)
t
i

i=1

The derived m.g.f follow Poisson distribution with ∑ μ i. The pdf of y is given by

e∑ μ ( ∑ μi )
y
i

f ( y )=
( y )!

Example 7: If X 1 , X 2 ,⋯ , X n follow bernoulli distribution with parameters p. Find the pdf of

y= X 1+ X 2 +⋯+ X n using m.g.f technique.

Solution: The pdf and m.g.f of binomial distribution

x 1− x
f ( x )= p q
 47

t
M x ( t )=q+ p e

The m.g.f of y is

M y ( t )=E ( ety )

M y ( t )=E ( e )
t ( X 1+ X 2+⋯+ X n)

M y ( t )=E ( e t x et x ⋯ e t x )
1 2 n

M y ( t )=E ( e ) E ( e ) ⋯ E ( e )
tx 1 tx tx2 n

n
M y ( t )=E [ etx ]
n
M y ( t )= [ q+ p e t ]

The derived m.g.f follows binomial distribution. The pdf of y is

f ( y )= ( ny) p q y n− y
 48

The distribution Function Technique

The distribution function technique is also called cumulative distribution function technique or
cdf technique.

Let X be a random variable with pdf f (x) and it is desired to find the pdf of y=h(x) . First we
find the distribution function of y as:

F ( y ) =P(Y ≤ y )

Second we obtain the pdf as:

d
f ( y )= F( y)
dy

Example 1: Let X be a continuous random variable with pdf is

f ( x )=2 x 0 ≤ x ≤ 1

Determine the m.g.f of Y =8 x 3 using c.d.f technique.

Solution: The pdf of X is

f ( x )=2 x 0 ≤ x ≤ 1
3
Y =8 x

F ( y ) =P(Y ≤ y )
3
F ( y ) =P(8 x ≤ y )
1
1 3
F ( y ) =P(x ≤ y )
2
 49

1
1 3
y
2
F ( y ) = ∫ f ( x ) dx
0

1
1 3
y
2
F ( y ) =2 ∫ xdx
0

1
1 3
y
[ ]
2
x 2
F ( y ) =2
2 ⋮
0
2
1
F ( y )= y 3
4

d
f ( y )= F( y)
dy
−1
1 2 3
f ( y )= × y
4 3
−1
1 3
( y )= y
6

Example 2: Let X having pdf f ( x )=4 x 3 0< x<1 . Determine the pdf of Y =−2 ln x 4 using cdf
technique.

Solution: The pdf is

f ( x )=4 x 3 0< x<1

The cdf is

F ( y ) =P ( Y ≤ y )

F ( y ) =P ( −2 ln x 4 ≤ y )

F ( y ) =P (−8lnx ≤ y )

( 1
F ( y ) =P lnx ≤− y
8 )
F ( y ) =P ( x ≤ e )
−1
y
8
 50

−1
y
8
e

F ( y ) =4 ∫ x dx
4

−1
y
8

[ ]
e
x4
F ( y ) =4 ⋮
4
0

F ( y ) =−( e )
−1 4
y
8

−1
y
2
F ( y ) =e
−1
1 2
y
f ( y )= e
2

Example 3: If X is a standard normal variable. Find the pdf of Y =x 2 using cdf technique.

Solution: The pdf of X is

−1 2
1 2
x
f ( x )= e −∞ < x <∞
√2 π
The cdf of X is

F ( y ) =P ( Y ≤ y )

F ( y ) =P ( x 2 ≤ y )

F ( y ) =P ( x ≤ ± √ y )

F ( y ) =P (− √ y ≤ x ≤ √ y )
√y
F ( y ) = ∫ f ( x ) dx
−√ y

√y −1 2
1 x
F ( y )= ∫e
√ π −√ y
2
2
dx

√y −1 2
2 x
F ( y )= ∫e
√2π 0
2
dx
 51

Let
2
x =u

2 xdx=du

du du
dx= → dx= 1
2x
2 u2
√y −1
2 u
du
F ( y )= ∫e
√2 π 0
2
1
2
2u
√y 1 −1
1 −1 u
F ( y )= ∫
√2π 0
u2 e 2
du

1 −1
1 2
−1
2
u
f ( y )= 1
u e
2
2 Γ
1
2()
The derived pdf follow chi square with 1 degree of freedom.

Example 4: If X follow exponential distribution. Find the pdf of Y =e x using cdf technique.

Solution: The pdf of X is

−λx
f ( x )= λ e 0< x <∞

The cdf of X is

F ( y ) =P ( Y ≤ y )

F ( y ) =P ( e x ≤ y )

F ( y ) =P ( x ≤lny )
lny
F ( y ) =λ ∫ e
− λx
dx
0

[ ]
−λx lny
e
F ( y ) =λ ⋮
−λ
0

F ( y ) =−[ e− λlny−e0 ]
 52

F ( y ) =1−e− λlny
−λ

F ( y ) =1−e ln y
−λ
F ( y ) =1− y
−(λ+1)
f ( y )=λ y