4.

Joint and Conditional Probability Distributions

So far we have been concerned with a single random variable, X and functions of X. Frequently, the
intersection of two or more events is of interest to an experimenter. For instance, we may observe the
income and expenditure of a household. Thus, we observe a pair of random variables (X; Y ). If we add
family size, we add a variable Z and observe (X; Y; Z). The observation of income, expenditure and family
size of a household represents the intersection of speci…c income-expenditure-family size measurements. We
would thus, be interested to study multivariate (the intersection of more than two events) and bivariate (the
intersection of two events).

Another important intersection is the one that occurs when sampling from a population. Suppose that
X1 ; X2 ; :::; Xn denote the outcomes on n successive trials in an experiment. The measurements could be the
income of n households or the measurement of n characteristics of a household. A speci…c set of outcomes,
or sample measurements, may be represented in terms of n events (X1 = x1 ; X2 = x2 ; :::; Xn = xn ). To make
inferences about the population from which the sample was drawn, we will need to calculate the probability
of the intersection (x1 ; x2 ; :::; xn ).

4.1 Joint and Marginal Distributions

De…nitions (4.1):

a). Let X and Y be two discrete random variables. Then

fX;Y (x; y) = P (X = x; Y = y)

is called the joint (bivariate) probability distribution for X and Y . The function f (x; y) will be referred
to as the joint probability mass function.

b). The marginal probability distribution of X is de…ned as:

X
fX (x) = P (X = x) = fX;Y (x; y);
y

and the marginal probability distribution of Y is

X
fY (y) = P (Y = y) = fX;Y (x; y)
x

c). Let X and Y be two random variables. The joint (bivariate) distribution function FX;Y (x; y) is given
by
FX;Y (x; y) = P (X x; Y y)

For discrete variables X and Y , F (x; y) has the form

x
X y
X x
X y
X
F (x; y) = f (x; y) = P (X = u; X = v)
u= 1 v= 1 u= 1 v= 1

1
 d). Two random variables will said to be jointly continuous if their joint distribution function F (x; y) is
continuous in both arguments.

e). Let X and Y be two continuous random variables with joint Z x distribution
Z y function F (x; y). If there
exists a nonnegative function f (x; y) such that F (x; y) = f (u; v) dvdu for any real numbers
1 1
x and y, then X and Y are said to be jointly continuous random variables. The function f (x; y) is
called the joint probability density function.
Z 1
f). The marginal Probability density function of X is de…ned as fX (x) = f (x; y) dy, and the marginal
Z 1 1

probability density function of Y is also de…ned as fY (y) = f (x; y) dx.

1

Properties of Joint Distribution Functions F (x; y):

1). FX;Y (x; 1) and FX;Y (4; 1) are the univariate functions, as functions of x and y, respectively.

2). FX;Y ( 1; y) = FX;Y (x; 1) = FX;Y ( 1; 1) = 0.

3). FX;Y (1; 1) = 1

Properties of Joint Probability Function

1). f (x; y) 0; 8x; y

PP
2). f (x; y) = 1
x y

Properties of Joint Density Function

1). f (x; y) 0; 8x; y

Z 1Z 1
2). f (x; y) dxdy = 1
1 1

Example (4.1): Throw two balanced dice. Let

X = the outcome on the 1st die, and
Y = the outcome on the 2nd die.
Then we de…ne
8
< 1
36 ; for x = 1; 2; : : : ; 6; y = 1; 2; : : : ; 6
f (x; y) = P (X = x; Y = y) =
: 0; otherwise
This can be shown in by the following table

X n Y 1 2 3 4 5 6
1 1 1 1 1 1
1 (1; 1) ; 36 (1; 2) ; 36 (1; 3) ; 36 (1; 4) ; 36 (1; 5) ; 36 (1; 6) ; 36
1 1 1 1 1 1
2 (2; 1) ; 36 (2; 2) ; 36 (2; 3) ; 36 (2; 4) ; 36 (2; 5) ; 36 (2; 6) ; 36
1 1 1 1 1 1
3 (3; 1) ; 36 (3; 2) ; 36 (3; 3) ; 36 (3; 4) ; 36 (3; 5) ; 36 (3; 6) ; 36
1 1 1 1 1 1
4 (4; 1) ; 36 (4; 2) ; 36 (4; 3) ; 36 (4; 4) ; 36 (4; 5) ; 36 (4; 6) ; 36
1 1 1 1 1 1
5 (5; 1) ; 36 (5; 2) ; 36 (5; 3) ; 36 (5; 4) ; 36 (5; 5) ; 36 (5; 6) ; 36
1 1 1 1 1 1
6 (6; 1) ; 36 (6; 2) ; 36 (6; 3) ; 36 (6; 4) ; 36 (6; 5) ; 36 (6; 6) ; 36

2
Example (4.2): In the above exercise let
X = the number of 4’s shown up
Y = the number of 1’s shown up

a). Now we would like to derive the joint probability distribution of X and Y .

b). The sample space consists of 36 points.

c). We can observe the number of 4’s and 1’s either 0 or 1 or 2 times in each case.

The random variable X = 0 for the following 25 points:

f(1; 1); (1; 2); (1; 3); (1; 5); (1; 6); (2; 1); (2; 2); (2; 3); (2; 5); (2; 6); (3; 1); (3; 2); (3; 3); (3; 5);

(3; 6); (5; 1); (5; 2)(5; 3); (5; 5); (5; 6); (6; 1); (6; 2); (6; 3); (6; 5); (6; 6)g

X = 1 for the following 10 points

f(1; 4); (2; 4); (3; 4); (4; 1); (4; 2); (4; 3); (4; 5); (4; 6); (5; 4); (6; 4)g

X = 2 is observed in only 1 point f(4; 4)g

Similarly, Y = 0 can be observed in 25 points

f(2; 2); (2; 3); (2; 4); (2; 5); (2; 6); (3; 2); (3; 3); (3; 4); (3; 5); (3; 6); (4; 2); (4; 3); (4; 4); (4; 5);

(4; 6); (5; 2); (5; 3); (5; 4); (5; 5); (5; 6); (6; 2); (6; 3); (6; 4); (6; 5); (6; 6)g ;

Y = 1 for the following 10 points

f(1; 2); (1; 3); (1; 4); (1; 5); (1; 6); (2; 1); (3; 1); (4; 1); (5; 1); (6; 1)g , and

Y = 2 is observed in only 1 point f(1; 1)g

Noting the following intersections between the random variables:

a). The intersection of X = 0 and Y = 0, has the following 16 points

f(2; 2); (2; 3); (2; 5); (2; 6); (3; 2); (3; 3); (3; 5); (3; 6); (5; 2)(5; 3); (5; 5); (5; 6); (6; 2); (6; 3); (6; 5); (6; 6)g

b). The intersection of X = 1 and Y = 0, has the following 8 points

f(2; 4); (3; 4); (4; 2); (4; 3); (4; 5); (4; 6); (5; 4); (6; 4)g

c). The intersection of X = 2 and Y = 0 has the following 1 points f(4; 4)g

d). The intersection of X = 0 and Y = 1, has the following 8 points

f(1; 2); (1; 3); (1; 5); (1; 6); (2; 1); (3; 1); (5; 1); (6; 1)g

3
 e). The intersection of X = 1 and Y = 1, has the following 2 points f(1; 4); (4; 1)g,

f). The intersection of X = 0 and Y = 2, has the following 1 point f(1; 1)g and

g). Note that when X = 2, Y = 1, cannot take values greater than 0 and vice versa, also when X = 1; Y = 2
cannot take values greater than 0 and there is no sample point where X = 2, Y = 2.

These results can be easily presented using the following table:

X nY 0 1 2 fX (x)
16 8 1 25
0 36 36 36 36
8 2 10
1 36 36 0 36
1 1
2 36 0 0 36
25 10 1
fY (y) 36 36 36 1

Example (4.3): A bowl contains 3 black, 2 red and 4 white balls. Two balls are selected at random without
replacement.

Let the r.v X denote number of black balls in the sample of 2 balls selected.

Let the r.v. Y denote number of red balls in the sample of 2 balls selected.
9
Since there are 9 balls in all and we are selecting two balls, there are 2 = 36 ways of doing so, and
3 2
x ways of selecting x black balls out of the 3 black balls, y ways of selecting y red balls out of the 2
red balls, and given that 2 x y is the number of white balls that can be selected, we can select white
4
balls in 2 x y ways.

The joint (bivariate) probability distribution for X and Y is given as

8 3 2 4
< (x)(y)(92 x y) ; x = 0; 1; 2; y = 0; 1; 2; x + y 2
f (x; y) = P (X = x; Y = y) = (2)
:
0 ; otherwise

This is a bivariate hypergeometric distribution.

3 2 4
0 0 2 0 0 6 1
f (0; 0) = 9 = =
2
36 6
3 2 4
1 0 2 1 0 12 1
f (1; 0) = 9 = =
2
36 3
3 2 4
2 0 2 2 0 3 1
f (2; 0) = 9 = =
2
36 12
3 2 4
0 1 2 0 1 8 2
f (0; 1) = 9 = =
2
36 9
3 2 4
0 2 2 0 2 1
f (0; 2) = 9 =
2
36
3 2 4
1 1 2 1 1 6 1
f (1; 1) = 9 = =
2
36 6

4
 Note that f (1; 2) = f (2; 1) = f (2; 2) = 0

This can be represented in tabular form as follows:

X nY 0 1 2 fX (x)
1 2 1 5
0 6 9 36 12
1 1 1
1 3 6 0 2
1 1
2 12 0 0 12
7 7 1
fY (y) 12 18 36 1

From the above examples we note the following properties

a). f (x; y) 0
PP
b). f (x; y) = 1
x y

We also observe that, for the last example

P (X = 0) = fX (0) = 5=12

P (X = 1) = fX (1) = 1=2

P (X = 2) = fX (2) = 1=12

These are the sums of the 3 columns. Thus we have

X fX (x)
5
0 12
1
1 2
1
2 12

This is the marginal probability distribution of X.

Similarly if we sum the three rows we get the marginal probability distribution of Y as follows:

P (Y = 0) = fY (0) = 7=12

P (Y = 1) = fY (1) = 7=18

P (Y = 2) = fY (2) = 1=36

and thus
Y fY (y)
7
0 12
7
1 18
1
2 36

5
4.2 Conditional Distribution and Independence
Conditional distribution: the conditional probability distribution of Y given X = x is de…ned as

fX;Y (x; y)
fY jX (yjx) = if fX (x) 6= 0.
fX (x)

The conditional distribution of X given Y = y is de…ned as

fX;Y (x; y)
fXjY (xjy) = if fY (y) 6= 0
fY (y)

Note that this de…nition holds for both discrete and continuous random variables.

Notice the similarity between the conditional probability and conditional probability distribution de…ned as

P (A \ B) fX;Y (x; y)
P (BjA) = ; and fY jX (yjx) = .
P (A) fX (x)

The notion of independence: when the information that the random variable X takes a particular value x
is irrelevant to the determination of the probability that another random variable Y takes a value y, we say
that Y is independent of X. Formally, two random variables, X and Y , are said to be independent if and
only if any one of the following three conditions hold:

i ). f (x; y) = fX (x) fY (y) ; 8x; y

ii ). fXjY (xjY = y) = fX (x) ; 8x; y

iii ). fY jX (yjX = x) = fY (y) ; 8x; y

4.3 Expectation
a). If X and Y are two discrete random variables with joint probability function f (x; y), then
XX X X
E (X) = xf (x; y) = x f (x; y)
x y x y
X
= xfX (x)
x

Similarly
XX X X
E (Y ) = yf (x; y) = y f (x; y)
x y y x
X
= yfY (y)
y

b). If X and Y are two continuous random variables with joint probability density function f (x; y),
then
Z 1 Z 1 Z 1 Z 1
E (X) = xf (x; y) dxdy = x f (x; y) dy dx
1 1 1 1
Z 1
= xfX (x) dx
1

6
 Similarly,
Z 1 Z 1 Z 1 Z 1
E (Y ) = yf (x; y) dxdy = y f (x; y) dx dy
1 1 1 1
Z 1
= yfY (y) dy
1

c). If g (x; y) is any function of X and Y , then

XX
E [g (x; y)] = g (x; y) f (x; y) ; if X and Y are discrete random variables, or
x y
Z 1 Z 1
= g (x; y) f (x; y) ; if X and Y are continuous random variables
1 1

4.3.1 Covariance and Correlation

An important measure when we have two random variables is that of covariation of X, and Y . It is de…ned
as follows:

cov (X; Y ) = E (X x) Y y ; where x = E (X) and y = E (Y )

XX
= (x x) y y f (x; y)
x y

In the case where the two random variables are continuous, the covariance between the two random variables
can be evaluated as follows:

cov (X; Y ) = E (X x) Y y
Z 1Z 1
= (x x) y y f (x; y) dxdy
1 1

Note the following:

XX
cov (X; Y ) = (x x) y y f (x; y)
x y
XX
= xy yx xy + x y f (x; y)
x y
XX X X X X XX
= xyf (x; y) y x f (x; y) x y f (x; y) + x y f (x; y)
x y x y y x x y
XX X X
= xyf (x; y) y xfX (x) x yfY (y) + x y
x y x y
= E (XY ) x y x y + x y = E (XY ) x y

We de…ne the coe¢ cient of correlation between X and Y , denoted by or x;y , as follows:

cov (X; Y ) cov (X; Y )

x;y =p =
var (X) V ar (Y ) x y

Properties of the coe¢ cient of correlation

a). 1 1.

7
 b). cov (X; Y ) = cov (Y; X), and

c). x;y = y;x

4.3.2 Conditional Expectation

Conditional expectation of Y given X = x is de…ned as

X
E (Y jX = x) = yfY jX (yjX = x) ; if is discrete
y
Z 1
= yfY jX (yjX = x) dy; if is continuous
1

Compare this with the unconditional mean of Y given as

X
E (Y ) = yfY (y) ; if is discrete
y
Z 1
= yfY (y) dy; if is continuous.
1

The conditional expectation of Y given X = x turns out to be a function of x. That is, E (Y jX = x) = m(x),
or is called the regression function of Y on X. It tells us how the mean of Y varies with changes in X.
Example (4.3): In our earlier example of the hypergeometric distribution we had the following results:
3 2 4
x y 2 x y
f (x; y) = P (X = x; Y = y) = 9 ; x = 0; 1; 2; y = 0; 1; 2; x + y 2
2

The results of this exercise were tabulated as

X nY 0 1 2 fX (x)
1 2 1 5
0 6 9 36 12
1 1 1
1 3 6 0 2
1 1
2 12 0 0 12
7 7 1
fY (y) 12 18 36 1

The problem is now to

a). …nd fY jX (yjX = x)

Y fY jX (yjX = 0) fY jX (yjX = 1) fY jX (yjX = 2)

1=6 2 1=3 2 1=12
0 5=12 = 5 1=2 = 3 1=12 =1
2=9 8 1=6 1 0
1 5=12 = 15 1=2 = 3 1=12 =0
1=36 1 0 0
2 5=12 = 15 1=2 =0 1=12 =0

b). …nd fXjY (xjY = y)

X fXjY (xjY = 0) fXjY (xjY = 1) fXjY (xjY = 2)

1=6 2 2=9 4 1=36
0 7=12 = 7 7=18 = 7 1=36 =1
1=3 4 1=6 3 0
1 7=12 = 7 7=18 = 7 1=36 =0
1=12 1 0 0
2 7=12 = 7 7=18 =0 1=36 =0

8
Note:

5 1 1 2
1). E (X) = 0 12 +1 2 +2 12 = 3

7 7 1 4
2). E (Y ) = 0 12 +1 18 +2 36 = 9

5 1 1 5
3). E X 2 = 02 12 + 12 2 + 22 12 = 6

7 7 1 1
4). E Y 2 = 02 12 + 12 18 + 22 36 = 2

2 2 5 2 2 7
5). x = E X2 [E (X)] = 6 3 = 18

2 2 1 4 2 49
6). y =E Y2 [E (Y )] = 2 9 = 162

1 2 5 1 1 1
7). E (XY ) = 0 0 6 +0 1 9 +0 2 12 +1 0 3 +1 1 6 + = 6

1 2 4 7
8). cov (X; Y ) = E (XY ) E (X) E (Y ) = 6 3 9 = 54

Note: The covariance ends up negative because of the restriction given by x + y 2. So that when x
increases, y must go down and thus, the negative relationship between the two random variables.

cov(X;Y ) 7=54 1
p
9). = =p = 7= p1
X;Y x y (7=18) (49=162) 7 7

10). Regression of Y on X is de…ned as m (x) = E (Y jX = x). Thus,

X 2 8 1 2
m (0) = E (Y jX = 0) = yfY jX (yjX = 0) = 0 +1 +2 =
y
5 15 15 3
X 2 1 1
m (1) = E (Y jX = 1) = yfY jX (yjX = 1) = 0 +1 +2 0=
y
3 3 3
X
m (2) = E (Y jX = 2) = yfY jX (yjX = 2) = 0 1+1 0+2 0=0
y

This can be tabulated as

x m (x)
2
0 3
1
1 3

2 0

Which could also be graphed as:

2 1
m (x) = x
3 3

As a result the slope of the regression of Y on X is linear with the slope of 1=3.

4.3.3 Independence and Expectation

Let X and Y be two random variables with probability function f (x; y), then

9
 a) If X and Y are independent, then E (XY ) = E (X) E (Y )

Proof:
XX XX
E (XY ) = xyf (x; y) = xyfX (x) fY (y)
x y x y
X X
= xfX (x) yfY (y) = E (X) E (Y )
x y

b). If X and Y are two independent random variables, then cov (X; Y ) = 0 and x;y =0

(a) Proof:
cov (X; Y ) = E (XY ) E (X) E (Y ) = E (X) E (Y ) E (X) E (Y ) = 0

Though independence of two random variables implies that they have zero correlation, the converse is not
necessarily true. That is, two random variables may have zero correlation; it does not necessarily mean that
they are independent.

Example (4.4): Let the joint probability distribution of two random variables X and Y be given as follows

X Y 1 0 1 fX (x)
1 1
1 0 4 0 4
1 1 1
0 4 0 4 2
1 1
1 0 4 0 4
1 1 1
fY (y) 4 2 4 1

E(X) = 0; E(Y ) = 0 and E(XY ) = 0 therefore cov(X; Y ) = 0.

However, P (X = 1; Y = 1) = f ( 1; 1) = 0 is not equal to

1 1 1
P (X = 1) P (Y = 1) = = :
4 4 16

Note also that

Y fY jX (yjX = 0) X fX (xjY = 0)
1=4 1 1=4 1
1 1=2 = 2 1 1=2 = 4
0 0 1
0 1=2 =0 0 1=2 = 2
1=4 1 1=4 1
1 1=2 = 2 1 1=2 = 4

1 1

The sum of the conditional distribution should add up to unity. Similarly, the sum of the marginal distrib-
ution also adds up to one.

If X and Y are independent fY jX (yjX = x) = fY (y) for all X, but in the above example this does not hold,
therefore, the two variables are not independent.

Sums of random variables

10
 i ). Let X and Y be two random variables and de…ne a new random variable Z as Z = X + Y

Proposition: var(Z) = var(X) + var(Y ) + 2vov(X; Y )

Proof:
h i h i
2 2
var(Z) = E (Z E (Z)) = E (X + Y E (X + Y ))
h i
2
= E ((X E (X)) + (Y E (Y )))
h i
2 2
= E (X E (X)) + (Y E (Y )) + 2 (X E (X)) (Y E (Y ))
h i h i
2 2
= E (X E (X)) + E (Y E (Y )) + 2E [(X E (X)) (Y E (Y ))]

= var(X) + var(Y ) + 2cov(X; Y )

ii ). Let X and Y be two random variables and de…ne a new random variable Z as Z = X + Y where
and are constants.
2 2
Proposition: var(Z) = var(X) + var(Y ) + 2 cov(X; Y )

Proof:
h i h i
2 2
var(Z) = E (Z E (Z)) = E ( X + Y E ( X + Y ))
h i
2
= E ( (X E (X)) + (Y E (Y )))
h i
2 2
= E 2 (X E (X)) + 2 (Y E (Y )) + 2 (X E (X)) (Y E (Y ))
h i h i
2 2
= 2
E (X E (X)) + 2 E (Y E (Y )) + 2 E [(X E (X)) (Y E (Y ))]
2 2
= var(X) + var(Y ) + 2 cov(X; Y )

Note if = 1 and = 1, then var (X + Y ) = var(X) + var(Y ) + 2cov(X; Y ); = 1 and = 1

implies that var(X Y ) = var(X) + var(Y ) 2cov(X; Y )

iii ). Generally, let X1 ; X2 ; : : : :; Xk be k random variables and de…ne a new random variable Z as Z =
1 X1 + 2 X2 + + k Xk , then let
2 3 2 3
1 X1
6 7 6 7
6 7 6 7
6 2 7 6 X2 7
=6 7
6 .. 7 ; X=6 7
6 .. 7
6 . 7 6 . 7
4 5 4 5
k Xk

The problem is to obtain var(Z).

0
Let Z = 1 X1 + 2 X2 + + k Xk = X, then
h i
2
var (Z) = var ( 0 X) = E ( 0 X 0
E (X))
h i
0
= E ( 0X 0
E (X)) (( 0 X 0
E (X)))

11
 0 0
= E (X E (X)) (X E (X))
0 0
= E (X E (X)) (X E (X))
0
= cov (X)

0
Note that and are 1 k and k 1 vectors, respectively, and cov (X) is a matrix. Namely,
2 3
11 12 1k
6 7
6 7
6 21 22 2k 7
cov (X) = 6
6 .. .. .. ..
7 ; variance covariance matrix of the vector X
7
6 . . . . 7
4 5
k1 k2 kk

Note that 11 = var(X1 ), 12 = cov(X1 ; X2 ), in general ij = cov(Xi ; Xj ). Consequently, var(Z) is a

scalar. Or, speci…cally,

2 2 2
var (Z) = 1 11 + 2 22 + + k kk +2 1 2 12 +2 1 3 13 + +2 1 k 1k

+2 2 3 23 + +2 2 k 2k + +2 k 1 k k 1;k

Note:

a). If Xi and Xj (i 6= j) are not correlated

2 2 2
var (Z) = 1 11 + 2 22 + + k kk

0
= diag 11 22 kk

2
b). If var(Xi ) = 8i, and if Xi and Xj (i 6= j) are not correlated

2 2 2 2 2 0
var (Z) = 1 + 2 + + k =

Exercise (4.1): Given an experiment of rolling a fair, four-sided die twice, let X denote the outcome of
the …rst die, and Y be the sum of the two rolls.

a) Tabulate f (x; y), and

2 2
b). Find E(X), E(Y ), x, y, cov(X; Y ), x;y , and E(Y jX) for X = 1; 2; 3; 4.

Theorem (4.1): Let X1 ; X2 ; : : : :; Xn be n random variable with joint probability density function f (x1 ; x2 ; : : : ::; xn )
and Yi = i (x1 ; x2 ; : : : ::; xn ) for i = 1; 2; : : : :; n. Let Xi = i (y1 ; y2 ; : : : ::; yn ) for i = 1; 2; : : : :; n be a one-
to-one inverse transformation, and let J be the Jacobian of the inverse transformation given by:
0 1
@ 1 @ 1 @ 1
B @y1 @y2 @y2 C
B @ 2 @ 2 @ 2 C
B @y1 @y2 @yn C
J =B
B .. .. .. ..
C;
C
B . . . . C
@ A
@ n @ n @ n
@y1 @y2 @yn

12
then the density function for Y1 ; Y2 ; : : : ::; Yn is given by:

h (y1 ; y2 ; : : : ::; yn ) = f ( 1 (y1 ; y2 ; : : : ::; yn ) ; 2 (y1 ; y2 ; : : : ::; yn ) ; : : : ::; n (y1 ; y2 ; : : : ::; yn )) kJk

where kJk is the absolute value of the determinant of J (Jacobian of transformation).

Example (4.5): Let X1 and X2 be two random variables with joint probability density function given by:
8
< e (x1 +x2 ) ; x > 0; x > 0
1 2
f (x1 ; x2 ) =
: 0 ; elsewhere
X1
Let Y1 = X1 + X2 ; and Y2 = X1 +X2 ; i.e.,
X1
1 (x1 ; x2 ) = X1 + X2 ; and 1 (x1 ; x2 ) =
X1 + X2

a). Find the joint probability density function for Y1 and Y2 .

b). Find the marginal probability density function Y2 .

Solution :

a). Joint probability density function for Y1 and Y2 is found as follows:

h (y1 ; y2 ) = f ( 1 (y1 ; y2 ) ; 2 (y1 ; y2 )) kJk

However,
X1 X1
Y2 = = ) X1 = Y1 Y2
X1 + X2 Y1
Y1 = X1 + X2 ) X2 = Y1 X1 = Y1 Y1 Y2

) X1 = 1 (y1 ; y2 ) = Y1 Y2

) X2 = 2 (y1 ; y2 ) = Y1 Y1 Y2
@ 1 @ 1
y2 y1
) jJj = @y1 @y2
= = y1
@ 2 @ 2
@y1 @y2 1 y2 y1

h (y1 ; y2 ) = f( 1 (y1 ; y2 ) ; 2 (y1 ; y2 )) kJk

(y1 y2 +y1 y1 y2 )
= e y = y 1 e y1
8 1
< y e y1 ; y > 0; 0 < y < 1
1 1 2
) h (y1 ; y2 ) =
: 0 ; Otherwise

b). The marginal probability density function Y2 is found as follows:

Z 1 Z 1
g (y2 ) = h (y1 ; y2 ) dy1 = y1 e y1 dy1 = 1
0 0
8
< 1; 0 < y < 1
2
) g (y2 ) =
: 0; elsewhere

13