Probability Theory

4.8 Transformations of Variables

Frequently in Statistics, one encounters the need to derive the probability distribution of a function of one

or more random variables.

Theorem#1: Suppose that X is a continuous random variable with probability distribution f (x). Let

Y = u(X) define a one-to-one correspondence between the values of X and Y so that the equation y = u(x)

can be uniquely solved for x in terms of y, say x = w(y). Then the probability distribution of Y is

g (y) = f [w (y)] |J| ,

where J = w′ (y) and is called the Jacobian of the transformation.

x
Question#1: Let X be a continuous random variable with probability distribution f (x) = 12 , 1 < x < 5.

Find the probability distribution of the random variable Y = 2X − 3.

Solution: The inverse solution of y = 2x − 3 yields x = (y + 3)/2, from which we obtain J = w′ (y) =

dx/dy = 1/2. Therefore, using Theorem 1, we find the density function of Y to be

(y + 3) /2 1 (y + 3)
g (y) = = , −1 < y < 7.
12 2 48

Question#2: If X has the exponential distribution given by f (x) = e−x , x > 0, find the probability
√
density of the random variable Y = X.
√
Solution: The inverse solution of y = x yields x = y 2 , from which we obtain J = dx/dy = 2y. Therefore,

using Theorem 1, we find the density function of Y to be

2 2
g (y) = e−y |2y| = 2ye−y , y > 0.

Question#3: Let X have the probability distribution f (x) = 1, 0 < x < 1. Show that the random variable

Y = −2 ln X has a chi-squared distribution with 2 degrees of freedom. Or If X has a uniform density with

α = 0 and β = 1, show that the random variable Y = −2 ln X has a gamma distribution. What are its

parameters?

Solution: The inverse solution of y = −2 ln x yields x = e−y/2 , from which we obtain J = dx/dy = − 21 e−y/2 .

Therefore, using Theorem 1, we find the density function of Y to be

1 1
g (y) = 1. − e−y/2 = e−y/2 , y > 0.
2 2

by Dr. Masood Anwar 102

 Probability Theory

Question#4: If, D, the diameter of a circle is selected at random from the density function

 
d
f (d) = k 1 − , 0 < d < 5,
5

(a) find the value of k so that f (d) is a probability density; (b) find the density function of the areas of the

circles so selected.

Solution: (a) Find k such that fD (d) is a valid PDF, it must satisfy:

Z ∞
fD (d) dd = 1
−∞

Given the support 0 < d < 5:

Z 5  
d
k 1− dd = 1
0 5
5 5
d2
Z    
d 25
k 1− dd = k d − =k 5− = k (5 − 2.5) = 2.5k
0 5 10 0 10

Solve for k
1 2
2.5k = 1 =⇒ k = =
2.5 5

(b) Find the PDF of the Areas A of the Circles

π 2
A= D
4

Thus, the diameter as a function of area is: r

4A
D=
π
25π
Since 0 < d < 5, the area a ranges from 0 < a < 4 .
q
π 2 4a
Given a = 4d , the inverse transformation is d = π .

The PDF of A is: r !

4a dd
fA (a) = fD ·
π da
r  −1/2
4a dd 1 4a 4 1
d= =⇒ = · =√
π da 2 π π πa

by Dr. Masood Anwar 103

 Probability Theory

q
4a
Now, substitute d = π into fD (d):

q 
√ 

r ! 4a 
4a 2 π
= 2 1− 2 a
fD = 1− √
π 5 5 5 5 π

 √ 
2 2 a 1 2 4
fA (a) = 1− √ ·√ = √ −
5 5 π πa 5 πa 25π

The PDF of the area A is:

2 4 25π
fA (a) = √ − , 0<a<
5 πa 25π 4

Question#5: If X = ln Y has a normal distribution with the mean µ and the standard deviation σ, find

the probability density of Y which is said to have the log-normal distribution.

Solution: Given: X = ln Y is normally distributed: X ∼ N (µ, σ 2 ). The PDF of X is:

(x − µ)2
 
1
fX (x) = √ exp −
σ 2π 2σ 2

We have to find the PDF of Y , denoted as fY (y).

The inverse relationship is x = ln y, so:

dx 1
=
dy y

The PDF of Y is given by:

dx
fY (y) = fX (ln y) ·
dy

(ln y − µ)2
 
1 1
fY (y) = √ exp − ·
σ 2π 2σ 2 y

Combining the terms, the PDF of Y is:

(ln y − µ)2
 
1
fY (y) = exp −√ , y > 0.
yσ 2π 2σ 2

Question#6: According to the Maxwell–Boltzmann law of theoretical physics, the probability density of

V , the velocity of a gas molecule, is

2
f (v) = kv 2 e−βv , v > 0,

where β depends on its mass and the absolute temperature and k is an appropriate constant. Show that
1 2
the kinetic energy U = 2 mV , where m the mass of the molecule is a random variable having a gamma

distribution.

by Dr. Masood Anwar 104

 Probability Theory

Solution: First, we’ll determine the normalization constant k. The PDF must satisfy:

Z ∞
2
k v 2 e−βv dv = 1.
0

dt
Let t = βv 2 , then dv = √
2 βt
, we get:

Z ∞ Z ∞  
t −t du k k 3
k e · √ = t1/2 e−t dt = Γ =1
0 β 2 βt 2β 3/2 0 2β 3/2 2

2β 3/2
=⇒ k =
.
Γ 32

Thus, the PDF of V is:

2β 3/2 2 −βv2
fV (v) =
v e .
Γ 32
q
The inverse solution of u = 21 mv 2 . yields v = 2u
m, from which we obtain J = dv
du = √1 .
2mu

Using the change of variables formula for PDFs:

r !
2u dv
fU (u) = fV · .
m du

2β 3/2
 
2u 1
e−β ( m ) · √
2u
fU (u) = , u > 0.
Γ 23


m 2mu

Therefore, the PDF of U is:

 3/2
1 2β 2βu
fU (u) = u1/2 e− m , u > 0.
Γ(3/2) m

3 m
This confirms that U follows a gamma distribution with shape 2 and scale 2β .

by Dr. Masood Anwar 105

 Probability Theory

Theorem#2: Suppose that X1 and X2 are continuous random variables with joint probability distribution

f (x1 , x2 ). Let Y1 = u1 (X1 , X2 ) and Y2 = u2 (X1 , X2 ) define a one-to one transformation between the points

(x1 , x2 ) and (y1 , y2 ) so that the equations y1 = u1 (x1 , x2 ) and y2 = u2 (x1 , x2 ) may be uniquely solved for x1

and x2 in terms of y1 and y2 , say x1 = w1 (yl , y2 ) and x2 = w2 (y1 , y2 ). Then the joint probability distribution

of Y1 and Y2 is

g (y1 , y2 ) = f [w1 (y1 ) , w2 (y2 )] |J| ,

where the Jacobian is the 2 × 2 determinant

∂x1 ∂x1
∂y1 ∂y2
J=
∂x2 ∂x2
∂y1 ∂y2

∂x1
and ∂y1 is simply the derivative of x1 = w1 (yl , y2 ) with respect to y1 with y2 held constant, referred to in

calculus as the partial derivative of x1 with respect to y1 . The other partial derivatives are defined in a

similar manner.

Question#:7 Let X1 and X2 be two continuous random variables with joint probability distribution

f (x1 , x2 ) = 4x1 x2 , 0 < x1 < 1, 0 < x2 < 1. Find the joint probability distribution of Y1 = X12 and

Y2 = X1 X2 .
√ √
Solution: The inverse solutions of y1 = x21 and y2 = x1 x2 are x1 = y1 and x2 = y2 / y1 , from which we

obtained
√

1/ 2 y1 0 1
J= = .
3/2 √ 2y1
−y2 /2y1 1/ y1

To determine the set B of points in the y1 y2 plane into which the set A of points in the x1 x2 plane is mapped,
√ √
we write x1 = y1 and x2 = y2 / y1 .

Then setting x1 = 0, x2 = 0, x1 = 1, and x2 = 1, the boundaries of set A are transformed to y1 = 0, y2 = 0,

√
y1 = 1, and y2 = y1 . The two regions are illustrated in Figure 4.2. Clearly, the transformation is one-to-one,
 √
mapping the set A = {(x1 , x2 ) |0 < x1 < 1, 0 < x2 < 1} into the set B = (y1 , y2 ) |0 < y1 < 1, 0 < y2 < y1 .

From Theorem 2 the joint probability distribution of Y1 and Y2 is

√ y2 1 2y2 √
g (y1 , y2 ) = 4 ( y1 ) √ = , 0 < y1 < 1, 0 < y2 < y1 .
y1 2y1 y1

Check that the density integrates to 1 over its support.

by Dr. Masood Anwar 106

 Probability Theory

Figure 4.2: Mapping set A into set B.

Question#8: If the joint density of X1 and X2 is given by f (x1 , x2 ) = 1, 0 < x1 < 1, 0 < x2 < 1. Find (a)

the joint probability distribution of Y1 = X1 + X2 and Y2 = X2 , (b) the marginal density of Y1 .

Solution: (a) The inverse solutions of y1 = x1 + x2 and y2 = x2 are x1 = y1 − y2 and x2 = y2 , from which

we obtained
1 −1
J= = 1.
0 1

Support of (Y1 , Y2 ):

From the original support 0 < x1 < 1 and 0 < x2 < 1, we derive:

1. Since x2 = y2 , 0 < y2 < 1.

2. From x1 = y1 − y2 and 0 < x1 < 1, we get: 0 < y1 − y2 < 1 =⇒ y2 < y1 < y2 + 1.

Additionally, since y1 = x1 + x2 and 0 < x1 , x2 < 1, the range for y1 is: 0 < y1 < 2.

Since y1 cannot exceed 2, we split into two cases:

- For 0 < y1 ≤ 1: 0 < y2 < y1 .

- For 1 < y1 < 2: y1 − 1 < y2 < 1.

Thus, the joint density is: 





1, 0 < y1 ≤ 1, 0 < y2 < y1 ,


g(y1 , y2 ) = 1, 1 < y1 < 2, y1 − 1 < y2 < 1,





0, otherwise.


and g(y1 , y2 ) is Uniform over two triangular regions.

by Dr. Masood Anwar 107

 Probability Theory

Figure 4.3:

(b) We find h(y1 ) by integrating g(y1 , y2 ) over y2 :

1. For 0 < y1 ≤ 1:
Z y1
h(y1 ) = 1 dy2 = y1 .
0

2. For 1 < y1 < 2:

Z 1
h(y1 ) = 1 dy2 = 1 − (y1 − 1) = 2 − y1 .
y1 −1

The marginal density of Y1 is: 





 y1 , 0 < y1 ≤ 1,


h(y1 ) = 2 − y1 , 1 < y1 < 2,





0,
 otherwise.

The marginal density h(y1 ) is triangular, peaking at y1 = 1.

Question#9: If the joint density of X1 and X2 is given by

f (x1 , x2 ) = e−(x1 +x2 ) , x1 > 0, x2 > 0.

X1
Find (a) the joint probability distribution of Y1 = X1 + X2 and Y2 = X1 +X2 , (b) the marginal density of Y2 .
x1
Solution: (a) The inverse solutions of y1 = x1 + x2 and y2 = x1 +x2 are x1 = y1 y2 and x2 = y1 (1 − y2 ),

by Dr. Masood Anwar 108

 Probability Theory

from which we obtained

y2 y1
J= = −y1
1 − y2 −y1

Since the transformation is one-to-one, mapping the region x1 > 0 and x2 > 0 in the x1 x2 -plane into the

region y1 > 0 and 0 < y2 < 1 in the y1 y2 -plane, we can use Theorem 2 and it follows that

g (y1 , y2 ) = e−y1 |−y1 | = y1 e−y1

for y1 > 0 and 0 < y2 < 1.

(b) Using the joint density obtained in part (a) and integrating out y1 , we get

Z ∞
h (y2 ) = y1 e−y1 dy1 = Γ2 = 1
0

for 0 < y2 < 1.

Question#10: Let X1 be the amount of premium gasoline (in 1,000 gallons) that a service station has in

its tanks at the beginning of a day, and X2 the amount that the service station sells during that day. If the

joint density of X1 and X2 is given by


1

200 , 0 < x2 < x1 < 20


f (x1 , x2 ) =

0,
 elsewhere

(a) Find the probability density of the amount that the service station has left in its tanks at the end of the

day. (b) Find the Expected amount that the service station has left in its tanks at the end of the day.

Solution: The amount left in the tanks at the end of the day (Y = X1 − X2 ).

We define the new random variables: Y = X1 − X2 , W = X2 .

Solve for x1 and x2 in terms of y and w: x1 = y + w, x2 = w.

The Jacobian matrix J of the transformation is:

   
∂x1 ∂x1
 ∂y ∂w  1 1
J = = .
∂x2 ∂x2
∂y ∂w 0 1

|J| = (1)(1) − (1)(0) = 1.

From the original support 0 < x2 < x1 < 20, we derive:

by Dr. Masood Anwar 109

 Probability Theory

1. Since x1 = y + w and x2 = w, the inequalities become:

0 < w < y + w < 20 =⇒ 0 < w < 20 − y, 0 < y < 20.

Additionally, since x2 = w > 0 and x1 = y + w > w, we have y > 0.

The joint density of (Y, W ) is:

1 1
g(y, w) = f (y + w, w) · |J| = ·1= .
200 200

Thus, the joint density is:


1



200 , 0 < w < 20 − y, 0 < y < 20,
g(y, w) =

0,
 otherwise.

The marginal density of Y , integrate g(y, w) over w:

20−y
20 − y 20 − y
Z
1
h(y) = dw = = , 0 < y < 20.
0 200 200 200

(b) Expected Amount Left in the Tanks (E[Y ])

Using the marginal density h(y):

20 20
20 − y
Z Z
1
E[Y ] = y· dy = (20y − y 2 ) dy.
0 200 200 0

20 20
y3
Z 
2 2 8000 8000 4000
(20y − y ) dy = 10y − = 10(400) − = 4000 − = .
0 3 0 3 3 3

1 4000 4000 20
E[Y ] = · = = ≈ 6.6.
200 3 600 3

20
(thousand gallons)
3

This makes sense: since the initial amount X1 is uniformly distributed up to 20,000 gallons and the sales

X2 are constrained by X1 , the average leftover is about one-third of the maximum capacity.

Question#11: The percentages of copper and iron in a certain kind of ore are, respectively, X1 and X2 .

by Dr. Masood Anwar 110

 Probability Theory

If the joint density of these two random variables is given by


3

11 (5x1 + x2 ), x1 > 0, x2 > 0, x1 + 2x2 < 2


f (x1 , x2 ) =

0,
 elsewhere

(a) Find the probability density of Y1 = X1 + X2 . (b) Find E(Y1 ), the expected total percentage of copper

and iron in the ore.

Solution: Let: Y2 = X2

The inverse solutions of y1 = x1 + x2 and y2 = x2 are x1 = y1 − y2 and x2 = y2 , from which we obtained

Jacobian of the transformation:

∂x1 ∂x1
∂y1 ∂y2 1 −1
J= = =1
∂x2 ∂x2
∂y1 ∂y2 0 1

The original constraints transform to:

1. x1 = y1 − y2 > 0 =⇒ y1 > y2 ,

2. x2 = y2 > 0,

3. x1 + 2x2 = y1 − y2 + 2y2 = y1 + y2 < 2.

Thus, the support of (Y1 , Y2 ) is 0 < y2 < y1 , y1 + y2 < 2

The joint PDF of (Y1 , Y2 ) is:

3 3
fY1 ,Y2 (y1 , y2 ) = fX1 ,X2 (y1 − y2 , y2 ) · |J| = (5(y1 − y2 ) + y2 ) = (5y1 − 4y2 )
11 11

To find fY1 (y1 ), integrate over y2 :

Z
fY1 (y1 ) = fY1 ,Y2 (y1 , y2 ) dy2
y2

Determine limits for y2 :

1. From y2 > 0 and y2 < y1 ,

2. From y1 + y2 < 2 =⇒ y2 < 2 − y1 .

Thus, y2 ranges from 0 to min(y1 , 2 − y1 ).

Case 1: 0 < y1 < 1

Here, y1 < 2 − y1 (since y1 < 1), so y2 ∈ (0, y1 ).

Case 2: 1 ≤ y1 < 2

Here, y1 ≥ 2 − y1 , so y2 ∈ (0, 2 − y1 ).

by Dr. Masood Anwar 111

 Probability Theory

For 0 < y1 < 1:

Z y1
3 3  y1 3 9 2
fY1 (y1 ) = (5y1 − 4y2 ) dy2 = 5y1 y2 − 2y22 0 = (5y12 − 2y12 ) = y
0 11 11 11 11 1

For 1 ≤ y1 < 2:

Z 2−y1
3 3  2−y1 3
5y1 y2 − 2y22 0 5y1 (2 − y1 ) − 2(2 − y1 )2


fY1 (y1 ) = (5y1 − 4y2 ) dy2 = =
0 11 11 11

Simplify:
3 3
10y1 − 5y12 − 8 + 8y1 − 2y12 = −7y12 + 18y1 − 8



=
11 11

Final PDF of Y1 : 
9 2

 y , 0 < y1 < 1
 11 1




fY1 (y1 ) = 3 2
 11 (−7y1 + 18y1 − 8), 1 ≤ y1 < 2




0,
 otherwise

(b)
Z 2
E(Y1 ) = y1 fY1 (y1 ) dy1
0

Split the integral based on the cases for fY1 (y1 ):

1. For 0 < y1 < 1:

1 Z 1  1
9 y14
Z  
9 2 9 9
y1 y1 dy1 = y13 dy1 = =
0 11 11 0 11 4 0 44

2. For 1 ≤ y1 < 2:

Z 2   Z 2
3 2 3
y1 (−7y1 + 18y1 − 8) dy1 = (−7y13 + 18y12 − 8y1 )dy1
1 11 11 1

Z Z Z
7
−7y13 dy1 = − y14 , 18y12 dy1 = 6y13 , −8y1 dy1 = −4y12
4

Evaluate from 1 to 2:

 2    
7 7 1 17
− y14 + 6y13 − 4y12 = (−28 + 48 − 16) − − + 6 − 4 = 4 − − =
4 1 4 4 4

3 17 51
· =
11 4 44
9 51 60 15
E(Y1 ) = + = = ≈ 1.36
44 44 44 11

by Dr. Masood Anwar 112

 Probability Theory

Question# 12. The random variables X and Y, representing the weights of creams and toffees, respectively,

in 1-kilogram boxes of chocolates containing a mixture of creams, toffees, and cordials, have the joint density

function

f (x, y) = 24xy, 0 ≤ x ≤ 1, 0 ≤ y ≤ 1, x + y ≤ 1.

(a) Find the probability density function of the random variable Z = X + Y . (b) Using the density function

of Z, find the probability that, in a given box, the sum of the weights of creams and toffees accounts for at

least 1/2 but less than 3/4 of the total weight.

Solution: (a) We’ll use the transformation method with an auxiliary variable. Let’s choose W = Y . The

inverse solutions of z = x + y and w = y are x = z − w and y = w from which we obtained Jacobian of the

transformation:  
 
∂x ∂x
 ∂z ∂w  1 −1
J = = 
∂y ∂y
∂z ∂w 0 1

The original support is: 0 ≤ x ≤ 1, 0 ≤ y ≤ 1, x+y ≤1

Under the transformation: 0 ≤ z − w ≤ 1 and 0≤w≤1 and z ≤ 1

This simplifies to: 0 ≤ w ≤ z ≤ 1

The Joint Density of (Z, W )

fZ,W (z, w) = fX,Y (z − w, w) · |J| = 24(z − w)w · 1 = 24w(z − w)

The marginal density of Z is obtained after integrating out W over its range 0 ≤ w ≤ z:

Z z Z z
fZ (z) = 24w(z − w) dw = 24 (zw − w2 ) dw
0 0

Compute the integral:

z
zw2 w3 z3 z3 z3
  
24 − = 24 − = 24 · = 4z 3
2 3 0 2 3 6

fZ (z) = 4z 3 , 0≤z≤1

1
(b) The probability that the sum of weights is at least 2 but less than 43 :

3/4 3/4  4  4
z4
  Z 
1 3 3 1 81 1 65
P ≤Z< = 4z 3 dz = 4 = − = − =
2 4 1/2 4 1/2 4 2 256 16 256

by Dr. Masood Anwar 113