1

ISyE 6759 – Fall 2004

Homework #2 Solutions

1. Suppose that the random variable X has p.d.f.

(
20
(x3/2 − x3 ) if 0 < x < 1
fX (x) = 3
0 otherwise

and that the conditional p.d.f. of Y given X = x is

( x √
4(x3/2 −x3 )
if 4x2 < y < 4 x
f (y|x) = .
0 otherwise

√ y2
√
y
Helpful Fact: 4x2 < y < 4 x if and only if 16
<x< 2
.

(a) Find E[Y |X = x].

Solution:
Z
E[Y |x] = yf (y|x) dy
<
Z 4√x
yx
= dy
4x2 4(x3/2− x3 )
2(x2 − x5 )
= . 2
x3/2 − x3

(b) Use (a) to find E[Y ].

Solution: By double expectation, we have

E[Y ] = E[E[Y |X]]

Z
= E[Y |x]fX (x) dx
<
Z 1
2(x2 − x5 ) 20 3/2
= (x − x3 ) dx (by part (a))
0 x3/2 − x3 3
40 Z 1 2
= (x − x5 ) dx = 20/9. 2
3 0

(c) Find the joint p.d.f. of X and Y .

 2

Solution:

f (x, y) = f (y|x)fX (x)

x 20 3/2
= (x − x3 )
4(x3/2 − x3 ) 3
5x √
= , for 0 < x < 1 and 4x2 < y < 4 x. 2
3

(d) Use (c) to find the p.d.f. of Y , i.e., fY (y).

Solution: Using the Helpful Fact from the statement of the problem, we have
Z
fY (y) = f (x, y) dx
<
Z √y/2
5x
= dx
3
y 2 /16
à !
5 y4
= y− , 0 < y < 4. 2
24 64

(e) Use (d) to find E[Y ].

Solution: Using the usual definition, we have

Z
E[Y ] = yfY (y) dy
<
Z 4 Ã !
5 y5
= y2 −
0 24 64
= 20/9. 2

2. Moment Generating Functions.

Find the m.g.f. of Z = X 2 .

(a) Suppose that X ∼ Nor(0, 1). q
R ∞ −ax2
Helpful Fact: −∞ e dx = π/a.

Solution:

MZ (t) = E[etZ ]
2
= E[etX ]
Z
2 1 2
= etx √ e−x /2 dx
< 2π
 3

· µ ¶ ¸
1 Z 1
= √ exp − − t x2 dx (if t < 1/2)
2π < 2
s
1 π
= √ 1 (by the Helpful Fact)
2π 2 − t
1
= √ , if t < 1/2. 2
1 − 2t
Pn
(b) Suppose that X1 , X2 , . . . , Xn ∼ Nor(0, 1). Find the m.g.f. of Y = i=1 Xi2 .

Solution: By (a) and the fact that Y is the sum of i.i.d. Xi ’s, we have

MY (t) = E[etY ] = (E[etX1 ])n = (1 − 2t)−n/2 . 2

(c) Find E[Y ].

Solution:
¯
d ¯
E[Y ] = MY (t)¯¯
dt t=0
¯
d ¯
−n/2 ¯
= (1 − 2t) ¯
dt ¯ t=0
n ¯
= n(1 − 2t)− 2 −1 ¯¯
t=0
= n. 2

(d) Find Var(Y ).

Solution: First of all, by part (c), we have

¯
2 d2 ¯
E[Y ] = 2
MY (t)¯¯
dt t=0
¯
d n ¯
= n(1 − 2t)− 2 −1 ¯¯
dt t=0 ¯
−n
¯
= (n + 2)n(1 − 2t) 2 −2 ¯¯
t=0
= (n + 2)n.

This implies Var(Y ) = E[Y 2 ] − (E[Y ])2 = (n + 2)n − n2 = 2n. 2

 4

(e) Find E[Y k ]. (Don’t simplify the resulting polynomial.)

Solution: Using a similar argument to that employed in part (d) above, we

find that E[Y 3 ] = (n + 4)(n + 2)n, and we notice a pattern. In fact,
E[Y k ] = (n + 2(k − 1)) · · · (n + 4)(n + 2)n. 2

3. Suppose that the number of accidents per week at an industrial plant is Pois(12).
Further, suppose that the numbers of workers injured in each accident are i.i.d.
Bin(10, 0.1) random variables. Assuming that the number of workers injured in
each accident is independent of the number of accidents that occur, find the ex-
pected value and variance of the number of injuries during a given week.

Solution: Let N ∼ Pois(12) denote the number of accidents, and suppose Xi ∼

Bin(10, 0.1) is the number of workers injured in accident i, i = 1, 2, . . . , N . Denote
P
the total number of injuries by T ≡ N i=1 Xi . Then, under the conditions of the
problem, we have
E[T ] = E[N ]E[X1 ] = 12 · 10(0.1) = 12. 2
Further,
Var(T ) = E[N ]Var(X1 ) + (E[X1 ])2 Var(N )
= 12 · 10(0.1)(0.9) + 12 · 12
= 22.8. 2

4. Every time Shaquille O’Neal shoots the basketball, the probability that he scores
is p. Suppose that the number of shots he takes in a particular game is Pois(λ)
distributed. Find the probability that he doesn’t score a basket during a particular
P
game. Hint: Use a conditioning argument and the fact that ∞ n a
n=0 (a /n!) = e .

Solution: Let N ∼ Pois(λ) denote the number of shots that Shaq takes, and
suppose
(
1 if he scores on shot i
Xi ∼ Bern(p) = , i = 1, 2, . . . , N.
0 if he misses on shot i
PN
Denote the total number of baskets by T ≡ i=1 Xi . Using a standard conditioning
argument, we have
∞
X
Pr(T = 0) = Pr(T = 0|N = n) Pr(N = n)
n=0
 5

∞
ÃN ¯ !
X X ¯
= Pr ¯
Xi = 0 ¯ N = n Pr(N = n)
n=0 i=1
∞
à n ¯ !
X X ¯
= Pr Xi = 0 ¯¯ N = n Pr(N = n)
n=0 i=1
∞
à n !
X X
= Pr Xi = 0 Pr(N = n)
n=0 i=1
(since N and the Xi ’s are independent)
∞
X
= Pr(Bin(n, p) = 0) Pr(N = n)
n=0
∞
à !
Xn 0 e−λ λn
= p (1 − p)n
n=0
0 n!
∞
X [λ(1 − p)] n
= e−λ
n=0 n!
= e−λ eλ(1−p)
= e−λp . 2

5. Suppose X1 , X2 , . . . , Xn are i.i.d. Pois(λ).

(a) Find the m.g.f. of Xi .

Solution: By the Law of the Unconscious Statistician,

MX (t) = E[etX ]
∞
X e−λ λn
= etn
n=0 n!
∞
X (λet )n
= e−λ
n=0 n!
t
= e−λ eλe
t
= eλ(e −1) . 2
Pn
(b) Use this m.g.f. to find the distribution of i=1 Xi .
Pn
Solution: Let Yn = i=1 Xi . By (a) and the fact that the Xi ’s are i.i.d., we
have
t
MYn (t) = [MX (t)]n = enλ(e −1) .
 6

This is the m.g.f. of a Pois(nλ) distribution. So by the uniqueness of m.g.f.’s

(at least in this class), we have Yn ∼ Pois(nλ). 2

(c) Suppose
³ λ =P1. Use the Central
´ Limit Theorem to find the approximate value
100
of Pr 95 ≤ i=1 Xi ≤ 105 .

Solution:
à !
95 − nλ Yn − nλ 105 − nλ
Pr(95 ≤ Y100 ≤ 105) = Pr √ ≤ √ ≤ √
nλ nλ nλ
µ ¶
95 − 100 Y100 − 100 105 − 100
= Pr ≤ ≤
10 10 10
≈ Pr (−0.5 ≤ Nor(0, 1) ≤ 0.5) (by the CLT)
= 2Φ(0.5) − 1 = 2(0.6915) − 1 = 0.383. 2

6. A prisoner is trapped in a cell containing three doors. The first door leads to a
tunnel that returns him to his cell after two days of travel. The second leads to
a tunnel that returns him to his cell after three days of travel. The third leads
immediately to freedom.
(a) Assuming that the prisoner will always select doors 1, 2, and 3 with prob-
abilities 0.5, 0.3, 0.2, what is the expected number of days until he reaches
freedom?

Solution: Let X denote the door chosen, and N the days spent in jail. Using
the usual conditioning argument, we have
3
X
E[N ] = E[N |X = k] Pr(X = k)
k=1
= (2 + E[N ])(0.5) + (3 + E[N ])(0.3) + (0)(0.2)
. = 1.9 + 0.8E[N ]
This implies that E[N ] = 9.5. 2

(b) Assuming that the prisoner is always equally likely to choose among those
doors that he has not used, what is the expected number of days until he
reaches freedom? (In this version, for instance, if the prisoner initially tries
door 1, then when he returns to the cell, he will now select only from doors 2
and 3.)
 7

Solution: Now let Ni denote the number of additional days in jail after
initially choosing door i. Then
1 1 1 5 1
E[N ] = (2 + E[N1 ]) + (3 + E[N2 ]) + (0) = + (E[N1 ] + E[N2 ]).
3 3 3 3 3
Meanwhile,
1 1 3
E[N1 ] = (3) + (0) =
2 2 2
and
1 1
E[N2 ] = (3) + (0) = 1.
2 2
These facts imply that E[N ] = 5/2. 2