Georgi Dinolov

AMS 203
Homework #2
11/12/2012

DeGroot and Schervish: 3.5.10, 3.6.10, 3.6.12, 3.7.2, 3.7.8, 3.8.14

3.5.10 Suppose that a point ( X, Y ) is chosen at random from the circle S defined as
follows:
S = {( x, y) : x2 + y2 1}.
a. Determine the joint p.d.f. of X and Y, the marginal p.d.f. of X, and the marginal
p.d.f. of Y.
b. Are X and Y independent?
a. The points ( x, y) are chosen at random, meaning that they are all equally likely, and
that f ( x, y) is uniform on the unit circle. Thus,
f ( x, y) =

1
,

since the radius of S is unity. The marginal distribution for a given Y is found by
integrating along all possible values for x, namely,
p
Z 1 y2
2 1 y2
1
f (y) =
dx =
.

1 y2
Similarly for X,
f (x) =

2 1 x2
dy =
.

Z 1 x 2
1

1 x 2

b. If X and Y are independent, f ( x ) f (y) = f ( x, y). However,

4 p
f ( x ) f (y) = 2 1 x2

1 y2 6 =

1
in general.

3.6.10 In a large collection of coins, the probability of X that a head will be obtained
when a coin is tossed varies from one coin to another, and the distribution of X in the
collection is specified by the following p.d.f.:

6x (1 x ) for 0 < x < 1,
f1 (x) =
0 otherwise
Suppose that a coin is selected at random from the collectin and tossed once, and that a
head is obtained. Determine the conditional p.d.f. of X for this coin.
In this problem we use Bayes Theorem. Let T1 = { h, t} be the event of tossing a coin for
the first time, with the two possible outcomes of head or tail. We seek f ( x | T1 = h). We
have
f ( x | T1 = h) = R 1
0

= R1
0

f ( T1 = h| x ) f ( x )
f ( T1 = h| x ) f ( x ) dx
6x2 (1 x )
6x2 (1 x ) dx

= R1
0

x [6x (1 x )]
x [6x (1 x )] dx

6x2 (1 x )
1/2

= 12x2 (1 x )


3.6.12 Let Y be the rate (calls per hour) at which calls arrive at a switchboard. Let X be
the number of calls during a two-hour period. Suppose that the marginal p.d.f. of Y is
 y
e
if y > 0,
f 2 (y) =
0
otherwise.
and that the conditional p.f. of X given Y = y is
(
(2y) x 2y
if x = 0, 1, . . . ,
x! e
g1 ( x | y ) =
0
otherwise.
a. Find the marginal p.f. of X. (You may use the formula

R
0

yk ey dy = k!)

b. Find the conditional p.d.f. g2 (y|0) of Y given X = 0.

c. Find the conditional p.d.f. g2 (y|1) of Y given X = 1.
d. For what values of y is g2 (y|1) > g2 (y|0)? Does this agree with the intuition that
the more calls you see, the higher you should think the rate is?
a. The marginal p.f. of X is found by integrating the product g1 ( x |y) f 2 (y) over all
admissible values of y,
f (x) =

Z
(2y) x 2y y
e e dy
g1 ( x |y) f 2 (y) dy =
x!

0
x Z

2
(3y) x e3y dy, substituting m = 3y
x
3 x! 0
Z
2x
2x
= x +1
(m) x em dm = x+1 x!
3 x! 0
3 x!
 x
1 2
=
3 3

b. We have g2 (y|0) = g1 (0|y) f 2 (y)/ f (0) so that

g2 ( y | 0 ) =

g1 ( 0 | y ) f 2 ( y )
e2y ey
=
= 3e3y
f (0)
1/3

c. We have g2 (y|1) = g1 (1|y) f 2 (y)/ f (1) so that

g2 ( y | 1 ) =

g1 ( 1 | y ) f 2 ( y )
2ye2y ey
=
= 9ye3y
f (1)
2/9

d. We seek y from the condition

1
3
This agrees with the intuition that that the more calls you see, the higher you should
think the rate is.
9ye3y > 3e3y

y>

3.7.2 Suppose that three random variables X1 , X2 , and X3 have mixed joint distribution
with p.f./p.d.f.:
 1+ x + x
cx1 2 3 (1 x1 )3 x2 x3 if 0 < x1 < 1 and x2 , x3 {0, 1},
f ( x1 , x2 , x3 ) =
0
otherwise.
(Notice that X1 has a continuous distribution and X2 and X3 have discrete distributions.)
Determine
a. the value of the constant c;
b. the marginal joint p.f. of X2 and X3 ;
c. the conditional p.d.f. of X1 given X2 = 1 and X3 = 1.
R
a. We need c such taht f = 1. This integral takes on the form
Z 1
Z
Z
3
f ( x1 , x2 , x3 ) dx1 dx2 dx3 = c
x1 (1 x1 ) dx1 +
0
Z 1

+
x12 (1 x1 )2 dx1 +
0


2
1
1
1
+
+
=c
= c
20 30 20
6

0
Z 1
0

x12 (1 x1 )2 dx1

3
x1 (1 x1 )dx1

c=6
b. The marginal joint p.f. of X2 and X3 is given by
f 2 ( x2 , x3 ) =

Z 1
0

f ( x1 , x2 , x3 )dx1 = 6

Z 1
0

x11+ x2 + x3 (1 x1 )3 x2 x3 dx1

This integral is difficult to compute, so we will rely on the fact that ( X1 , X2 ) have
discrete distributions. The integrals in Part a will be used to compute the probabilities.
Z 1

 
1
3
= 0.3
f 2 (0, 0) = 6
x1 (1 x1 ) dx1 = 6
20
0
Z 1

 
1
2
2
f 2 (0, 1) = 6
= 0.2
x1 (1 x1 ) dx1 = 6
30
0
Z 1

 
1
2
2
f 2 (1, 0) = 6
= 0.2
x1 (1 x1 ) dx1 = 6
30
0
Z 1

 
1
3
f 2 (1, 1) = 6
x1 (1 x1 )dx1 = 6
= 0.3
20
0
Thus,


f 2 ( x2 , x3 ) =

0.2
0.3

if ( x2 , x3 ) {(1, 0), (0, 1)}

if ( x2 , x3 ) {(0, 0), (1, 1)}
4

c. If x (0, 1),
3/2( x13 (1 x1 )1 )
f ( x1 , x2 = 1, x3 = 1)
=
f 2 (1, 1)
3/40


= 20 x13 (1 x1 )1

f 3 ( x1 | x2 = 1, x3 = 1) =

Else, f 3 ( x1 | x2 = 1, x3 = 1) = 0.


3.7.8 Suppose that the p.d.f. of a random variable X is as follows:

 1 n x
for x > 0
n! x e
f (x) =
0
otherwise.
Suppose also that for any given value X = x ( x > 0), the n random variables Y1 , . . . , Yn
are i.i.d. and the conditional p.d.f. g of each of them is as follows:
 1
for 0 < y < x
x
g(y| x ) =
0
otherwise.
Determine
a. the marginal joint p.d.f. of Y1 , . . . , Yn and
b. the conditional pdf of X for any given values of Y1 , . . . , Yn .
For any given value of X, the random variables Y1 , . . . , Yn are iid, each with a pdf g(y| x ).
Therefore, the conditional joind pdf of Y1 , . . . , Yn given that X = x is
 1
for 0 < yi < x, i = 1, . . . , n,
xn
h ( y1 , . . . , y n | x ) = g ( y1 | x ) g ( y n | x ) =
0
otherwise
The joing pdf is positive if and only if each yi > 0 and x is greater than every yi , that is
x > m = max{y1 , . . . , yn }
a. For y1 > 0 (i = 1, . . . , n), the marginal pdf of Y1 , . . . , Yn is
g0 ( y 1 , . . . , y n ) =

f ( x )h(y1 , . . . , yn | x ) dx =

Z
1
1 x
e dx = em
m

n!

n!

b. For yi > 0 (i = 1, . . . , n), the conditional pdf of X given that Yi = yi (i = 1, . . . , n)

is
 ( xm)
f ( x ) h ( y1 , . . . , y n | x )
e
for x > m,
g1 ( x | y 1 , . . . , y n ) =
=
0
otherwise.
g0 ( y 1 , . . . , y n )


3.8.14 Let X have the uniform distribution on the interval [ a, b], and let c > 0. Prove
that cX + d has the uniform distribution on the interval [ca + d, cb + d].
Let Y = cX + d. The inverse transformation is x = (y d)/c. Assume that c > 0. The
derivative of the inverse is 1/c. The pdf of Y is
g(y) = f ([y d]/c)/c = [c(b a)]1 , for a (y d)/c b.
It is easy to see that a (y d)/c b if and only if ca + d y cb + d, so g is the
pdf of the uniform distribution on the interval [ca + d, cb + d]. If c < 0, the distribution
of Y would be uniform on the interval [cb + d, ca + d]. If c = 0, the distribution of Y is
degenerate at the value d, ie P(Y = d) = 1.