Fundamentals of Applied Probability and Random Processes

Instructor’s Manual

2nd Edition

Oliver C. Ibe

1
Preface
This manual has been prepared to accompany the second edition of Fundamentals of Applied
Probability and Random Processes. It contains the solutions to all the problems in the book. The
problem numbers are the same as those in the book. In some cases, alternative methods have
been provided for solving a problem. The instructor can choose the method that appeals the most
to them.

Although a great effort has been made to minimize the number of errors that occur in this
manual, there is no guarantee that it is error-free. Therefore, I would appreciate comments from
instructors who use the manual regarding any errors that they identify.

Oliver C. Ibe
Lowell, Massachusetts
December 2014

2
Contents
Preface
Chapter 1 Basic Probability Concepts
Chapter 2 Random Variables
Chapter 3 Moments of Random Variables
Chapter 4 Special Probability Distributions
Chapter 5 Multiple Random Variables
Chapter 6 Functions of Random Variables
Chapter 7 Transform Methods
Chapter 8 Introduction to Descriptive Statistics
Chapter 9 Introduction to Inferential Statistics
Chapter 10 Introduction to Random Processes
Chapter 11 Linear Systems with Random Inputs
Chapter 12 Special Random Processes

3
Chapter 1: Basic Probability Concepts

Chapter Summary: This chapter deals with basic concepts in probability including sample
space and events, elementary set theory, conditional probability, independent events, basic
combinatorial analysis, and applications of probability in reliability analysis other systems.

Section 1.2: Sample Space and Events

1.1 A fair die is rolled twice. Find the probability of the following events:
(a) The second number is twice the first
(b) The second number is not greater than the first
(c) At least one number is greater than 3.
Solution: Let X denote the outcome of the first roll of the die, and let Y denote the outcome of
the second roll. Then ( x, y ) denotes the event {X = x and Y = y}.

(a) Let U denote the event that the second number is twice the first; that is, y = 2x. Then U
can be represented by

U = {(1, 2), (2, 4), (3, 6)}

Since there are 36 equally likely sample points in the event space, the probability of
event U is given by

3 1
P[U ] = =
36 12

(b) Let V denote the event that the second number is greater than the first. Then V can be
represented by

(1, 2), (1, 3), (1, 4), (1, 5), (1, 6), (2, 3), (2, 4), (2, 5), (2, 6), 
V = 
(3, 4), (3, 5), (3, 6), (4, 5), (4, 6), (5, 6) 

Since there are 15 sample points in the event space, the probability of event V is given
by

15 5
P[V ] = =
36 12

The event that the second number is not greater than the first is V , and its probability is
given by
4
 5 7
P[V ] = 1 − P[V ] = 1 − =
12 12

Note that we can also solve the problem directly as follows:

(1, 1), (2, 1), (2, 2), (3, 1), (3, 2), (3, 3), (4, 1), (4, 2), (4, 3), (4, 4), 
V = 
(5, 1), (5, 2), (5, 3), (5 4), (5, 5), (6, 1), (6, 2), (6, 3), (6, 4), (6, 5), (6, 6)

Since there are 21 sample points in the event space, we obtain

21 7
P[V ] = =
36 12

(c) Let W denote the event that at least one number is greater than 3. If we use “na” to
denote that an entry is not applicable, then W can be represented by

 na na na (1, 4) (1, 5) (1, 6) 

 na na na (2, 4) (2, 5) (2, 6)

 na na na (3, 4) (3, 5) (3, 6) 
W = 
(4, 1) (4, 2) (4, 3) (4, 4) (4, 5) (4, 6)
(5, 1) (5, 2) (5, 3) (5, 4) (5, 5) (5, 6) 
 
(6, 1) (6, 2) (6, 3) (6, 4) (6, 5) (6, 6) 

Since there are 27 sample points in the event space, the probability of W is given by

27 3
P[W ] = =
36 4

1.2 Two distinct dice A and B are rolled. What is the probability of each of the following
events?
(a) At least one 4 appears
(b) Just one 4 appears
(c) The sum of the face values is 7
(d) One of the values is 3 and the sum of the two values is 5
(e) One of the values is 3 or the sum of the two values is 5

Solution: Let (a, b) denote the event {A = a and B = b}.

(a) Let U denote the event that at least one 4 appears. Then we have that

U = {(1, 4), (2, 4), (3, 4), (4, 4), (5, 4), (6, 4), (4, 1), (4, 2), (4, 3), (4, 5), (4, 6)}

Because there are 11 sample points in the event space, the probability of U is given by

5
 11
P[U ] =
36

(b) Let V denote the even that just one 4 appears. Then we have the following event
space:

V = {(1, 4), (2, 4), (3, 4), (5, 4), (6, 4), (4, 1), (4, 2), (4, 3), (4, 5), (4, 6)}

Because there are 10 sample points in the event space, the probability of V is given by

10 5
P[V ] = =
36 18

(c) Let W denote the event that the sum of the face values is 7. Then we have that

W = {(1, 6), (2, 5), (3, 4), (4, 3), (5, 2), (6, 1)}

Because there are 6 sample points in the event space, the probability of W is given by

6 1
P[W ] = =
36 6

(d) Let X denote the event that one of the values is 3 and the sum of the two values is 5.
Then we have that

X = {(2, 3), (3, 2)}

Because there are 2 sample points in the event space, the probability of X is given by

2 1
P[ X ] = =
36 18

(e) Let Y denote the event that one of the values is 3 or the sum of the two values is 5. Let
G denote the event that one of the values is 3, and let H denote the event that the sum
of the two values is 5. Then we have that

G = {(1, 3), (2, 3), (4, 3), (5, 3), (6, 3), (3, 1), (3, 2), (3, 4), (3, 5), (3, 6)}
H = {(1, 4), (2, 3), (3, 2), (4, 1)}

Now, Y is the union of G and H; that is, Y = G ∪ H and

P[Y ] = P[G ∪ H ] = P[G ] + P[ H ] − P[G ∩ H ]

From earlier results we have that P[G ∩ H ] = P[ X ]. Thus, we obtain

6
 10 4 2 12 1
P[Y ] = P[G ] + P[ H ] − P[ X ] = + − = =
36 36 36 36 3

1.3 Consider an experiment that consists of rolling a die twice.

(a) Plot the sample space Ω of the experiment
(b) Identify the event A, which is the event that the sum of the two outcomes is equal to
6.
(c) Identify the event B, which is the event that the difference between the two outcomes
is equal to 2.

Solution: We represent the outcomes of the experiment by the two-dimensional diagram shown
below where the horizontal axis represents the outcome of the first roll and the vertical axis
represents the outcome of the second roll. The events A and B are indicated in the diagram.

B
6 (4, 6)
(1, 6) (2, 6) (3, 6) (5, 6) (6, 6)

5
(1, 5) (2, 5) (3, 5) (4, 5) (5, 5) (6, 5)

4
(1, 4) (2, 4) (3, 4) (4, 4) (5, 4) (6, 4)
Second Roll

3
(1, 3) (2, 3) (3, 3) (4, 3) (5, 3) (6, 3)

2
(1, 2) (2, 2) (3, 2) (4, 2) (5, 2) (6, 2)

1
(1, 1) (2, 1) (3, 1) ( 4, 1) (5, 1) (6, 1)
A
1 2 3 4 5 6
First Roll

1.4 A 4-sided fair die is rolled twice. What is the probability that the outcome of the first roll
is greater than the outcome of the second roll?

Solution: We represent the outcomes of the experiment by the two-dimensional diagram shown
below where the horizontal axis represents the outcome of the first roll and the vertical axis
represents the outcome of the second roll. Let A denote the event that the outcome of the first
roll is greater than the outcome of the second roll.

7
 4
(1, 4) (2, 4) (3, 4) (4, 4)

3
Second Roll

(1, 3) (2, 3) (3, 3) (4, 3)

2 A
(1, 2) (2, 2) (3, 2) (4, 2)

1
(1, 1) (2, 1) (3, 1) ( 4, 1)

1 2 3 4
First Roll

Because all the outcomes are equally likely, the probability that the outcome of the first roll is
greater than the outcome of the second roll is

6 3
P[ A] = =
16 8

1.5 A coin is tossed until the first head appears, and then the experiment is stopped. Define a
sample space for the experiment.

Solution: The experiment can stop after the first toss if the outcome is a head (H). If the first toss
results in a tail (T), we try again and stop if the outcome is a head or continue if the outcome is a
tail, and so on. Thus, the sample space of the experiment is as shown below.

Λ
1.6 A coin is tossed four times and observed to be either a head or a tail each time. Describe
the sample space for the experiment.

Solution: The sample space for the experiment is as shown below:

8
 H
H T
H H
T
T
H H
H T
T H
T
T
H
H
T
H
H
T
T T
H H
T
T H
T
T

1.7 Three friends, Bob, Chuck, and Dan take turns (in that order) throwing a die until the first
“six” appears. The person that throws the first six wins the game, and the game ends.
Write down a sample space for this game.

Solution: Let B denote the event that Bob wins the game, C the event that Chuck wins the game,
and D the event that Dan wins the game. Then B, C , D denote the complements of B, C and D,
respectively. The sample space of the game is as follows:

B C D B C D Λ
Section 1.3: Definitions of Probability

1.8 A small country has a population of 17 million people of whom 8.4 million are male and
8.6 million are female. If 75% of the male population and 63% of the female population
are literate, what percentage of the total population is literate?

Solution: Let M denote the number of literate males in millions and F the number of literate
females in millions. Then we have that

M = 0.75 × 8.4 = 6.3

F = 0.65 × 8.6 = 5.418

Thus, the number of literate people in the population is M + F = 11.718. Therefore, the
percentage p of literate people in the population is
9
 11.718
p= × 100 = 68.93
17

1.9 Let A and B be two independent events with P[ A] = 0.4 and P[ A ∪ B] = 0.7. What is
P[B] ?

Solution: We know that P[ A ∪ B] = P[ A] + P[ B ] − P[ A ∩ B ]. Because A and B are independent

we know that P[ A ∩ B ] = P[ A]P[ B]. Therefore, we have that

P[ A ∪ B] = P[ A] + P[ B ] − P[ A]P[ B ] = P[ A] + P[ B]{1 − P[ A]}

From this we obtain

P[ A ∪ B] − P[ A] 0.7 − 0.4 0.3

P[ B] = = = = 0.5
1 − P[ A] 0.6 0.6

1.10 Consider two events A and B with known probabilities P[ A], P[ B] and P[ A ∩ B ]. Find
the expression for the event that exactly one of the two events occurs in terms of
P[ A], P[ B] and P[ A ∩ B ].

Solution: Recall that P[ A ∪ B] denotes the probability that either event A occurs or event B
occurs or both events occur. Thus, if Z is the event that exactly one of them occurs, then

P[ Z ] = P[ A ∪ B] − P[ A ∩ B] = P[ A] + P[ B] − 2 P[ A ∩ B]

Second Method: Another way to see this result is by noting that

P[ A \ B] = P[ A − B ] = P[ A] − P[ A ∩ B] is the probability that only the portion of A that is
disjoint from B occurs; that is, the probability that points in A that are not in B occur. Similarly,
P[ B \ A] = P[ B − A] = P[ B ] − P[ A ∩ B] is the probability that only the portion of B that is
disjoint from A occurs; that is, the probability that points in B that are not in A occur. Since the
events A \ B and B \ A are disjoint (or mutually exclusive), we have that

P[ Z ] = P[{ A − B} ∪ {B − A}] = P[ A − B] + P[ B − A] = P[ A] + P[ B] − 2 P[ A ∩ B]

Third Method: We observe that Z = ( A ∩ B ) ∪ ( B ∩ A ). Since A ∩ B and B ∩ A are disjoint

events, we obtain

P[ Z ] = P[ A ∩ B ] + P[ B ∩ A ] = P[ A − B] + P[ B − A] = P[ A] + P[ B] − 2 P[ A ∩ B]

Comment: Note that the problem specifically requires the answer in terms of P[ A], P[ B] and
P[ A ∩ B ]. Also, we might be tempted to solve the problem in the following manner:

10
 P[ Z ] = P[ A]{1 − P[ B ]} + P[ B]{1 − P[ A]} = P[ A] + P[ B] − 2 P[ A]P[ B]

However, this result is correct only if A and B are independent events because the implicit
assumption we make here is that P[ A ∩ B ] = P[ A]P[ B ] = P[ A]{1 − P[ B]}, which implies that A
and B are independent, which in turn means that the events A and B are independent.

1.11 Two events A and B have the following probabilities: P[ A] = 1 / 4, P[ B | A] = 1 / 2 and

P[ A | B] = 1 / 3. Compute (a) P[ A ∩ B ], (b) P[B] and (c) P[ A ∪ B ].

Solution: The solution is as follows:

P[ A ∩ B] 1 1 1
P[ B | A] = ⇒ PA ∩ B] = P[ A]P[ B | A] = × = = 0.125
P[ A] 4 2 8
P[ A ∩ B] P[ A ∩ B ] (1 / 8) 3
P[ A | B] = ⇒ P[ B ] = = = = 0.375
P[ B ] P[ A | B ] (1 / 3) 8
1 3 1 1
P[ A ∪ B] = P[ A] + P[ B ] − P[ A ∩ B] = + − = = 0.5
4 8 8 2

1.12 Two events A and B have the following probabilities: P[ A] = 0.6, P[ B ] = 0.7 and
P[ A ∩ B] = p. Find the range of values that p can take.

Solution: We know that P[ A ∪ B] = P[ A] + P[ B ] − P[ A ∩ B ] ≤ 1. Thus,

0.6 + 0.7 − p ≤ 1 ⇒ p ≥ 0.3

Also, P[ A ∩ B] ≤ min{P[ A], P[ B ]} ⇒ p ≤ 0.6. Thus, we obtain

0.3 ≤ p ≤ 0.6

1.13 Two events A and B have the following probabilities: P[ A] = 0.5, P[ B ] = 0.6 and
P[ A ∩ B ] = 0.25. Find the value of P[ A ∩ B ].

Solution: From the De Morgan’s first law we know that

P[ A ∩ B ] = P[ A ∪ B ] = 1 − P[ A ∪ B] = 0.25 ⇒ P[ A ∪ B] = 0.75

But P[ A ∪ B] = P[ A] + P[ B ] − P[ A ∩ B]. Thus,

P[ A ∩ B] = P[ A] + P[ B ] − P[ A ∪ B] = 0.5 + 0.6 − 0.75 = 0.35

11
1.14 Two events A and B have the following probabilities: P[ A] = 0.4, P[ B] = 0.5 and
P[ A ∩ B] = 0.3. Calculate the following:
(a) P[ A ∪ B]
(b) P[ A ∩ B ]
(c) P[ A ∪ B ]

Solution: We have that

(a) P[ A ∪ B] = P[ A] + P[ B] − P[ A ∩ B] = 0.4 + 0.5 − 0.3 = 0.6

(b) P[ A ∩ B ] = P[ A] − P[ A ∩ B] = 0.4 − 0.3 = 0.1
(c) From the De Morgan’s second law, P[ A ∪ B ] = P[ A ∩ B] = 1 − P[ A ∩ B] = 0.7

1.15 Christie is taking a multiple-choice test in which each question has four possible answers.
She knows the answers to 40% of the questions and can narrow the choices down to two
answers 40% of the time. If she knows nothing about the remaining 20% of the questions
and merely guesses the answers to these questions, what is the probability that she will
correctly answer a question chosen at random from the test?

Solution: We use the tree diagram to solve the problem. First, we note that there are two cases to
consider for the problem:

(a) Christie does not answer questions she knows nothing about
(b) She answers all questions, resorting to guesswork on those she knows nothing about.

Under case (a), we assume that after Christie has narrowed down the choices to two answers, she
flips a coin to guess the answer. That is, given that she can narrow the choices down to two
answers, the probability of getting the correct answer is 0.5. Thus, the tree diagram for this case
is as follows:

12
Thus, the probability p that she will correctly answer a question chosen at random from the test is
given by

p = (0.4 × 1.0) + (0.4 × 0.5) = 0.60

Under case (b), she adopts the same strategy above for those questions that she can narrow down
the answer to two questions; that is, the final answer is based on flipping a fair coin. For those
she knows nothing about, she is equally likely to choose any one of the four answers. Thus, the
tree diagram for this case is as follows:

Completely 1.0 Correct

Knows
0.4

0.5 Correct
0.4 Partially Knows
0.5 Wrong

0.2 0.25 Correct

Completely
Doesn’t Know
0.75 Wrong

Thus, the probability p that she will correctly answer a question chosen at random from the test is
given by

p = (0.4 × 1.0) + (0.4 × 0.5) + (0.2 × 0.25) = 0.65

We can also solve the problem using the conditional probability approach. We define the
following probabilities:

• P[C ] is the probability that Christie answers a question correctly

• P[CK ] is the probability that she completely knows the answer
• P[PK ] is the probability that she partially knows the answer
• P[K ] is the probability that she does not know the answer

Then under case (a), we have that

P[C ] = P[C | CK ]P[CK ] + P[C | PK ]P[ PK ] + P[C | K ]P[ K ]

= 1.0 × 0.4 + 0.5 × 0.4 + 0.0 × 0.2 = 0.6

Similarly, under case (b), we have that

13
 P[C ] = P[C | CK ]P[CK ] + P[C | PK ]P[ PK ] + P[C | K ]P[ K ]
= 1.0 × 0.4 + 0.5 × 0.4 + 0.25 × 0.2 = 0.65

1.16 A box contains 9 red balls, 6 white balls, and 5 blue balls. If 3 balls are drawn
successively from the box, determine the following:
(a) The probability that they are drawn in the order red, white, and blue if each ball is
replaced after it has been drawn.
(b) The probability that they are drawn in the order red, white, and blue if each ball is not
replaced after it has been drawn.

Solution: The total number of balls is 20. Let P[R] denote the probability that a red ball is
drawn, P[W ] the probability that a white ball is drawn, and P[B] the probability that a blue ball
is drawn.

(a) If drawing is made with replacement, then the probability of drawing a red ball, a white
ball and a blue ball, in that order is

9 6 5 27
P[ R → W → B] = P[ R]P[W | R]P[ B | R, B ] = P[ R]P[W ]P[ B] =× × = = 0.03375
20 20 20 800
(b) If drawing is made without replacement, then the probability of drawing a red ball, a
white ball and a blue ball, in that order is

9 6 5 3
P[ R → W → B] = P[ R]P[W | R]P[ B | R, B ] = × × = = 0.0395
20 19 18 76

1.17 Let A be the set of positive even integers, let B be the set of positive integers that are
divisible by 3, and let C be the set of positive odd integers. Describe the following events:
(a) E1 = A ∪ B
(b) E 2 = A ∩ B
(c) E3 = A ∩ C
(d) E 4 = ( A ∪ B ) ∩ C
(e) E5 = A ∪ (B ∩ C )

Solution:
(a) E1 = A ∪ B is the set of positive integers that are either even integers or integers that are
divisible by 3 or both
(b) E 2 = A ∩ B is the set of positive integers that are both even integers and integers that are
divisible by 3
(c) E3 = A ∩ C is the set of positive integers that are both even integers and odd integers,
which is a null set since no integer can be both even and odd at the same time
(d) E 4 = ( A ∪ B ) ∩ C is the set of positive integers that are odd integers and are either even
integer or divisible by 3. (This reduces to the set of odd integers that are divisible by 3.)
14
 (e) E5 = A ∪ (B ∩ C ) is the set of positive integers that are either even or are both divisible
by 3 and are odd.

1.18 A box contains 4 red balls labeled R1 , R2 , R3 and R4 ; and 3 white balls labeled W1 , W2
and W3 . A random experiment consists of drawing a ball from the box. State the
outcomes of the following events:
(a) E1 , the event that the number on the ball (i.e., the subscript of the ball) is even
(b) E 2 , the event that the color of the ball is red and its number is greater than 2
(c) E3 , the event that the number on the ball is less than 3
(d) E 4 = E1 ∪ E3
(e) E5 = E1 ∪ (E 2 ∩ E3 )

Solution:
(a) E1 = {R2 , R4 , W2 }
(b) E 2 = {R3 , R4 }
(c) E3 = {R1 , R2 , W1 , W2 }
(d) E 4 = E1 ∪ E3 = {R1 , R2 , R4 , W1 , W2 }
(e) E5 = E1 ∪ (E 2 ∩ E3 ) = E1 = {R2 , R4 , W2 }

1.19 A box contains 50 computer chips of which 8 are known to be bad. A chip is selected at
random and tested.
(a) What is the probability that it is bad?
(b) If a test on the first chip shows that it is bad, what is the probability that a second
chip selected at random will also be bad, assuming the tested chip is not put back into
the box?
(c) If the first chip tests good, what is the probability that a second chip selected at
random will be bad, assuming a tested chip is not put back into the box?

Solution: The number of good chips is 50 − 8 = 42.

(a) The probability that the selected chip is bad is p = 8 / 50 = 0.16
(b) Let X be the event that the first chip is bad and Y the event that the second chip is bad. If
the tested chip is not put back into the box, then there are 49 chips left after the first chip
has been removed and 7 of them are bad. Thus,

7 1
P[Y | X ] = = = 0.1428
49 7

(c) If the first chip tests good and a tested chip is not put back into the box, the probability
that a second chip selected at random is bad is given by

15
 8
P[Y | X ] = = 0.1633
49
Section 1.5: Elementary Set Theory

1.20 A set Ω has four members: A, B, C, and D. Determine all possible subsets of Ω.

Solution: Ω = {A, B, C , D}. Thus, all the possible subsets of Ω are as follows:

The null set Ø along with the following non-empty subsets

{A}, {B}, {C}, {D}, {A, B}, {A, C}, {A, D}, {B, C}, {B, D}, {C , D},
{A, B, C}, {A, B, D}, {A, C , D}, {B, C , D}, {A, B, C , D}
1.21 For three sets A, B, and C, use the Venn diagram to show the areas corresponding to the
sets
(a) ( A ∪ C ) − C
(b) B ∩ A
(c) A ∩ B ∩ C
( )
(d) A ∪ B ∩ C

Solution: Let Ω denote the universal set. We assume that the three sets A, B and C have
intersections as shown below.

B Ω

(a) ( A ∪ C ) − C is the shaded area shown below:

16
 B Ω

(b) B ∩ A is the shaded area shown below:

B Ω

(c) A ∩ B ∩ C is the shaded area shown below:

B Ω

( )
(d) A ∪ B ∩ C = (A ∩ B ) ∩ C is the shaded area shown below:

17
 B Ω

1.22 A universal set is given by Ω = {2, 4, 6, 8, 10, 12, 14}. If we define two sets A = {2, 4, 8}
and B = {4, 6, 8, 12}, determine the following:
(a) A
(b) B − A
(c) A ∪ B
(d) A ∩ B
(e) A ∩ B
(f) ( A ∩ B ) ∪ (A ∩ B )
Solution:
(a) A = Ω − A = {6, 10, 12, 14}
(b) B − A = {6, 12}
(c) A ∪ B = {2, 4, 6, 8, 12}
(d) A ∩ B = {4, 8}

(e) A ∩ B = {6, 12} = B − A

(f) ( A ∩ B ) ∪ (A ∩ B ) = {4, 8} ∪ {6, 12} = {4, 6, 8, 12} = B

1.23 Consider the switching networks shown in Figure 1.18. Let E k denote the event that
switch S k is closed, k = 1, 2, 3, 4. Let E AB denote the event that there is a closed path
between nodes A and B. Express E AB in terms of the E k for each network.

18
 S1 S2 S3 S4

S1
S1 S3
S2

S3
S2 S4
S4

S1 S2 S3

S4

Figure 1.18: Figure for Problem 1.23

Solution:

(a) We are given the figure:

S1 S2 S3 S4

This is a serial system that requires all switches to be closed for the path to exist. Thus,

E AB = E1 ∩ E 2 ∩ E3 ∩ E 4
(b) We are given the figure:

S1 S3

S2 S4

This is a series-parallel system that requires that either all the upper path be closed or
those in the lower path be closed or all the switches in both paths be closed. Thus,

E AB = ( E1 ∩ E 2 ) ∪ ( E 3 ∩ E 4 )
19
 (c) We are given the figure:

S1

S2

S3

S4
This is a system of parallel connections that requires at least one switch to be closed.
Thus,

E AB = E1 ∪ E 2 ∪ E3 ∪ E 4
(d) We are given the figure:

This is another series-parallel system that requires switches S1 , S 3 and either S 2 or S 4

or both to be closed. Thus,
E AB = E1 ∩ ( E 2 ∪ E 4 ) ∩ E3 = ( E1 ∩ E 2 ∩ E3 ) ∪ ( E1 ∩ E 4 ∩ E3 )

1.24 Let A, B and C be three events. Write out the expressions for the following events in
terms of A, B and C using set notation:
(a) A occurs but neither B nor C occurs.
(b) A and B occur, but not C.
(c) A or B occurs, but not C.
(d) Either A occurs and B does not occur, or B occurs and A does not occur.

Solution:
(a) The event “A occurs but neither B nor C occurs” is given by

A ∩ B ∩ C = A ∩ (B ∪ C) = A − (B ∪ C)

(b) The event “A and B occur, but not C” is given by

A ∩ B ∩ C = ( A ∩ B) − C

20
 (c) The event “A or B occurs, but not C” is given by

( A ∪ B) ∩ C = ( A ∪ B) − C

(d) The event “either A occurs and B does not occur, or B occurs and A does not occur” is
given by

( A ∩ B ) ∪ ( A ∩ B)

Section 1.6: Properties of Probability

1.25 Mark and Lisa registered for Physics 101 class. Mark attends class 65% of the time and
Lisa attends class 75% of the time. Their absences are independent. On a given day, what
is the probability that
(a) at least one of them is in class?
(b) exactly one of them is in class?
(c) Mark is in class, given that only one of them is in class?

Solution: Let M denote the event that Mark attends class and L the event that Lisa attends class.
Then M is the event that Mark is absent from class and L is the event that Lisa is absent from
class.
(a) The probability p that at least one of them is in class is the complement of the probability
that neither of them is in class and is given by

p = 1 − P[ M ∩ L ] = 1 − P[ M ]P[ L ] = 1 − (0.35)(0.25) = 0.9125

(b) The probability q that exactly one of them is in class is the probability that either Mark is
in class and Lisa is absent or Mark is absent and Lisa is in class. This is given by

q = P[( M ∩ L ) ∪ ( M ∩ L)] = P[ M ∩ L ] + P[ M ∩ L] = P[ M ]P[ L ] + P[ M ]P[ L]

= (0.65)(0.25) + (0.35)(0.75) = 0.425
The second equality in the first line follows from the fact that the events M ∩ L and
M ∩ L are mutually exclusive, and the third equality follows from the fact that M and L
are independent.

(c) Let X denote the event that exactly one of them is in class. The probability that Mark is in
class, given that only one them is in class, is given by

P[ M ∩ X ] P[ M ∩ L ] (0.65)(0.25)
P[ M | X ] = = = = 0.3823
P[ X ] P[ M ∩ L ] + P[ M ∩ L] (0.65)(0.25) + (0.35)(0.75)

1.26 The probability of rain on a day of the year selected at random is 0.25 in a certain city.
The local weather forecast is correct 60% of the time when the forecast is rain and 80%
21
 of the time for other forecasts. What is the probability that the forecast on a day selected
at random is correct?

Solution: Let R denote the event that it rains and CF the event that the forecast is correct. First,
we use the tree diagram, as follows:

CF
0.6

R
0.25 0.4
CF

0.8 CF
0.75
R
0.2
CF
The probability of a correct forecast is given by

P[CF ] = (0.25)(0.6) + (0.75)(0.8) = 0.75

We can also use the direct conditional probability approach as follows:

P[CF ] = P[CF | R]P[ R] + P[CF | R ]P[ R ] = (0.6)(0.25) + (0.8)(0.75) = 0.75

1.27 53% of the adults in a certain city are female, and 15% of the adults are unemployed
males.
(a) What is the probability that an adult chosen at random in this city is an employed
male?
(b) If the overall unemployment rate in the city is 22%, what is the probability that a
randomly selected adult is an employed female?

Solution: Let M denote the event that an adult is a male, F the event that an adult is a female, E
the event that an adult is employed, and U the event that an adult is unemployed. Then we have
that P[ F ] = 0.53, P[ M ] = 1 − P[ F ] = 0.47.
(a) Let m denote the probability that a male is unemployed. Thus, the probability that an
unemployed adult is a male is 0.47m. From the problem we have that
0.47 m = 0.15 ⇒ m = 0.15 / 0.47 = 0.32. Thus, the probability that a randomly selected
adult is an employed male is 0.47 × (1 − 0.32) = 0.32. This is illustrated by the following
tree diagram where E stands for “employed” and U stands for unemployed:

22
 1− f

1− m

(b) The probability that an adult is unemployed is given by

P[U ] = 0.22 = 0.53 f + 0.47m = 0.53 f + (0.47)(0.32)

0.22 − (0.47)(0.32)
⇒ f = = 0.1313
0.53
Thus, the probability that a randomly selected adult is an employed female is

0.53(1 − f ) = 0.4604

1.28 A survey of 100 companies shows that 75 of them have installed wireless local area
networks (WLANs) on their premises. If three of these companies are chosen at random
without replacement, what is the probability that each of the three has installed WLANs?

75 74 73
Solution: The probability of this event is × × = 0.417.
100 99 98
Section 1.7: Conditional Probability

1.29 A certain manufacturer produces cars at two factories labeled A and B. 10% of the cars
produced at factory A are found to be defective, while 5% of the cars produced at factory
B are defective. If factory A produces 100,000 cars per year and factory B produces
50,000 cars per year, compute the following:
(a) The probability of purchasing a defective car from the manufacturer
(b) If a car purchased from the manufacturer is defective, what is the probability that it
came from factory A?

Solution: Let A denote the event that a randomly selected car is produced in factory A, and B the
event that it is produced in factory B. Let D denote the event that a randomly selected car is
defective. Now, we are given the following:

23
 100,000 2
P[ A] = =
100,000 + 50,000 3
50,000 1
P[ B] = =
100,000 + 50,000 3
P[ D | A] = 0.10
P[ D | B ] = 0.05

(a) The probability of purchasing a defective car from the manufacturer is given by

P[ D] = P[ D | A]P[ A] + P[ D | B]P[ B ] = (0.1)( 23 ) + (0.05)( 13 ) = 0.25

3 = 0.083

(b) Given that a car purchased from the manufacturer is defective, the probability that it
came from factory A is given by

P[ A ∩ D] P[ D | A]P[ A] (0.1)( 23 )
P[ A | D] = = = 0.25 = 0.80
P[ D] P[ D] (3)
1.30 Kevin rolls two dice and tells you that there is at least one 6. What is the probability that
the sum is at least 9?

Solution: Let X denote the event that there is at least one 6, and Y the event that the sum is at
least 9. Then X and Y can be represented as follows:

X = {(1, 6), (2, 6), (3, 6), (4, 6), (5, 6), (6, 6), (6, 5), (6, 4), (6, 3), (6, 2), (6, 1)}
Y = {(3, 6), (4, 5), (4, 6), (5, 4), (5, 5), (5, 6), (6, 3), (6, 4), (6, 5), (6, 6)}

The probability that the sum is at least 9, given that there is at least one 6 is given by

P[Y ∩ X ]
P[Y | X ] =
P[ X ]

Now, P[ X ] = 11 / 36, and the event Y ∩ X can be represented by

Y ∩ X = { (3, 6), (4, 6), (5, 6), (6, 6), (6, 5), (6, 4), (6, 3)}

Thus, P[Y ∩ X ] = 7 / 36, and we have that

(7 / 36) 7
P[Y | X ] = = = 0.6364
(11 / 36) 11

24
1.31 Chuck is a fool with probability 0.6, a thief with probability 0.7, and neither with
probability 0.25.
(a) What is the probability that he is a fool or a thief but not both?
(b) What is the conditional probability that he is a thief, given that he is not a fool?

Solution: Let F denote the event that Chuck is a fool and T the event that he is a thief. Then we
know that P[ F ] = 0.6, P[T ] = 0.7, and P[ F ∩ T ] = 0.25. From the De Morgan’s first law,

P[ F ∩ T ] = P[ F ∪ T ] = 1 − P[ F ∪ T ] = 0.25 ⇒ P[ F ∪ T ] = 0.75

Because P[ F ∪ T ] = P[ F ] + P[T ] − P[ F ∩ T ] = 0.75 = 0.6 + 0.7 − P[ F ∩ T ], we obtain

P[ F ∩ T ] = 0.55
(a) We know that P[ F ∪ T ] is the probability that Chuck is either a fool or a thief or both.
Therefore, the probability that he is either a fool or a thief but not both is given by

P[ F ∪ T ] − P[ F ∩ T ] = 0.20

Note that we can obtain the same result by noting that the probability that he is either a
fool or a thief but not both is the probability of the union of the events that he is a fool
and not a thief, and he is a thief and not a fool. That is, the required result is

P[( F ∩ T ) ∪ ( F ∩ T )] = P[ F ∩ T ] + P[ F ∩ T ]

where the equality is due to the fact that the events are mutually exclusive. But

P[ F ∩ T ] = P[ F ] − P[ F ∩ T ] = 0.05
P[ F ∩ T ] = P[T ] − P[ F ∩ T ] = 0.15

Adding the two probabilities gives the result we obtained earlier.

(b) The probability that he is a thief, given that he is not a fool, is given by

P[T ∩ F ] 0.15 3
P[T | F ] = = = = 0.375
P[ F ] 0.40 8

1.32 Studies indicate that the probability that a married man votes is 0.45, the probability that
a married woman votes is 0.40, and the probability that a married woman votes given that
her husband does is 0.60. Compute the following probabilities:
(a) Both a man and his wife vote.
(b) A man votes given that his wife votes also.

25
Solution: Let M denote the event that a married man votes, and W the event that a married
woman votes. We are given that P[ M ] = 0.45, P[W ] = 0.40 and P[W | M ] = 0.60.

(a) The probability that both a man and his wife vote is P[ M ∩ W ], which is given by

P[ M ∩ W ] = P[W | M ]P[ M ] = (0.60)(0.45) = 0.27

(b) The probability that a man votes, given that his wife votes is given by

P[ M ∩ W ] 0.27
P[ M | W ] = = = 0.675
P[W ] 0.40

1.33 Tom is planning to pick up a friend at the airport. He has figured out that the plane is late
80% of the time when it rains, but only 30% of the time when it does not rain. If the
weather forecast that morning calls for a 40% chance of rain, what is the probability that
the plane will be late?

Solution: Let L denote the event that the plane is late and R the event that the forecast calls for
rain. Then we are given that P[ R ] = 0.4, P[ L | R ] = 0.8, and P[ L | R ] = 0.3. Therefore, the
probability that the plane will be late is

P[ L] = P[ L | R ]P[ R ] + P[ L | R | P[ R ] = (0.8)(0.4) + (0.3)(0.6) = 0.5

1.34 Consider the communication channel shown in Figure 1.19. The symbols transmitted are
0 and 1. However, three possible symbols can be received: 0, 1, and E. Thus, we define
the input symbol set as X = {0, 1} and the output symbol set as Y = {0, 1, E}. The
transition (or conditional) probability P[Y | X ] is the probability that Y is received, given
that X was transmitted. In particular, P[0 | 0] = 0.8, P[1 | 0] = 0.1 and P[ E | 0] = 0.1.
Similarly, P[0 | 1] = 0.2, P[1 | 1] = 0.7 and P[ E | 1] = 0.1. If P[ X = 0] = P[ X = 1] = 0.5,
determine the following:
(a) P[Y = 0], P[Y = 1] and P[Y = E ]
(b) If 0 is received, what is the probability that 0 was transmitted?
(c) If E is received, what is the probability that 1 was transmitted?
(d) If 1 is received, what is the probability that 1 was transmitted?

26
 Figure 1.19: Figure for Problem 1.34

Solution:
(a) The probabilities that 0, 1 and E are received are given, respectively, by

P[Y = 0] = P[Y = 0 | X = 0]P[ X = 0] + P[Y = 0 | X = 1]P[ X = 1]

= (0.8)(0.5) + (0.2)(0.5) = 0.5
P[Y = 1] = P[Y = 1 | X = 0]P[ X = 0] + P[Y = 1 | X = 1}P[ X = 1]
= (0.1)(0.5) + (0.7)(0.5) = 0.4
P[Y = E ] = P[Y = E | X = 0]P[ X = 0] + P[Y = E | X = 1]
= (0.1)(0.5) + (0.1)(0.5) = 0.1

(b) Given that 0 is received, the probability that 0 was transmitted is given by

P[( X = 0) ∩ (Y = 0)] P[Y = 0 | X = 0]P[ X = 0] (0.8)(0.5)

P[ X = 0 | Y = 0] = = = = 0.8
P[Y = 0] P[Y = 0] 0.5

(c) Given that E was received, the probability that 1 was sent is given by

P[( X = 1) ∩ (Y = E )] P[Y = E | X = 1]P[ X = 1] (0.1)(0.5)

P[ X = 1 | Y = E ] = = = = 0.5
P[Y = E ] P[Y = E ] 0.1

(d) Given that 1 was received, the probability that 1 was transmitted is given by

P[( X = 1) ∩ (Y = 1)] P[Y = 1 | X = 1]P[ X = 1] (0.7)(0.5)

P[ X = 1 | Y = 1] = = = = 0.875
P[Y = 1] P[Y = 1] 0.4

1.35 A group of students consists of 60% men and 40% women. Among the men, 30% are
foreign students, and among the women, 20% are foreign students. A student is randomly
selected from the group and found to be a foreign student. What is the probability that the
student is a woman?

27
Solution: Let M denote the event that a student is a man, W the event that a student is a woman
and F the event that the student is a foreign student. We are given that P[ M ] = 0.6, P[W ] = 0.4,
P[ F | M ] = 0.3 and P[ F | W ] = 0.2. The probability that a randomly selected student who is
found to be a foreign student is a woman is given by

P[W ∩ F ] P[ F | W ]P[W ] (0.2)(0.4)

P[W | F ] = = = = 0.3077
P[ F ] P[ F | W ]P[W ] + P[ F | M ]P[ M ] (0.2)(0.4) + (0.3)(0.6)

1.36 Joe frequently gets into trouble at school, and past experience shows that 80% of the time
he is guilty of the offense he is accused of. Joe has just gotten into trouble again, and two
other students, Chris and Dana, have been called into the principal’s office to testify
about the incident. Chris is Joe’s friend and will definitely tell the truth if Joe is innocent,
but he will lie with probability 0.2 if Joe is guilty. Dana does not like Joe and so will
definitely tell the truth if Joe is guilty, but will lie with probability 0.3 if Joe is innocent.
(a) What is the probability that Chris and Dana give conflicting testimonies?
(b) What is the probability that Joe is guilty, given that Chris and Dana give conflicting
testimonies?

Solution: Let J denote the event that Joe is innocent, C the event that Chris testifies that Joe is
innocent, D the event that Dana testifies that Joe is innocent, and X the event that Chris and Dana
give conflicting testimonies. We draw the following tree diagram that illustrates the process:

D J CD

J DC
J D
C J DC

(a) The probability that Chris and Dana give conflicting testimonies is given by

P[ X ] = P[ J C D ∪ J D C ] = P[ J C D ] + P[ J D C ] = 0.6 + 0.16 = 0.22

(b) The probability that Joe is guilty, given that Chris and Dana give conflicting testimonies
is given by

P[ J ∩ X ] P[ J D C ] 0.16 8
P[ J | X ] = = = = = 0.7273
P[ X ] P[ X ] 0.22 11
28
1.37 Three car brands A, B, and C have all the market share in a certain city. Brand A has 20%
of the market share, brand B has 30%, and brand C has 50%. The probability that a brand
A car needs a major repair during the first year of purchase is 0.05, the probability that a
brand B car needs a major repair during the first year of purchase is 0.10, and the
probability that a brand C car needs a major repair during the first year of purchase is
0.15.
(a) What is the probability that a randomly selected car in the city needs a major repair
during its first year of purchase?
(b) If a car in the city needs a major repair during its first year of purchase, what is the
probability that it is a brand A car?

Solution: Let A denote the event that a car is a brand A car, B the event that a car is a brand B
car, C the event that a car is a brand C car, and R the event that it needs a major repair during the
first year of purchase. We are given that

P[ A] = 0.2, P[ B] = 0.3, P[C ] = 0.5, P[ R | A] = 0.05, P[ R | B] = 0.10, P[ R | C ] = 0.15

(a) The probability that a randomly selected car needs a major repair during its first year of
purchase is given by

P[ R] = P[ R | A]P[ A] + P[ R | B]P[ B] + P[ R | C ]P[C ]

= (0.05)(0.2) + (0.10)(0.3) + (0.15)(0.5) = 0.115

(b) Given that a car in the city needs a major repair during the first year of its purchase, the
probability that it is a brand A car is given by

P[ A ∩ R] P[ R | A]P[ A] (0.05)(0.2) 2
P[ A | R] = = = = = 0.0870
P[ R ] P[ R ] 0.115 23

Section 1.8: Independent Events

1.38 If I toss two coins and tell you that at least one is heads, what is probability that the first
coin is heads?

Solution: The sample space of the experiment is Ω = {HH , HT , TH , TT }. Let A denote the
event that at least one coin came up heads and B the event that the first coin came up heads. Then
A = {HH , HT , TH } and B = {HH , HT }. Since B is a subset of A, the probability that the first
coin came up heads, given that there is at least one head, is given by

P[ A ∩ B] P[ B ] 1
2
P[ B | A] = = = 2
3
= = 0.667
P[ A] P[ A] 4
3

29
1.39 Assume that we roll two dice and define three events A, B, and C, where A is the event
that the first die is odd, B is the event that the second die is odd, and C is the event that
the sum is odd. Show that these events are pairwise independent but the three are not
independent.

Solution: The sample space of the experiment and the events A, B and C are shown in the
following figure.
Second Roll

To show that these events are pairwise independent we proceed as follows:

18 1 18 1 18 1
P[ A] = = , P[ B] = = , P[C ] = =
36 2 36 2 36 2
9 1
P[ A ∩ B ] = = = P[ A]P[ B ]
36 4
9 1
P[ A ∩ C ] = = = P[ A]P[C ]
36 4
9 1
P[ B ∩ C ] = = = P[ B]P[C ]
36 4

30
Since P[ A ∩ B] = P[ A]P[ B], we conclude that A and B are independent. Similarly, since
P[ A ∩ C ] = P[ A]P[C ], we conclude that A and C are independent. Finally, since
P[ B ∩ C ] = P[ B]P[C ], we conclude that B and C are independent. Thus, the events are pairwise
independent.

Now, A ∩ B ∩ C = Ø. Thus, P[ A ∩ B ∩ C ] = 0 ≠ P[ A]P[ B]P[C ]. Therefore, we conclude that

the three events are not independent.

1.40 Consider a game that consists of two successive trials. The first trial has outcome A or B,
and the second trial has outcome C or D. The probabilities of the four possible outcomes
of the game are as follows:

Outcome AC AD BC BD
Probability 1/3 1/6 1/6 1/3

Determine in a convincing way if A and C are statistically independent.

Solution: Let P[ A] = a, P[C | A] = c and P[C | B] = d . Then we represent the outcome of the
experiment by the following tree diagram:

ac = 1 / 3

1− c
a(1−c) = 1/ 6

(1 − a )d = 1 / 6
1− a

1− d (1− a)(1− d) = 1/ 3

Now,

1 1 1
a(1 − c) = = a − ac = a − ⇒ a =
6 3 2
1
2
c = 13 =
2
3
1 d 1
(1 − a )d = = ⇒d =
6 2 3

31
If A and C are statistically independent, then the outcome of the second trial should be
independent of the outcome of the first trial, and we should have that P[C | A] = P[C | B]; that is,
we should have that c = d . Since this is not the case here, that is, since P[C | A] ≠ P[C | B], we
conclude that A and C are not independent.

1.41 Suppose that two events A and B are mutually exclusive and P[ B] > 0. Under what
conditions will A and B be independent?

Solution: Since the two events are mutually exclusive, we have that

P[ A ∩ B] = P[ A | B]P[ B] = 0

Since P[ B] > 0, we have that P[ A | B] = 0. For A and B to be independent, we must have that
P[ A ∩ B] = P[ A]P[ B]. Since P[ B] > 0, we have that P[ A] = 0.

Section 1.10: Combinatorial Analysis

1.42 Four married couples bought tickets for eight seats in a row for a football game.
(a) In how many different ways can they be seated?
(b) In how many ways can they be seated if each couple is to sit together with the hus-
band to the left of his wife?
(c) In how many ways can they be seated if each couple is to sit together?
(d) In how many ways can they be seated if all the men are to sit together and all the
women are to sit together?
Solution:
(a) Since there are 8 different people, there are 8!= 40320 different ways that they can be
seated.
(b) Since there are 4 couples and each couple can be arranged in only one way, there are
4!= 24 different ways they can be seated if each couple is to sit together with the
husband to the left of his wife.
(c) In this case there is no restriction on how each couple can sit next to each other; the
husband can sit to the right or to the left of his wife. Thus, for each arrangement of the
couples, there are 2! ways of arranging each couple. Also, for each sitting arrangement of
a couple, there are 4!= 24 ways of arranging all the couples. Therefore, the number of
ways that they can be seated, if each couple is to sit together, is given by

4! × (2 !) 4 = 24 × 16 = 384

(d) If all the men sit together and all the women sit together, then there are two groups of
people that can be arranged in 2! ways. Within each group, there are 4! ways of
arranging the members for every arrangement of men and women. Therefore, the number
of arrangements is given by
32
 2! × (4 !) 2 = 2 × 24 × 24 = 1152

1.43 A committee consisting of 3 electrical engineers and 3 mechanical engineers is to be

formed from a group of 7 electrical engineers and 5 mechanical engineers. Find the
number of ways in which this can be done if
(a) any electrical engineer and any mechanical engineer can be included.
(b) one particular electrical engineer must be on the committee.
(c) two particular mechanical engineers cannot be on the same committee.

Solution:
(a) If any electrical engineer and any mechanical engineer can be included, the number of
committees that can be formed is

 7   5  7! 5!
  ×   = × = 350
 3   3  4!3! 2!3!
(b) If one particular electrical engineer must be on the committee, then we need to select 2
more electrical engineers on the committee. Thus, the number of committees that can be
formed is

 6  5 6! 5!
  ×   = × = 150
 2   3  4!2! 2!3!
(c) If two particular mechanical engineers cannot be on the same committee, then we
consider the number of committees that include both of them, which is the number of
committees where we need to choose one mechanical engineer from the remaining 3. The
number of such committees is

 7   3  7! 3!
  ×   = × = 105
 3  1  4!3! 2!1!
Thus, the number of committees that do not include both mechanical engineers together
is given by

 7   5   7   3 
  ×   −   ×   = 350 − 105 = 245
 3   3   3  1 

1.44 Use Stirling’s formula to evaluate 200!

Solution: The Stirling’s formula is given by n !≈ 2πn × n n × e − n . Thus, we have that

33
 200!≈ 400π × 200 200 × e −200 = 20 π × 200 200 × e −200
log (200!) ≈ log (20) + 12 log(π ) + 200 log (200) − 200 log(e)
= 1.30103 + 0.24857 + 460.20600 − 86.85890 = 374.89670
= 374 + 0.89670

Taking the antilog we obtain

200!≈ 7.88315 × 10 374

1.45 A committee of 3 members is to be formed consisting of one representative from labor,

one from management, and one from the public. If there are 7 possible representatives
from labor, 4 from management, and 5 from the public, how many different committees
can be formed?

Solution: The number of committees that can be formed is

 7   4  5
  ×   ×   = 7 × 4 × 5 = 140
1  1  1 

1.46 There are 100 U.S. senators, two from each of the 50 states.
(a) If 2 senators are chosen at random, what is the probability that they are from the same
state?
(b) If 10 senators are randomly chosen to form a committee, what is the probability that
they are all from different states?

Solution: The number of ways of indiscriminately choosing two senators from the 50 states is

100  100! 100 × 99

  = = = 4950
 2  2 !98! 2

(a) The probability that 2 senators chosen at random are from the same state is
49
 50   2   2 
  ×   ×  
 1   2   0  50 1
= = = 0.0101
100  4950 99
 
 2 

where the first combinatorial term in the numerator is the number of ways of choosing
the state from where the senators come, the second term is the number of ways of
choosing the 2 senators from that state, and the third term accounts for the fact that no
senator is chosen from any of the remaining 49 states.

34
 (b) The probability of randomly choosing 10 senators who are all from different states can be
obtained as follows. There are C (50,10) ways of choosing the 10 states where the 10
senators come from, and for each of these states there are C (2,1) ways of choosing 1
senator out of the 2 senators. For the remaining 40 states, no senators are chosen.
Therefore, the probability of this event is given by
10 40
 50   2   2 
  ×   ×  
10  1   0  50 ! × 210 × 90! × 10 !
= = 0.60766
100  100! × 10 ! × 40!
 
 2 

1.47 A committee of 7 people is to be formed from a pool of 10 men and 12 women.

(a) What is the probability that the committee will consist of 3 men and 4 women?
(b) What is the probability that the committee will consist of all men?

Solution:
(a) The probability that the committee will consist of 3 men and 4 women is given by

10  12 
  ×  
 3   4  = 10 ! × 12 ! × 7 !× 15! = 0.3483
 22  22! × 3! × 7 ! × 4!× 8!
 
7 

(b) The probability that the committee will consist of all men is given by

10  12  10 

  ×    
 7   0  =  7  = 10 ! × 7 !× 15! = 0.0007
 22   22  22! × 3! × 7 !
   
7  7 

1.48 Five departments in the college of engineering, which are labeled departments A, B, C, D
and E, send three delegates each to the college’s convention. A committee of four
delegates, selected by lot, is formed. Determine the probability that
(a) Department A is not represented on the committee.
(b) Department A has exactly one representative on the committee.
(c) Neither department A nor department C is represented on the committee.

Solution:
(a) The probability that department A is not represented on the committee is the probability
of choosing 0 delegates from department A and 4 from the other 12 delegates, which is
given by
35
  3  12  12 
  ×    
 0   4  =  4  = 12! × 11!× 4! = 33 = 0.36264
15  15  15! × 4! × 8! 91
   
4 4

(b) The probability that department A has exactly one representative on the committee is the
probability that 1 delegate is chosen out of the 3 from department A and 3 delegates are
chosen out of the 12 delegates from the other departments, which is given by

 3  12  12 
  ×   3 ×  
1   3  =  3  = 3 × 12! × 11!× 4 ! = 44 = 0.48352
15  15  15! × 3! × 9 ! 91
   
4 4

(c) The probability that neither department A nor department C is represented on the
committee is the probability that all chosen delegates are from departments B, D and E;
and this is given by
2
 3   9   9 
  ×    
 0   4  =  4  = 9! × 11!× 4 ! = 0.0923
15  15  15! × 4 ! × 5!
   
4 4

Section 1.11: Reliability Applications

1.49 Consider the system shown in Figure 1.20. If the number inside each box indicates the
probability that the component will independently fail within the next two years, find the
probability that the system fails within two years.

Figure 1.20: Figure for Problem 1.49

36
Solution: We first define the problem in terms of reliability functions and obtain the following
diagram:

R4 = 0.98
R1 = 0.95 R2 = 0.99

R5 = 0.98

R3 = 0.99
R6 = 0.97

Next we combine the series-connected R1 and R2 into R12 = R1 R2 = 0.9405, and combine the
parallel-connected R4 , R5 and R6 into a composite system with reliability function R456 , which
is defined as follows:
R456 = 1 − (1 − R4 )(1 − R5 )(1 − R6 ) = 0.99999
The reduced figure is as shown below:

R12

R 456

R3

The final reduction combines R12 and R3 to form R123 = 1 − (1 − R12 )(1 − R3 ) = 0.999405, which
is now in series with R456 , as shown below:

R123 R 456

Thus, the reliability of the system is given by

R = R123 × R456 = 0.999405 × 0.99999 = 0.999395

Therefore, the probability that the system fails within two years is 1 − R = 0.0006 .

37
1.50 Consider the structure shown in Figure 1.21. Switches S1 and S 2 are in series, and the
pair is in parallel with a parallel arrangement of switches S 3 and S 4 . Their reliability
functions are R1 (t ), R2 (t ), R3 (t ) and R4 (t ), respectively. The structure interconnects
nodes A and B. What is the reliability function of the composite system in terms of
R1 (t ), R2 (t ), R3 (t ) and R4 (t ) if the switches fail independently?

S1 S2

S3

S4

Figure 1.21: Figure for Problem 1.50

Solution: We start by observing that because S1 and S 2 are in series, we can reduce them into a
composite switch S12 whose reliability function is R12 (t ) = R1 (t ) R2 (t ). Thus, we obtain the
following system:

S 12

A S3 B

S4

Thus, the reliability function of the system is given by

R(t ) = 1 − {1 − R12 (t )}{1 − R3 (t )}{1 − R4 (t )} = 1 − {1 − R1 (t ) R2 (t )}{1 − R3 (t )}{1 − R4 (t )}

1.51 Consider the network shown in Figure 1.22 that interconnects nodes A and B. The
switches labeled S1 , S 2 , Κ , S 8 have the reliability functions R1 (t ), R2 (t ), Κ , R8 (t ),
respectively. If the switches fail independently, find the reliability function of the
composite system.

38
 S3
S1 S2
S4

S5

S7
S6
S8

Figure 1.22: Figure for Problem 1.51

Solution: We start with the first level of system reduction by combining switches S1 and S 2 to
form switch S12 whose reliability function is R12 (t ) = R1 (t ) R2 (t ). We also combine S 3 and S 4 to
form switch S 34 , and combine S 7 and S 8 to form switch S 78 . Their respective reliability
functions are as follows:

R34 (t ) = 1 − {1 − R3 (t )}{1 − R4 (t )}
R78 (t ) = 1 − {1 − R7 (t )}{1 − R8 (t )}

The structure of the reduced system is as shown in the figure below.

S1 S2 S 34

S5

S6 S 78

39
Next, we combine switches S12 and S 34 to form composite switch S1234 , and we combine
switches S 6 and S 78 to form composite switch S 678 . The reliability functions of these composite
switches are given by

R1234 (t ) = R12 (t ) R24 (t ) = R1 (t ) R2 (t ){1 − {1 − R3 (t )}{1 − R4 (t )}}

R678 (t ) = R6 (t ) R78 (t ) = R6 (t ){1 − {1 − R7 (t )}{1 − R8 (t )}}

The final structure is a system of three components in parallel, as shown below.

S 1234

A S5 B

S678

Thus, the reliability function of the system is given by

R(t ) = 1 − {1 − R1234 (t )}{1 − R5 (t )}{1 − R678 (t )}

1.52 Consider the network shown in Figure 1.23 that interconnects nodes A and B. The
switches labeled S1 , S 2 , Κ , S 8 have the reliability functions R1 (t ), R2 (t ), Κ , R8 (t ),
respectively. If the switches fail independently, find the reliability function of the
composite system.

S1 S3 S5

S7 S8

S2 S4 S6

Figure 1.23: Figure for Problem 1.52

40
Solution: We consider the following 4 cases associated with the bridge switches S 7 and S 8 :
(A) Switches S 7 and S 8 do not fail by time t; the probability of this event is
P[ A] = R7 (t ) R8 (t )
(B) Switch S 7 fails, but switch S 8 does not fail by time t; the probability of this event is
P[ B] = {1 − R7 (t )}R8 (t )
(C) Switch S 7 does not fail, but switch S 8 fails by time t; the probability of this event is
P[C ] = R7 (t ){1 − R8 (t )}
(D) Switches S 7 and S 8 fail by time t; the probability of this event is
P[ D] = {1 − R7 (t )}{1 − R8 (t )}
Case A is equivalent to the following system:

S1 S3 S5

S2 S4 S6

This in turn can be reduced to the following system:

S 12 S 34 S56

The respective reliability functions of the switches labeled S12 , S 34 and S 56 are given
respectively as follows:
R12 (t ) = 1 − {1 − R1 (t )}{1 − R2 (t )}
R34 (t ) = 1 − {1 − R3 (t )}{1 − R4 (t )}
R56 (t ) = 1 − {1 − R5 (t )}{1 − R6 (t )}

Thus, the reliability function for Case A, R A (t ), is given by

R A (t ) = R12 (t ) R34 (t ) R56 (t )

where R12 (t ), R34 (t ) and R56 (t ) are as previously defined.

41
Case B is equivalent to the following system:

S1 S3 S5

S2 S4 S6

This can be further reduced as follows:

S 13 S5

S 24 S6

The respective reliability functions of the switches labeled S13 and S 24 are given respectively as
follows:

R13 (t ) = R1 (t ) R3 (t )
R24 (t ) = R2 (t ) R4 (t )

Finally we can reduce the system to the following structure:

S1234 S56

where the reliability functions of switches S1234 and S 56 are given by

R1234 (t ) = 1 − {1 − R13 (t )}{1 − R24 (t )}

R56 (t ) = 1 − {1 − R5 (t )}{1 − R6 (t )}

42
Thus, the reliability function for Case B, RB (t ), is given by

RB (t ) = R1234 (t ) R56 (t )

Case C is equivalent to the following system:

S1 S3 S5

S2 S4 S6

This can be further reduced as follows:

S1 S 35

S2 S46

The respective reliability functions of the switches labeled S 35 and S 46 are given respectively as
follows:
R35 (t ) = R3 (t ) R5 (t )
R46 (t ) = R4 (t ) R6 (t )

Finally we can reduce the system to the following structure:

S 12 S 3456

43
where the reliability functions of switches S12 and S 3456 are given by

R12 (t ) = 1 − {1 − R1 (t )}{1 − R2 (t )}
R3456 (t ) = 1 − {1 − R35 (t )}{1 − R46 (t )}

Thus, the reliability function for Case C, RC (t ), is given by

RC (t ) = R12 (t ) R3456 (t )

Case D is equivalent to the following system:

S1 S3 S5

S2 S4 S6

This can be further reduced as follows:

S135

S246

The respective reliability functions of the switches labeled S135 and S 246 are given respectively
as follows:
R135 (t ) = R1 (t ) R3 (t ) R5 (t )
R246 (t ) = R2 (t ) R4 (t ) R6 (t )

Thus, the reliability function for Case D, RD (t ), is given by

44
 RD (t ) = 1 − {1 − R135 (t )}{1 − R246 (t )}

Finally, the reliability function of the system is given by

R(t ) = R A (t ) P[ A] + RB (t ) P[ B ] + RC (t ) P[C ] + RD (t ) P[ D]

where P[ A], P[ B], P[C ], P[ D], R A (t ), RB (t ), RC (t ) and RD (t ) are as previously defined.

1.53 Consider the network shown in Figure 1.24 that interconnects nodes A and B. The
switches labeled S1 , S 2 , Κ , S 7 have the reliability functions R1 (t ), R2 (t ), Κ , R7 (t ),
respectively. If the switches fail independently, find the reliability function of the
composite system.

S1

S2 S4

S7
S5

S3 S6

Figure 1.24: Figure for Problem 1.53

Solution: The network is equivalent to the following system:

This can now be reduced to the following system:

45
 S1

S7

S23 S 456

The reliability functions of the switches labeled S 23 and S 456 are as follows:

R23 (t ) = 1 − {1 − R2 (t )}{1 − R3 (t )}
R456 (t ) = 1 − {1 − R4 (t )}{1 − R5 (t )}{1 − R6 (t )}

The reliability function of the switches labeled S1 , S 23 and S 456 is given by

R123456 (t ) = 1 − {1 − R1 (t )}{1 − R23 (t ) R456 (t )}

The final configuration is as shown below:

S123456 S7

Thus, the reliability function of the system is given by

R(t ) = R123456 (t ) R7 (t )

46
Chapter 2: Random Variables

Chapter Summary: This chapter discusses random variables including events defined by
random variables, discrete random variables, continuous random variables, cumulative
distribution function, probability mass function of discrete random variables, and probability
density function of continuous random variables.

Section 2.4: Distribution Functions

2.1 Bob claims that he can model his experimental study of a process by the following CDF:

0 −∞ < x ≤1
FX ( x ) = 
[
B 1 − e
− ( x −1)
] 1< x < ∞

(a) For what value of B is the function a valid CDF?

(b) With the above value of B, what is FX (3) ?
(c) With the above value of B, what is P[2 < X < ∞] ?
(d) With the above value of B, what is P[1 < X ≤ 3] ?

Solution:
(a) For the function to be a valid CDF, it must satisfy the condition

FX (∞) = 1 = B{1 − e −∞ } = B{1 − 0} ⇒ B = 1

(b) FX (3) = 1 = B{1 − e − ( 3−1) } = 1 − e −2 = 0.86466

(c) P[2 < X < ∞] = FX (∞) − FX (2) = 1 − FX (2) = 1 − {1 − e −1 } = e −1 = 0.3679

(d) P[1 < X < 3] = FX (3) − FX (1) = {1 − e −2 }− {1 − e −0 } = 1 − e −2 = 0.86466

2.2 The CDF of a random variable X is given by

0 x<0

FX ( x) = 3 x 2 − 2 x 3 0 ≤ x <1
1 x ≥1


What is the PDF of X?

Solution: The PDF of X is given by

1
 d 6 x − 6 x 2 0 ≤ x <1
f X ( x) = FX ( x ) = 
dx 0 otherwise

2.3 A random variable X has the CDF

0 x<0
FX ( x ) = 
1 − e − x / 2σ
2 2
x≥0

where σ is a positive constant.

(a) Find P[σ ≤ X ≤ 2σ ]
(b) Find P[ X > 3σ ].

Solution:
(a) P[σ ≤ X ≤ 2σ ] = FX (2σ ) − FX (σ ) = 1 − e −4σ
2
/ 2σ 2
{
− 1 − e −σ
2
/ 2σ 2
}= e −0.5
− e −2 = 0.4712

(b) P[ X > 3σ ] = 1 − FX (3σ ) = e −9σ / 2σ 2

= e −4.5 = 0.0111
2

2.4 The CDF of a random variable T is given by

0 t<0

FT (t ) = t 2 0 ≤ t <1
1 t ≥1


(a) What is the PDF of T?

(b) What is P[T > 0.5] ?
(c) What is P[0.5 < T ≤ 0.75] ?
Solution:
(a) The PDF of T is given by
d 2t 0 ≤ x <1
f T (t ) = FT (t ) = 
dt 0 otherwise

(b) P[T > 0.5] = 1 − P[T ≤ 0.5] = 1 − FT (0.5) = 1 − (0.5) 2 = 1 − 0.25 = 0.75
(c) P[0.5 < T ≤ 0.75] = FT (0.75) − FT (0.5) = (0.75) 2 − (0.5) 2 = 0.3125

2.5 The CDF of a continuous random variable X is given by

2
 0 x < − π2

FX ( x) = k {1 + sin( x)} − π2 ≤ x < π
2
1 x≥ π
 2

(a) Find the value of k.

(b) Find the PDF of X.
Solution:
(a) FX ( π2 ) = 1 = k {1 + sin ( π2 )} = 2k ⇒ k = 0.5
Alternatively we can proceed by first obtaining the PDF of X and integrating it over the
full range of X, as follows:
d
f X ( x) = FX ( x) = k cos( x), − π2 ≤ x ≤ π2
dx
π

k cos( x)dx = k [sin( x)]−2 π = 2k ⇒ k =

∞ π

∫ f X ( x)dx = 1 = ∫ 2 1
−∞ π
−2 2
2

(b) As shown above, the PDF of X is given by

d k cos( x) − π2 ≤ x ≤ π
f X ( x) = FX ( x ) =  2

dx 0 otherwise

2.6 The CDF of a random variable X is given by

0 x<2

FX ( x ) =  4
1 − x 2 x≥2

(a) Find P[ X < 3]

(b) Find P[4 < X < 5].
Solution:
(a) Because X is a continuous random variable, P[ X < 3] = P[ X ≤ 3] = FX (3) = 1 − 94 = 5
9

(b) P[4 < X < 5] = FX (5) − FX (4) = {1 − 254 } − {1 − 164 } = 100

9
= 0.09

2.7 The CDF of a discrete random variable K is given by

0.0 k < −1
0.2 −1 ≤ k < 0

FK (k ) = 
0.7 0 ≤ k <1
1.0 k ≥1

3
 (a) Draw the graph of FK (k )
(b) Find p K (k ), the PMF of K.
Solution:
(a) The graph of FK (k ) is as follows:
FK (k)

(b) To find the PMF of K, we observe that it has nonzero values at those values of k where
the value of the CDF changes, and its value at any such point is equal to the change in the
value of the CDF. Thus,
0.2 k = −1
0.5 k =0

p K (k ) = 
0.3 k =1
0 otherwise

2.8 The random variable N has the CDF

0.0 n < −2
0.3 −2≤ n<0

FN (n) = 0.5 0≤n<2
0.8 2≤n<4

1.0 n≥4
(a) Draw the graph of FN (n)
(b) Find p N (n), the PMF of N.
(c) Draw the graph of p N (n).
Solution:
4
(a) The graph of FN (n) is as follows:
FN (n)

(b) The PMF of N is given by

0.3 n = −2
0.2 n=0

p N (n) = 0.3 n=2
0.2 n=4

0 otherwise
(c) The graph of p N (n) is as follows:
pN (n)

5
2.9 The CDF of a discrete random variable Y is given by
0.0 y<2
0.3 2≤ y<4

FY ( y ) = 
0.8 4≤ y<6
1.0 y≥6

(a) What is P[3 < Y < 4] ?

(b) What is P[3 < Y ≤ 4] ?
Solution:
(a) P[3 < Y < 4] = FY (4 − ) − FY (3) = 0.3 − 0.3 = 0. We have used 4 − to indicate the fact that
we are looking at values that are close to but not equal to 4. Another way to see this is to
first obtain the PMF of Y, which is given by
0.3 y=2
0.5 y=4

pY ( y ) = 
0.2 y=6
0 otherwise
Thus, P[3 < Y < 4] = pY (3) = 0
(b) P[3 < Y ≤ 4] = FY (4) − FY (3) = 0.8 − 0.3 = 0.5 = pY (4)

2.10 Determine the PMF of the random variable Y if its CDF is given by

0.0 y<0
0.50 0≤ y<2

FY ( y ) = 0.75 2≤ y<3
0.90 3≤ y <5

1.0 y≥5

Solution: The PMF of Y is given by

0.50 y=0
0.25 y=2

pY ( y ) = 0.15 y=3
0.10 y=5

0 otherwise

6
2.11 The CDF of a discrete random variable X is given as follows:

0 x<0
1
4 0 ≤ x <1

FX ( x) =  12 1≤ x < 3
5 3≤ x< 4
8
1 x≥4

(a) Determine p X (x), the PMF of X, and draw its graph.

(b) Determine the values of (i) P[ X < 2] and (ii) P[0 ≤ X < 4].
Solution:
(a) The PMF of X is given by
 14 x=0
1
4 x =1

p X ( x) =  18 x=3
3 x=4
8
0 otherwise
The graph of the PMF is as shown below:
pX (x)

1.0

0.8

0.6

0.4

0.2

x
0 1 2 3 4

(b) (i) P[ X < 2] = p X (0) + p X (1) = 14 + 14 = 1

2

(ii) P[0 ≤ X < 4] = p X (0) + p X (1) + p X (3) = 14 + 14 + 18 = 5

8

7
Section 2.5: Discrete Random Variables
2.12 Let the random variable K denote the number of heads in 4 flips of a fair coin.
(a) Plot the graph of p K (k ).
(b) What is P[ K ≥ 3] ?
(c) What is P[2 ≤ K ≤ 4] ?
Solution:
The PMF of K can be obtained as follows. p K (k ) is the probability that there are k heads and,
therefore, 4 − k tails. Since the probability of a head on any toss is 12 and the outcomes of the
flips are independent, the probability of k heads in 4 flips of the coin is
p K (k ) = C (4, k )( 12 ) k ( 12 ) 4−k = C (4, k )( 12 ) 4 , where the combinatorial term C (4, k ) is used to
account for the possible combinations of k heads and 4 − k tails. Thus, we have that

 161 k =0
1
4 k =1
 3 k=2
p K (k ) =  81
4 k =3
1 k=4
 16
0 otherwise
(a) The graph of p K (k ) is as follows:
pK (k)

3
8

1
4

1
8

k
0 1 2 3 4

(b) P[ K ≥ 3] = p K (3) + p K (4) = 165

(c) P[2 ≤ K ≤ 4] = p K (2) + p K (3) + p K (4) = 16
11

8
2.13 Ken was watching some people play poker, and he wanted to model the PMF of the
random variable N that denotes the number of plays up to and including the play in which
his friend Joe won a game. He conjectured that if p is the probability that Joe wins any
game and the games are independent, then the PMF of N is given by

p N (n) = p (1 − p ) n −1 n = 1, 2, Κ

(a) Show that p N (n) is a proper PMF.

(b) Find the CDF of N.
Solution:
(a) For p N (n) to be a proper PMF it must satisfy the condition ∑ n
p N (n) = 1. Thus, we
evaluate the sum:
∞ ∞
 1  p
∑ p ( n ) = 1 = p ∑ (1 − p ) n −1 = p   = =1
1 − (1 − p)  p
N
n =1 n =1

Thus, we conclude that p N (n) is a proper PMF.

(b) The CDF of N is given by

n n n −1
1 − (1 − p) n 
FN (n) = P[ N ≤ n] = ∑ p N (k ) = p ∑ (1 − p ) k −1 = p ∑ (1 − p ) m = p  
k =1 k =1 m =0  1 − (1 − p) 
= 1 − (1 − p ) n n = 1, 2, Κ

2.14 The discrete random variable K has the following PMF:

b k =0
2b k =1

p K (k ) = 
3b k=2
0 otherwise

(a) What is the value of b?

(b) Determine the values of (i) P[ K < 2], (ii) P[ K ≤ 2] and (iii) P[0 < K < 2].
(c) Determine the CDF of K.

Solution:
(a) ∑k p K (k ) = 1 = b + 2b + 3b = 6b ⇒ b = 1
6

(b) P[ K < 2] = p K (0) + p K (1) = 3b = 12

P[ K ≤ 2] = p K (0) + p K (1) + p K (2) = 6b = 1
P[0 < K < 2] = p K (1) = 2b = 1
3

9
 (c) The CDF of K is given by

0 k <0
1 0 ≤ k <1
6
FK (k ) =  1
2 1≤ k < 2
1 k≥2


2.15 A student got a summer job at a bank, and his assignment was to model the number of
customers who arrive at the bank. The student observed that the number of customers K
that arrive over this period had the PMF
 λk −λ
 e k = 0, 1, 2, Κ ; λ > 0
p K ( k ) =  k!
0 otherwise


(a) Show that p K (k ) is a proper PMF.

(b) What is P[ K > 1] ?
(c) What is P[2 ≤ K ≤ 4] ?

Solution:
(a) To show that p K (k ) is a proper PMF, we must have that it sums to 1 over all values of k;
that is,
∞ ∞
λk
k =0
∑ p K ( k ) = e −λ ∑
k =0 k !
= e −λ e λ = 1

Thus, we conclude that p K (k ) is a proper PMF.

(b) P[ K > 1] = 1 − P[ K ≤ 1] = 1 − p K (0) − p K (1) = 1 − e − λ {1 + λ }

 λ 2 λ3 λ 4 
−λ
(c) P[2 ≤ K ≤ 4] = p K (2) + p K (3) + p K (4) = e  + + 
2 6 24 

2.16 Let X be the random variable that denotes the number of times we roll a fair die until the
first time the number 5 appears. Find the probability that X = k , if the outcomes of the
rolls are independent.

Solution: Since the probability that the number 5 appears in any roll is 16 , the probability that
X = k is the probability that we had no number 5 in the previous k − 1 rolls and the number 5 in
the kth roll. Since the outcomes of the different rolls are independent, we have that

P[ K = k ] = ( 56 ) ()
k −1 1
6

10
2.17 The PMF of a random variable X is given by p X ( x) = bλ x / x!, x = 0, 1, 2, Κ , where
λ > 0. Find the values of (a) the parameter b, (b) P[ X = 1] and (c) P[ X > 3].

Solution:
(a) To obtain b we know that
∞ ∞
λx
∑p
x =0
X ( x) = 1 = b∑
x =0 x!
= be λ ⇒ b = e −λ

(b) P[ X = 1] = p X (1) = λe − λ
 λ 2 λ3 
(c) P[ X > 3] = 1 − P[ X ≤ 3] = 1 − {p X (0) + p X (1) + p X (2) + p X (3)} = 1 − e −λ 1 + λ + + 
 2 6

2.18 A random variable K has the PMF

5 
p K (k ) =  (0.1) k (0.9) 5− k k = 0, 1, Κ , 5
k 
Obtain the values of the following:

(a) P[ K = 1]
(b) P[ K ≥ 1]
Solution:
5
(a) P[ K = 1] = p K (1) =  (0.1)1 (0.9) 5−1 = 5(0.1)(0.9) 4 = 0.32805
1 
(b) P[ K ≥ 1] = 1 − P[ K < 1] = 1 − P[ K = 0] = 1 − (0.9) 5 = 0.40951

2.19 A biased four-sided die has faces labeled 1, 2, 3, and 4. Let the random variable X denote
the outcome of a roll of the die. Extensive testing of the die shows that the PMF of X is
given by
0.4 x =1
0.2 x=2

p X ( x) = 
0.3 x=3
0.1 x=4

(a) Find the CDF of X.

(b) What is the probability that a number less than 3 appears on a roll of the die?
(c) What is the probability of obtaining a number whose value is at least 3 on a roll of the
die?
Solution:
11
 (a) The CDF of X is given by
0.0 x <1
0.4 1≤ x < 2

FX ( x) = 0.6 2≤ x<3
0.9 3≤ x<4

1.0 x≥4
(b) P[ X < 3] = p X (1) + p X (2) = 0.6
(c) P[ X ≥ 3] = 1 − P[ X < 3] = 1 − 0.6 = 0.4

2.20 The number N of calls arriving at a switchboard during a period of one hour has the PMF
10 n −10
p N ( n) = e n = 0, 1, 2, Κ
n!
(a) What is the probability that at least 2 calls arrive within one hour?
(b) What is the probability that at most 3 calls arrive within one hour?
(c) What is the probability that the number of calls that arrive within one hour is greater
than 3 but less than or equal to 6?
Solution:
(a) P[ N ≥ 2] = 1 − P[ N < 2] = 1 − {p N (0) + p N (1)} = 1 − e −10 {1 + 10} = 0.9992

−10  10 2 10 3 
(b) P[ N ≤ 3] = p N (0) + p N (1) + p N (2) + p N (3) = e 1 + 10 + +  = 0.01034
 2 6 

−10 10 4 10 5 10 6 
(c) P[3 < N ≤ 6] = p N (4) + p N (5) + p N (6) = e  + +  = 0.1198
 24 120 720 

2.21 Assume that the random variable K denotes the number of successes in n trials of an
experiment. The probability of success in any trial of the experiment is 0.6, and the PMF of
K is given by
n
p K (k ) =  (0.6) k (0.4) n − k k = 0, 1, Κ , n; n = 1, 2, Κ
k 
(a) What is the probability of at least 1 success in 5 trials of the experiment?
(b) What is the probability of at most 1 success in 5 trials of the experiment?
(c) What is the probability that the number of successes is greater than 1 but less than 4 in
5 trails of the experiment?
Solution:
(a) P[ K ≥ 1, n = 5] = 1 − p K (0) n =5 = 1 − (0.4) 5 = 0.98976
(b) P[ K ≤ 1, n = 5] = [ p K (0) + p K (1)]n =5 = (0.4) 5 + 5(0.6)(0.4) 4 = 0.08704

12
 (c) P[1 < K < 4, n = 5] = [ p K (2) + p K (3)]n =5 = 10(0.6) 2 (0.4) 3 + 10(0.6) 3 (0.4) 2 = 0.576

2.22 Prove that the function p(x) is a legitimate PMF of a discrete random variable X, where p(x)
is defined by
2  1  x
   x = 0, 1, 2, Κ
p( x) =  3  3 
0 otherwise

Solution: We have that

2  1  2 1  2 3 
∞ x
2 ∞ 1
∑ p ( x) =
x =0
∑  
3 x =0  3 
=  1 = 2 =   =1
3 1 − 3  3  3  3  2 

Thus, p (x) is a valid PMF.

Section 2.6: Continuous Random Variables

2.23 Consider the following function.
a (1 − x 2 ) −1 < x < 1
g ( x) = 
0 otherwise

(a) Determine the value of a that makes g(x) a valid probability density function.
(b) If X is the random variable with this PDF, determine the value of P[0 < K < 0.5].

Solution:
(a) For g (x ) to be a valid PDF we must have that

∫
∞

x = −∞
g ( x)dx = 1 = ∫
1

x = −1
a (1 − x 2 )dx = a x − [ ]
x3
1
3 −1 = 4a
3 ⇒a= 3
4

(b) If X is the random variable with this PDF, then

0.5

x =0
0.5

x =0
3
4
x3 
3
4
3 0
0.5
P[0 < X < 0.5] = ∫ g ( x)dx = ∫ 1 − x dx =  x −  = 0.34375( 2
)

2.24 The PDF of a continuous random variable X is defined as follows for λ > 0,
bxe − λx 0≤x<∞
f X ( x) = 
0 otherwise

(a) What is the value of b?

(b) What is the CDF of X?

13
 (c) What is P[0 ≤ X ≤ 1 / λ ] ?

Solution:
(a) To obtain the value of b we have that
∞ ∞
∫ f X ( x)dx = 1 = b ∫ xe −λx dx
x = −∞ x =0

Let u = x ⇒ du = dx, and let dv = e − λx dx ⇒ v = − e λ . Thus,

− λx

∞
∞  xe −λx ∞ e − λx  b  e − λx  ∞ b
− λx
b∫ xe dx = 1 = b − +∫ dx  = −  = 2
x =0
 λ 0
x =0 λ  λ  λ  0 λ

From this we obtain b = λ2 .

(b) The CDF of X is given by
x x
FX ( x) = P[ X ≤ x] = ∫ f X ( w)dw = λ2 ∫ we −λw dw
w= −∞ w= 0

Let u = w ⇒ du = dw, and let dv = e − λw dw ⇒ v = − e λ . Thus,

− λw

 we −λw x
x e − λw   e − λw 
x
− λx − λx − λx
FX ( x ) = λ  −
2
+∫ dw = −λxe + λ −  = 1 − e − λxe , x ≥ 0
 λ 0
w=0 λ   λ 0

(c) P[0 ≤ X ≤ 1 / λ ] = FX ( λ1 ) = 1 − e −1 − e −1 = 1 − 2e −1 = 0.26424

2.25 Find the PDF of the continuous random variable X with the following CDF:

0 x<0

FX ( x ) =  2 x 2 − x 3 0 ≤ x <1
1 x ≥1

Solution: The PDF is given by

d 4 x − 3 x 2 0 ≤ x <1
f X ( x) = FX ( x ) = 
dx 0 otherwise

2.26 A random variable X has the following PDF, where K > 0:

14
 0 x <1
 K ( x − 1) 1≤ x < 2

f X ( x) = 
 K (3 − x) 2≤ x<3
0 x≥3

(a) What is the value of K?

(b) Sketch f X (x)
(c) What is the CDF of X?
(d) What is P[1 ≤ X ≤ 2] ?

Solution:
(a) The value of K that makes f X (x) a valid PDF can be obtained as follows:

∞  x 2 
2
 x 2  
3

∫ x=−∞ f X ( x)dx = 1 = K ∫ x=1 (x − 1)dx + ∫ x=2 (3 − x )dx = K  2 − x + 3x − 2  
 
2 3

 1 2

1 1 
= K +  = K
2 2
Thus, K = 1.

(b) The plot of f X (x) is as follows:

f X (x)

(c) The CDF of X is given by

15
 0 x <1
 x
 K ∫ x =1 (u − 1)du 1≤ x < 2

FX ( x) = P[ X ≤ x] = 
 K  ∫ (u − 1)du + ∫ (3 − u )du 
2 x
2≤ x<3
  u =1 u =2 
1 x≥3
0 x <1
 2
x − x + 1 1≤ x < 2

= 2 2
2
3 x − x − 7 2≤ x<3
 2 2

1 x≥3

1 1
(d) P[1 ≤ X ≤ 2] = FX (2) − FX (1) = −0 =
2 2

2.27 A random variable X has the CDF

0 x < −1

FX ( x) =  A(1 + x) −1 ≤ x < 1
1 x ≥1


(a) What is the value of A?

(b) With the above value of A, what is P[ X > 14 ] ?
(c) With the above value of A, what is P[−0.5 ≤ X ≤ 0.5] ?

Solution:
(a) We know that FX (1) = 1 ⇒ 2 A = 1 ⇒ A = 12 . Alternatively,

d
f X ( x) = FX ( x) = A, −1 ≤ x < 1
dx
∞ 1 1
1= ∫ f X ( x)dx = ∫ Adx = 2 A ⇒ A =
−∞ −1 2

5 3
(b) P[ X > 14 ] = 1 − P[ X ≤ 14 ] = 1 − FX ( 14 ) = 1 − A(1 + 14 ) = 1 − =
8 8
1
(c) P[−0.5 ≤ X ≤ 0.5] = FX (0.5) − FX (−0.5) = {(1 + 0.5) − (1 − 0.5)} = 1
2 2

16
2.28 The lifetime X of a system in weeks is given by the following PDF:

0.25e −0.25 x x≥0

f X ( x) = 
0 otherwise

(a) What is the probability that the system will not fail within two weeks?
(b) Given that the system has not failed by the end of the fourth week, what is the
probability that it will fail between the fourth and sixth weeks?

Solution: The CDF of X is given by

FX ( x ) = ∫
x

u = −∞
x
[
f X (u )du = 0.25∫ e −0.25u du = − e −0.25u
u =0
]
x
0 = 1 − e −0.25 x

(a) The probability that the system will not fail within two weeks is given by

P[ X > 2] = 1 − P[ X ≤ 2] = 1 − FX (2) = 1 − {1 − e −0.5 } = e −0.5 = 0.6065

(b) Given that the system has not failed by the end of the fourth week, the probability that it
will fail between the fourth and sixth weeks is given by

P[(4 < X < 6 ) ∩ ( X > 4 )] P[4 < X < 6] FX (6) − FX (4)

P[4 < X < 6 | X > 4] = = =
P[ X > 4] P[ X > 4] 1 − FX (4)
e −1 − e −1.5
= = 1 − e −0.5 = 0.3935
e −1

where the second equality follows from the fact that the event {4 < X < 6} is a subset of
the event {X > 4}.

2.29 A military radar is set at a remote site with no repair facility. The time T until the radar fails
in years has the PDF given by
0.2e −0.2t t≥0
f T (t ) = 
0 otherwise

What is the probability that the radar lasts for at least four years?

Solution: The probability that the radar lasts for at least four years is given by

P[T ≥ 4] = ∫
∞

t =4
∞
[
f T (t )dt = 0.2∫ e −0.2t dt = − e −0.2t
t =4
]
∞
4 = e −0.2 ( 4 ) = e −0.8 = 0.4493

2.30 The PDF of a random variable X is given by

17
 A
 x > 10
f X ( x) =  x 2
0 otherwise

(a) What is the value of A?

(b) With the above value of A, determine the CDF of X.
(c) With the above value of A, what is P[ X > 20] ?

Solution:
(a) Since f X (x) is a valid PDF, we have that

∞
∞  1  ∞  1 A
∫ −∞
f X ( x)dx = A∫  2 dx =
 
10 x
A −  =
 x  10 10
⇒ A = 10

(b) The CDF of X is given by

x 0 x ≤ 10
x  10 
x  10  
FX ( x) = ∫ f X (u )du = ∫  2 du = −  =  10
−∞ −10 u
   u 10 1 − x > 10
 x

 10  1
(c) P[ X > 20] = 1 − P[ X ≤ 20] = 1 − FX (20) = 1 − 1 −  =
 20  2

2.31 Assume that X is a continuous random variable with the following PDF:

 A(3x 2 − x 3 ) 0≤ x<3
f X ( x) = 
0 otherwise

(a) What is the value of A?

(b) Find P[1 < X < 2].

Solution:
(a) Since f X (x) is a valid PDF, we have that
3

∫
∞

−∞
f X ( x )dx = A∫
3
( )
 x4   81 27 A
3 x − x dx = A x 3 −  = A27 −  =
2 3
⇒ A=
4
0
 4 0  4 4 27

(b) The probability that X lies between 1 and 2 is given by

18
 2

P[1 < X < 2] = ∫

2

1
f X ( x)dx =
4 2 2
27 ∫ 1
3 (
x − x 3
)
dx =
4  3 x4 
x − 
27  4 1
4   1  13
=  (8 − 4) − 1 −  =
27   4  27

2.32 A random variable X has the PDF

k (1 − x 4 ) −1 ≤ x ≤ 1
f X ( x) = 
0 otherwise

(a) Find the value of k.

(b) Find the CDF of X.
(c) Find P[ X < 12 ].
Solution:
(a) Since f X (x) is a valid PDF, we have that
1

∫
∞

−∞
1
( 
) x5   1   1 
f X ( x)dx = 1 = k ∫ 1 − x dx = k  x −  = k 1 −  −  − 1 + 
−1
4

 5  −1  5   5 
 2  8k 5
= k 2 −  = ⇒k=
 5 5 8

(b) The CDF of X is given by

0 x ≤ −1
 x

FX ( x ) = ∫
x

−∞
5 x
( 5  u5 
)5  x5 4 
f X (u )du = ∫ 1 − u du = u −  =   x − + 
8 −−1
4

8 5  −1  8  5 5
−1 ≤ x < 1

1 x ≥1

5  1 ( 12 ) 4  5  1 1
5
4
(c) P[ X < 12 ] =  − + =  − +  = 0.8086
8  2 5 5  8  2 160 5 

2.33 The PDF of a random variable X is given by

x 0 < x <1

f X ( x ) = 2 − x 1≤ x ≤ 2
0 otherwise


(a) Find the CDF of X.

(b) Find P[0.2 < X < 0.8].
19
 (c) Find P[0.6 < X < 1.2].

Solution: The PDF of X is as shown below:

f X (x)

(a) The CDF of X is given by

0 x<0
 x
∫ x = 0 udu 0 ≤ x <1

FX ( x) = P[ X ≤ x] =  1
 ∫ udu + ∫ (2 − u )du 
x
1≤ x < 2
 u =0 u =1 
1 x≥2
0 x<0
 2
x 0 ≤ x <1

= 2 2
2 x − x − 1 1≤ x < 2
 2

1 x≥2

0.64 0.04
(b) P[0.2 < X < 0.8] = FX (0.8) − FX (0.2) = − = 0.3
2 2

 1.44  0.36 1.36 0.36

(c) P[0.6 < X < 1.2] = FX (1.2) − FX (0.6) = 2.4 − − 1 − = − = 0.5
 2  2 2 2

2.34 The mileage (in thousands of miles) that car owners get with a certain type of tire is a
random variable X with PDF

 Ae − x / 20 x≥0
f X ( x) = 
0 otherwise

(a) Find the value of A.

20
 (b) Find the CDF of X.
(c) What is P[ X < 10] ?
(d) What is P[16 < X < 24] ?
Solution:
(a) Since f X (x) is a valid PDF, we have that
[ ]
∞
∞ ∞ 1
∫ f X ( x)dx = 1 = A∫ e − x / 20 dx = 20 A − e − x / 20 = 20 A ⇒ A =
−∞ x =0 0 20
(b) The CDF of X is given by
x<0
FX ( x) = ∫
x
f X (u )du =
1 x −u / 20
20 ∫ u =0
e du = − e −u / 20 [ ] x
0
0
= − x / 20
1 − e x≥0
u = −∞

−0.5
(c) P[ X < 10] = FX (10) = 1 − e = 0.3935
(d) P[16 < X < 24] = FX (24) − FX (16) = e −0.8 − e −1.2 = 0.4493 − 0.3012 = 0.1481

2.35 Assume that X is a continuous random variable with the following PDF:

0 x < 0.5
f X ( x) =  −2 ( x −0.5)
ke x ≥ 0.5

(a) Find the value of k.

(b) Find the CDF of X.
(c) What is P[ X < 1.5] ?
(d) What is P[1.2 < X < 2.4] ?

Solution:
(a) Since f X (x) is a valid PDF, we have that
∞
∞ ∞
− 2 ( x − 0.5)  e −2 x  ke1e −1 k
∫ f X ( x)dx = 1 = k ∫ dx = ke −  = = ⇒k=2
1
e
−∞ x = 0.5
 2  0.5 2 2

(b) The CDF of X is given by

x < 0.5
FX ( x ) = ∫
x x
f X (u )du = 2e1 ∫ e −2u du = −e1 e −2u [ ] x
0.5
0
= − 2 ( x − 0.5)
1 − e x ≥ 0.5
u = −∞ u = 0.5

(c) P[ X < 1.5] = FX (1.5) = 1 − e −2 = 0.8647

(d) P[1.2 < X < 2.4] = FX (2.4) − FX (1.2) = e −2( 0.7 ) − e −2(1.9) = e −1, 4 − e −3.8 = 0.2242

21
Chapter 3: Moments of Random Variables

Chapter Summary: This chapter discusses moments of random variables including the concepts
of expectation and variance, higher moments, conditional expectation, and the Chebyshev and
Markov inequalities.

Section 3.2: Expected Values

3.1 Find the mean and variance of the random variable whose triangular PDF is given in Figure
3.4.
f X (x)

1
2

x
0 2 4

Figure 3.4: PDF of X for Problem 3.1

Solution: The PDF of X is given by

 4x 0≤ x<2

f X ( x) = 1 − 4x 2≤ x<4
0 otherwise

Thus,
2 4
 x3   x2 x3 
E[ X ] = ∫ xf X ( x)dx = ∫ x( )dx + ∫ x(1 − )dx =   +  − 
∞ 2 4
x x
−∞ 4 4
0 2
 12  0  2 12  2
8  16 64   4 8 
= + − − −  = 2
12  2 12   2 12 
2 4
 x4   x3 x4 
E[ X ] = ∫ x f X ( x)dx = ∫ x ( )dx + ∫ x (1 − )dx =   +  − 
∞ 2 4
2 2 2 x 2 x
−∞ 4 4
0 2
 16  0  3 16  2
16  64 256   8 16  14
= + − − −  =
16  3 16   3 16  3
14 2
σ X2 = E[ X 2 ] − (E[ X ])2 = − 4 =
3 3

1
3.2 An insurance company has 1000 policies of men of age 50. The company’s estimate is that
the probability that a man of age 50 dies within one year is 0.02. What is the expected
number of claims that the company can expect from the beneficiaries of these men within
one year?

Solution: Let N be a random variable that denotes the number of claims that will be filed in one
year. Then the expected value of N is

E[ N ] = (1000)(0.02) = 20

3.3 A class has 20 students whose heights are as follows: There are 4 students with a height of
5.5 feet, there are 5 students with a height of 5.8 feet, there are 3 students with a height of 6.0
feet, there are 5 students with a height of 6.2 feet, and there are 3 students with a height of
6.5 feet. If a student is randomly selected from the class, what is his expected height?

Solution: Let X be the random variable that denotes the height of a student. Then the PMF of X
is given by
 204 x = 5.5
5
 20 x = 5.8
 3 x = 6.0
p X ( x) =  205
 20 x = 6.2
3 x = 6.5
 20
0 otherwise
Thus, the expected height of a student selected at random from the class is

E[ X ] = 5.5( 204 ) + 5.8( 205 ) + 6.0( 203 ) + 6.2( 205 ) + 6.5( 203 ) = 5.975

3.4 The time it takes a machine to perform an operation depends on the state it enters when the
start button is pushed. There are three known states: fast, moderate, and slow. When it is in
the fast state, it takes 2 minutes to perform the operation. When it is in the moderate state, it
takes 4 minutes to perform the operation. When it is in the slow state, it takes 7 minutes to
perform the operation. Studies indicate that the machine goes into the fast state when the start
button is pushed 60% of the time; it goes into the moderate state 25% of the time; and it goes
into the slow state 15% of the time. What is the expected time it takes the machine to
perform the operation?

Solution: Let T denote the time it takes the machine to perform an operation. Then the PMF of T
is given by

2
 0.60 t=2
0.25 t=4

pT (t ) = 
0.15 t=7
0 otherwise

Thus, the expected time it takes to perform a random operation is given by

E[T ] = 2(0.60) + 4(0.25) + 7(0.15) = 3.25

3.5 A student has three methods of solving a problem. It takes him 1 hour to solve the problem
using method A. It takes him 45 minutes to solve the problem using method B. It takes him
30 minutes to solve the problem using method C. His friends discovered that the student uses
method A 10% of the time, method B 40% of the time, and method C 50% of the time. What
is the expected time it takes the student to solve a problem?

Solution: Let X denote the time it takes the student to solve a problem. Then the PMF of X is
given by

0.1 x = 60
0.4 x = 45

p X ( x) = 
0.5 x = 30
0 otherwise

Thus, the expected time it takes to perform a random operation is given by

E[ X ] = 60(0.1) + 45(0.4) + 30(0.5) = 39

3.6 Consider the following game that involves tossing a fair die. If the outcome of a toss is an
even number, you win $2. If the outcome is 1 or 3, you lose $1. If the outcome is 5, you lose
$3. What are the expected winnings?
Solution: Let X denote the amount won in a game. Then, the PMF of X is given by

 16 x = −3
2
 x = −1
p X ( x) =  63
6 x=2
0 otherwise


Thus, the expected amount won in a game is given by

E[ X ] = 2( 63 ) − 1( 62 ) − 3( 16 ) = 1
6

3
3.7 Three vans are used to carry 45 students in the applied probability class to participate in a
state competition on random phenomena. The first van carried 12 students, the second van
carried 15 students, and the third van carried 18 students. Upon their arriving at the place
where the competition was taking place, one student was randomly selected from the entire
group to receive a gift certificate for a free lunch at a local restaurant. What is the expected
number of students in the van that carried the selected student?

Solution: Let K be a random variable that denotes the number of students in a van. Then the
PMF of K is given by

 12
45 k = 12
 15
 45 k = 15
p K (k ) =  18
 45 k = 18
0 otherwise


Thus, the expected number of students in the van that carried the selected student is given by

E[ K ] = 12(12
45 ) + 15( 45 ) + 18( 45 ) = 16.4
15 18

3.8 Find the expected value of a discrete random variable N whose PMF is given by
p N (n) = p (1 − p ) n −1 n = 1, 2, Κ

Solution: The expected value of N is

∞
E[ N ] = ∑ np (n) = p ∑ n(1 − p) n −1
n n =1
Now,
∞
1 1
∑ (1 − p)
n =1
n
=
1 − (1 − p )
−1 = −1
p
d ∞ ∞
d ∞

∑
dp n =1
(1 − p ) n = ∑ (1 − p ) n = −∑ n(1 − p ) n −1
n =1 dp n =1

Thus,
∞
d ∞ d 1   1 
E[ N ] = p ∑ n(1 − p) n −1 = − p ∑ (1 − p ) n = − p  − 1 = − p − 2 
n =1 dp n =1 dp  p   p 
1
=
p

3.9 Find the expected value of a discrete random variable K whose PMF is given by

4
 5 k e −5
p K (k ) = k = 0, 1, 2, Κ
k!
Solution: The expected value of K is given by
∞
5k ∞
5k ∞
5 k −1 ∞
5m
E[ K ] = ∑ kp K (k ) = e −5 ∑ k = e −5 ∑ = 5e −5 ∑ = 5e −5 ∑
k k =0 k! k =1 ( k − 1)! k =1 ( k − 1)! m = 0 m!
−5 5
= 5e e = 5

3.10 Find the expected value of a continuous random variable X whose PDF is given by
f X ( x ) = 2e −2 x x≥0

Solution: The expected value of X is given by

∞ ∞ ∞
E[ X ] = ∫ xf X ( x)dx = ∫ 2 xe −2 x dx = 2∫ xe −2 x dx
−∞ 0 0

Let u = x ⇒ du = dx, and let dv = e −2 x dx ⇒ v = − 12 e −2 x . Thus,

 xe − 2 x  ∞ 1 ∞   e −2 x  ∞ 1
−2 x

2  0 2 ∫0
E[ X ] = 2−  + e dx = −  =
   2  0 2

3.11 Assume that a random variable X takes on discrete values x1 , x 2 , Κ , x n with probabilities
p1 , p 2 , Κ , p n respectively. That is, P[ X = xi ] = pi , i = 1, 2, Κ , n. The “entropy” of
X, which is also defined as the amount of information provided by an observation of X, is
defined by
n
 1  n
H ( X ) = ∑ p i log  = −∑ pi log( pi )
i =1  pi  i =1

When the logarithm is taken base 2, the unit of entropy is bits. Let X represent the outcome
of a single roll of a fair die. What is the entropy of X in bits?

Solution: If the random variable X represents the outcome of a single roll of a fair die, then we
have that pi = 1 / 6, i = 1, 2, Κ , 6. Thus, the entropy of X is given by

n
 1  6 1 1
H ( X ) = ∑ p i log  = ∑ log(6 ) = 6  log(6 ) = log(6 ) = 2.5850
i =1  pi  i =1 6 6
Section 3.4: Moments of Random Variables and the Variance
3.12 A random variable X assumes two values 4 and 7 with probabilities p and q, respectively,
where q = 1 − p. Determine the mean and standard deviation of X.

5
Solution: The PMF of X is given by

p x=4

p X ( x) = 1 − p x=7
0 otherwise


Since the standard deviation is the square root of the variance, we have that

E[ X ] = ∑ xp X ( x) = 4 p + 7(1 − p) = 7 − 3 p
x

E[ X ] = ∑ x 2 p X ( x) = 16 p + 49(1 − p ) = 49 − 33 p
2

σ = E[ X 2 ] − (E[ X ])2 = 49 − 33 p − (7 − 3 p )2 = 9 p(1 − p )

2
X

σ X = 9 p (1 − p) = 3 p(1 − p )

3.13 Find the mean and variance of the discrete random variable X with the following PMF:
 25 x=3
p X ( x) =  3
5 x=6

Solution: The mean and variance are given by

E[ X ] = ∑ xp X ( x) = 3( 25 ) + 6( 35 ) = 24
5 = 4.8
x

E[ X 2 ] = ∑ x 2 p X ( x) = 9( 52 ) + 36( 35 ) = 126
5 = 25.2
x

σ = E[ X 2 ] − (E[ X ])2 = 25.2 − 23.04 = 2.16

2
X

3.14 Let N be a random variable with the following CDF:

0 n <1
0.2 1≤ n < 2

FN (n) = 0.5 2≤n<3
0.8 3≤n<4

1 n≥4

(a) What is the PMF of N?

(b) What is the expected value of N?
(c) What is the variance of N?
Solution:
(a) The PMF of N is given by

6
 0.2 n =1
0.3 n=2

p N (n) = 0.3 n=3
0.2 n=4

0 otherwise
(b) The expected value of N is
E[ N ] = 1(0.2) + 2(0.3) + 3(0.3) + 4(0.2) = 0.2 + 0.6 + 0.9 + 0.8 = 2.5
(c) The second moment of N is given by

E[ N 2 ] = 12 (0.2) + 2 2 (0.3) + 3 2 (0.3) + 4 2 (0.2) = 0.2 + 1.2 + 2.7 + 3.2 = 7.3

Thus, the variance of N is given by

σ N2 = E[ N 2 ] − (E[ N ])2 = 7.3 − (2.5)2 = 7.3 − 6.25 = 1.05

3.15 Let X be the random variable that denotes the outcome of tossing a fair die once.
(a) What is the PMF of X?
(b) What is the expected value of X?
(c) What is the variance of X?
Solution:
(a) The PMF of X is p X ( x) = 16 , x = 1, 2, Κ , 6
(b) The expected value of X is E[ X ] = 1
6 {1 + 2 + 3 + 4 + 5 + 6} = 216 = 72 = 3.5
(c) The variance of X is obtained as follows:

E[ X 2 ] = 1
6 {1
2
+ 2 2 + 32 + 4 2 + 52 + 6 2 } = 1
6 {1 + 4 + 9 + 16 + 25 + 36} = 91
6
2
91  7  91 49 182 − 147 35
σ X2 = E[ X 2 ] − (E[ X ])2 = −  = − = =
6 2 6 4 12 12

3.16 Suppose the random variable X has the PDF f X ( x) = ax 3 , 0 < x < 1.
(a) What is the value of a?
(b) What is the expected value of X?
(c) What is the variance of X?
(d) What is the value of m so that P[ X ≤ m] = 0.5 ?
Solution:
(a) The value of a is obtained as follows:
7
 1
∞ 1  x4  a
∫−∞ X = = ∫0 =   = ⇒a=4
3
f ( x ) dx 1 a x dx a
 4 0 4
(b) The expected value of X is given by
1
∞  x5 
1 a 4
E[ X ] = ∫ xf X ( x)dx = a ∫ x dx = a   = = = 0.8
4
−∞ 0
 5 0 5 5

(c) The variance of X is obtained as follows:

1
∞  x6 
1 a 4 2
E[ X ] = ∫ x f X ( x)dx = a ∫ x dx = a   = = =
2 2 5
−∞ 0
 6 0 6 6 3
2 2 − 1.92 0.08
σ X2 = E[ X 2 ] − (E[ X ])2 =
− 0.64 = = = 0.0267
3 3 3
(d) The value of m such that P[ X ≤ m] = 0.5 is obtained as follows:
x
x u 4 
x ax 4
FX ( x) = P[ X ≤ x] = ∫ f X (u )du = a ∫ u dx = a   =3
= x4
−∞ 0
 4 0 4
FX (m) = m 4 = 0.5 ⇒ m 2 = 0.5 = 0.7071
m = 0.7071 = 0.8409

3.17 A random variable X has the CDF

0 x <1

FX ( x) = 0.5( x − 1) 1≤ x < 3
1 x≥3


(a) What is the PDF of X?

(b) What is the expected value of X?
(c) What is the variance of X?
Solution:
d 0.5 1≤ x ≤ 3
(a) The PDF of X is given by f X ( x) = FX ( x ) = 
dx 0 otherwise
(b) The expected value of X is given by
3
∞ 3 x2   9 − 1
E[ X ] = ∫ xf X ( x)dx = 0.5∫ xdx = 0.5  = 0.5 =2
−∞ 1
 2 1  2 
(c) The variance of X is obtained as follows:

8
 3
∞ 3  x3   27 − 1 13
E[ X ] = ∫ x f X ( x)dx = 0.5∫ x dx = 0.5  = 0.5
2 2 2
=
−∞ 1
 3 1  3  3
13 13 − 12 1
σ X2 = E[ X 2 ] − (E[ X ])2 = − 4 = =
3 3 3

3.18 A random variable X has the PDF f X ( x) = x 2 / 9, 0 ≤ x ≤ 3. Find the mean, variance, and
third moment of X.

Solution: The mean, variance and third moment of X are obtained as follows:

3
∞ 1 3 1  x4  1  81 9
E[ X ] = ∫ xf X ( x)dx = ∫ x 3 dx =   =   =
−∞ 9 0 9  4 0 9  4  4
3
∞ 1 3 1  x5  1  243  27
E[ X ] = ∫ x f X ( x)dx = ∫ x 4 dx =   = 
2 2
=
−∞ 9 0 9  5  0 9  5  5
27 81 432 − 405 27
σ X2 = E[ X 2 ] − (E[ X ])2 = − = =
5 16 80 80
3
∞ 1 3 1  x6  1  729  81 27
E[ X ] = ∫ x f X ( x)dx = ∫ x 5 dx =   = 
3 3
= = = 13.5
−∞ 9 0 9  6  0 9  6  6 2

3.19 Assume the random variable X has the PDF f X ( x) = λe − λx , x ≥ 0. Find the third moment
of X, E[ X 3 ].

Solution: The third moment is given by

∞ ∞
E[ X 3 ] = ∫ x 3 f X ( x)dx = λ ∫ x 3 e −λx dx
−∞ 0

Let u = x 3 ⇒ du = 3 x 2 dx, and let dv = e − λx dx ⇒ v = − λ1 e − λx . Thus,

 x 3 e −λx  ∞ 3 ∞  ∞
E[ X ] = λ −
3 2 − λx 2 − λx
 + ∫0 x e dx  = 3∫0 x e dx
 λ 0 λ 

Let u = x 2 ⇒ du = 2 xdx, and let dv = e − λx dx ⇒ v = − λ1 e − λx . Thus,

 x 2 e −λx  ∞ 2 ∞  6 ∞
− λx − λx
E[ X ] = 3− + ∫  = ∫0 xe dx
3
 xe dx
 λ 0 λ 0
 λ
Let u = x ⇒ du = dx, and let dv = e − λx dx ⇒ v = − λ1 e − λx . Thus,

9
 ∞ ∞
6  xe −λx  1 ∞ −λx  6 ∞
− λx 6  e − λx 
λ  0 λ ∫0 ∫
E[ X ] = −
3
+ e dx = 2 e dx = 2 −
λ   λ 0 λ  λ  0
6
= 3
λ

3.20 Suppose X is a random variable with PDF f X (x), mean E[ X ], and variance σ X2 . If we
define the random variable Y = X 2 , determine the mean and variance of Y in terms of the
mean, variance, and other higher moments of X.

Solution: The mean and variance of Y are obtained as follows:

E[Y ] = E[ X 2 ] = σ X2 + (E[ X ])
2

E[Y 2 ] = E[ X 4 ]
{
σ Y2 = E[Y 2 ] − (E[Y ])2 = E[ X 4 ] − σ X2 + (E[ X ])2 }
2

( )
= E[ X 4 ] − σ X2
2
− 2σ X2 (E[ X ]) − (E[ X ])
2 4

3.21 The PDF of a random variable X is given by f X ( x) = 4 x(9 − x 2 ) / 81, 0 ≤ x ≤ 3. Find the
mean, variance, and third moment of X.

Solution: The mean, variance, and third moment of X are obtained as follows:

3
∞ 4 3
( 4 
) x5  4 
E[ X ] = ∫ xf X ( x)dx = ∫ 9 x 2 − x 4 dx = 3x 3 −  = 81 −
−∞ 81 0 81  5  0 81 
243  8
= = 1.6
5  5
3
2
−∞
∞
2 4 3
81 0
(
E[ X ] = ∫ x f X ( x)dx = ∫ 9 x 3 − x 5 dx = 
81  4
)
4  9x 4 x 6 
−  = 
4  729 729 
6  0 81  4
−
6 
=3

σ X2 = E[ X 2 ] − (E[ X ])2 = 3 − (1.6 )2 = 3 − 2.56 = 0.44

3
3
−∞
∞
3 4 3
81 0
(
E[ X ] = ∫ x f X ( x)dx = ∫ 9 x 4 − x 6 dx = 
81  5
)
4 9x5 x 7 
−  = 
4  2187 2187  216
7  0 81  5
− =
7  35

Section 3.5: Conditional Expectations

3.22 If the PDF of a random variable X is given by f X ( x) = 4 x(9 − x 2 ) / 81, 0 ≤ x ≤ 3, find the
conditional expected value of X, given that X ≤ 2.

Solution: The conditional expected value of X, given that X ≤ 2 can be obtained as follows;

10
 x
x 4 x
FX ( x) = ∫ f X (u )du = ∫ 9u − u 3 du = 
0 81 0
(
4  9u 2 u 4 
81  2
− 
4 0
)
0 x<0
1

[
=  18 x 2 − x 4
81
0≤ x<3 ]

1 x≥3
2
4  3 x5  4  32 
∫
2
xf X ( x)dx
4
∫0 (
2
2 4

)
9 x − x dx 81 3 x − 5 
 0 81  24 − 
5  88
= 81
−∞
E[ X | X ≤ 2] = = = =
P[ X ≤ 2] F X ( 2) 1
[72 − 16] 1
[72 − 16] 70
81 81
= 1.2571

3.23 The PDF of a continuous random variable X is given by f X ( x) = 2e −2 x , x ≥ 0. Find the

conditional expected value of X, given that X ≤ 3.

Solution: The conditional expected value of X, given that X ≤ 3 can be obtained as follows:
x x
FX ( x) = ∫ f X (u )du = ∫ 2e −2u du = − e −2u
0 0
[ ]
x
0 = 1 − e −2 x , x ≥ 0
3 3 3
∫ xf X ( x)dx ∫ 2 xe − 2 x dx 2 ∫ xe − 2 x dx
−∞
E[ X | X ≤ 3] = = =
0 0

P[ X ≤ 3] FX (3) 1 − e −6

Let u = x ⇒ du = dx, and let dv = e −2 x dx ⇒ v = − 12 e −2 x . Thus,

 xe − 2 x  3 1 3 
3
2∫ xe −2 x
dx = 2− 
2 0 2 0
+ ∫ e −2 x
dx 
1
= −3e −6 + − e −2 x
2
[ ]
3
0 =
1 − 7e − 6
2
0
 

Therefore,
3
2∫ xe −2 x dx 1 − 7e −6
E[ X | X ≤ 3] = = = 0.4925
0

1 − e −6 2(1 − e −6 )

3.24 The PDF of a random variable X is given by

0.1 30 ≤ x ≤ 40
f X ( x) = 
0 otherwise

Find the conditional expected value of X, given that X ≤ 35.

Solution: The conditional expected value of X, given that X ≤ 35 is given by
11
 35
35
 0.1x 2  35

E[ X | X ≤ 35] =
∫−∞ xf X ( x)dx = ∫30 0.1xdx =  2  30 = 1 {35 + 30} = 32.5
P[ X ≤ 35]
∫
35
0 .1dx [0.1x ]30
35
2
30

3.25 A fair die is rolled once. Let N denote the outcome of the experiment. Find the expected
value of N, given that the outcome is an even number.

Solution: Let Y denote the event that the outcome is an even number. Then the expected value of
N, given that the outcome is an even number is given by

∑ np N ( n) 1
{2 + 4 + 6} 2+4+6
E[ N | Y ] = n∈Y
= 6
= =4
∑p n∈Y
N ( n) 1
6 + 16 + 16 3

3.26 The life of a lightbulb in months is denoted by a random variable X with the PDF
f X ( x) = 0.5e −0.5 x , x ≥ 0. Find the conditional expected value of X, given that X ≤ 1.5.

Solution: The conditional expected value of X, given that X ≤ 1.5 is given by

1.5 1.5
∫ xf X ( x)dx ∫ 0.5 xe −0.5 x dx
−∞
E[ X | X ≤ 1.5] = =
0

P[ X ≤ 1.5] 1.5
0.5e −0.5 x dx
∫ 0

Now, ∫
1.5

0
[
0.5e −0.5 x dx = − e −0.5 x ]1.5
0 = 1 − e −0.75 . Also, let u = x ⇒ du = dx, and let

dv = e −0.5 x dx ⇒ v = − 01.5 e −0.5 x = −2e −0.5 x . Thus,

{[ }
1.5

∫
1.5
0.5 xe − 0.5 x
dx = 0.5 − 2 xe ]
− 0.5 x 1.5
0 + 2∫ e
1.5
− 0.5 x
dx = −1.5e − 0.75  e −0.5 x 
+ −  = 2 − 3.5e
− 0.75
0 0
 0.5  0
Therefore,
1.5
∫ 0.5 xe −0.5 x dx 2 − 3.5e −0.75
E[ X | X ≤ 1.5] = = = 0.6571
0
1.5
0.5e −0.5 x dx 1 − e −0.75
∫ 0

Sections 3.6 and 3.7: Markov and Chebyshev Inequalities

3.27 A random variable X has the PDF f X ( x) = 2e −2 x , x ≥ 0. Obtain an upper bound for
P[ X ≥ 1] using the Markov inequality.

Solution: The Markov inequality states that P[ X ≥ a ] ≤ 1a E[ X ]. Now, the mean of X is given by

12
 ∞ 1
E[ X ] = ∫ 2 xe −2 x dx =
0 2

1
Thus, P[ X ≥ 1] ≤ E[ X ] = .
2

3.28 A random variable X has the PDF f X ( x) = 2e −2 x , x ≥ 0. Obtain an upper bound for
P[| X − E[ X ] | ≥ 1].

Solution: To obtain an upper bound for P[| X − E[ X ] | ≥ 1] we use the Chebyshev inequality,
which states that

σ X2
P[| X − E[ X ] | ≥ a] ≤
a2

From previous examples we have that

∞ 1
E[ X ] = ∫ 2 xe − 2 x dx =
0 2
∞ 1
E[ X 2 ] = ∫ 2 x 2 e − 2 x dx =
0 2

Thus, the variance of X is given by σ X2 = E[ X 2 ] − (E[ X ]) = 12 − 14 = 14 . Therefore,

2

σ X2 1
P[| X − E[ X ] | ≥ 1] ≤ =
1 4

3.29 A random variable X has a mean 4 and variance 2. Use the Chebyshev inequality to obtain
an upper bound for P[| X − 4 | ≥ 2].

Solution: According to the Chebyshev inequality,

σ X2
P[| X − E[ X ] | ≥ a] ≤
a2

Thus, the upper bound for P[| X − 4 | ≥ 2] is given by

σ X2 2 1
P[| X − 4 | ≥ 2] ≤ 2
= =
2 4 2

13
3.30 A random variable X has the PDF
1
 1≤ x ≤ 4
f X ( x) =  3
0 otherwise

Use the Chebyshev inequality to estimate P[| X − 2.5 | ≥ 2].

Solution: The variance of X is obtained as follows:
4
∞ 1 4 1  x2  5
E[ X ] = ∫ xf X ( x)dx = ∫ xdx =   = = 2.5
−∞ 3 1 3  2 1 2
4
∞ 1 4 1  x3 
E[ X ] = ∫ x f X ( x)dx = ∫ x 2 dx =   = 7
2 2
−∞ 3 1 3  3 1
3
σ X2 = E[ X 2 ] − (E[ X ])2 = 7 − (2.5)2 = 7 − 6.25 = 0.75 =
4

σ X2 ( 34 ) 3
Thus, P[| X − 2.5 | ≥ 2] ≤ 2
= = = 0.1875.
2 4 16

14
Chapter 4: Special Probability Distributions

Chapter Summary: This chapter discusses special random variables and their distributions.
These include the Bernoulli distribution, binomial distribution, geometric distribution, Pascal
distribution, hypergeometric distribution, Poisson distribution, exponential distribution, Erlang
distribution, uniform distribution, and normal distribution.
Section 4.3: Binomial Distribution
4.1 Suppose four dice are tossed. What is the probability that at most one 6 appears?

Solution: The probability of a six on a toss of a die is p = 16 . Let N (4) be a random variable that
denotes the number of sixes that appear in tossing the 4 dice. Since the outcome of each die is
independent of the outcome of any other die, N (4) has a binomial distribution. Thus, the PMF
of N (4) is given by
n 4−n
4 n 4−n  4  1   5 
p N ( 4 ) (n) =   p (1 − p ) =      n = 0, 1, 2, 3, 4
n  n  6   6 

Therefore, the probability that at most one 6 appears is given by

0 4 1 3 4 3
 4  1   5   4  1  5 5  1  5 
P[ N (4) ≤ 1] = p N ( 4) (0) + p N ( 4) (1) =      +      =   + 4  
 0  6   6  1  6  6 6  6  6 
3
 5   5 4   125  3 
=   + =   = 0.86806
 6   6 6   216  2 

4.2 An equipment consists of 9 components, each of which will independently fail with a
probability of p. If the equipment is able to function properly when at least 6 of the
components are operational, what is the probability that it is functioning properly?

Solution: Let K (9) be a random variable that denotes the number of operational components out
of the 9 components. Then K (9) has a binomial distribution with the PMF

9 
p K ( 9) (k ) =  (1 − p ) k p 9− k k = 0, 1, Κ , 9
k 

Let A denote the event that at least 6 of the components are operational. Then the probability of
event A is given by

1
 9 9
9 
P[ A] = P[ K (9) ≥ 6] = ∑ p K ( 9) (k ) = ∑  (1 − p ) k p 9−k
k =6 k =6  k 

9! 9! 9! 9!
= (1 − p ) 6 p 3 + (1 − p ) 7 p 2 + (1 − p ) 8 p 1 + (1 − p) 9 p 0
6!3! 7!2! 8!1! 9!0!
= 84 p 3 (1 − p ) 6 + 36 p 2 (1 − p) 7 + 9 p(1 − p ) 8 + (1 − p) 9

4.3 A fair coin is tossed three times. Let the random variable X denote the number of heads that
turn up. Determine the mean and variance of X.
Solution: If we assume that the outcomes are independent, then the PMF of X is given by
x 3− x 3
3   3  1   1   3  1 
p X ( x) =   p x (1 − p ) 3− x =      =    x = 0, 1, 2, 3
 x  x  2   2   x  2 

Thus, the mean and variance of X are given by

3
E[ X ] = 3 p =
2
3
σ X2 = 3 p (1 − p) =
4

4.4 A certain student is known to be late to the Signals and Systems class 30% of the time. If the
class meets four times a week, find
(a) the probability that the student is late for at least three classes in a given week.
(b) the probability that the student will not be late at all during a given week.

Solution: Let Y (4) be a random variable that denotes the number of times in the 4 meeting times
a week that the student is late. Then Y (4) has a binomial distribution with a success probability
p = 0.3 and the PMF is given by

4 4 
pY ( 4) ( y ) =   p y (1 − p ) 4− y =  (0.3) (0.7 )
4− y
y = 0, 1, 2, 3, 4
y

 y  y

(a) The probability that the student is late for at least three classes in a given week is given
by

4  4
P[Y (4) ≥ 3] = pY ( 4) (3) + pY ( 4 ) (4) =  (0.3) (0.7 ) +  (0.3) (0.7 )
3 1 4 0

3   4
= 4(0.3) (0.7 ) + (0.3) = 0.0837
3 4

(b) The probability that the student will not be late at all during a given week is given by

2
 4
P[Y (4) = 0] = pY ( 4) (0) =  (0.3) (0.7 ) = (0.7 ) = 0.2401
0 4 4

0
 

4.5 A multiple-choice exam has 6 problems, each of which has 3 possible answers. What is the
probability that John will get 4 or more correct answers by just guessing?

Solution: Let N (6) be a random variable that denotes the number of correct answers that John
gets out of the 6 problems. Since each problem has three possible answers, the probability of
getting a correct answer to a question is p = 13 . If we assume that John’s performance is
independent from one question to another, then N (6) has a binomial distribution, and the PMF
is given by

n 6− n
6  6  1   2 
p N ( 6 ) (n) =   p n (1 − p ) 6− n =      n = 0, 1, Κ , 6
n  n  3   3 

Thus, the probability that John will get 4 or more correct answers by just guessing is given by

P[ N (6) ≥ 4] = p N ( 6) (4) + p N ( 6 ) (5) + p N ( 6) (6)

4 2 5 1 6
 6  1   2   6  1   2   6  1   2 
=      +      +      =
0
( )
15 × 2 2 + (6 × 2 ) + 1
 4  3   3   5  3   3   6  3   3  36
73
= = 0.1001
729

4.6 A block of 100 bits is to be transmitted over a binary channel with a probability of bit error
of p = 0.001. What is the probability that 3 or more bits are received in error?

Solution: Let K (100) be a random variable that denotes the number of bits among the 100 bits
that are received in error. Given that the probability of bit error is p = 0.001 and the channel
treatment of each bit is independent of other bits, K (100) has a binomial distribution with the
PMF

100  k 100 
p K (100) (k ) =   p (1 − p )100− k =  (0.001) k (0.999 p )100− k k = 0, 1, Κ , 100
 k   k 
Thus, the probability that 3 or more bits are received in error is given by

3
P[ K (100) ≥ 3] = 1 − P[ K (100) < 3] = 1 − {p K (100) (0) + p K (100) (1) + p K (100) (2)}
100  100  100  
= 1 −  (0.001)0 (0.999)100 +  (0.001)1 (0.999)99 +  (0.001)2 (0.999)98 
 0  1   2  
{
= 1 − (0.999)100 + 100(0.001)(0.999) 99 + 4950(0.001) 2 (0.999) 98 }
= 0.00015

4.7 An office has 4 phone lines. Each is busy about 10% of the time. Assume that the phone lines
act independently.
(a) What is the probability that all 4 phones are busy?
(b) What is the probability that 3 of the phones are busy?
Solution: Let Y (4) be a random variable that denotes the number of busy phone lines among the
4 phone lines. Since each phone line acts independently and the probability that a phone line is
busy is 0.1, Y (4) has a binomial distribution with PMF

4 4 
pY ( 4) ( y ) =   p y (1 − p ) 4− y =  (0.1) (0.9 )
4− y
y = 0, 1, 2, 3, 4
y

 y  y

(a) The probability that all 4 phones are busy is given by

P[Y (4) = 4] = pY ( 4) (4) = (0.1) = 0.0001

4

(b) The probability that 3 of the phones are busy is given by

4
P[Y (4) = 3] = pY ( 4 ) (3) =  (0.1) (0.9 ) = 4 (0.1) (0.9) = 0.0036
3 1 3

3
 

4.8 The laptops made by the XYZ Corporation have a probability of 0.10 of being defective as
they come out of the assembly line. The ABC company has purchased 8 of these laptops for
office use.
(a) What is the PMF of K, the number of defective laptops out of the 8 that the ABC
company purchased?
(b) What is the probability that at most one laptop is defective out of the 8?
(c) What is the probability that exactly 1 laptop is defective out of the 8?
(d) What is the expected number of defective laptops out of the 8?
Solution:
(a) K is a binomially distributed random variable, and its PMF is given by
8  8 
p K (k ) =   p k (1 − p ) 8− k =  (0.1) (0.9)
8−k
y = 0, 1, Κ , 8
k

k  k 

4
 (b) The probability that at most one laptop is defective out of the 8 is given by
8  8
P[ K ≤ 1] = p K (0) + p K (1)] =  (0.1) (0.9 ) +  (0.1) (0.9)
0 8 1 7

0
  1
 
= (0.9 ) + 8(0.1)(0.9 ) = 1.7(0.9) = 0.8131
8 7 7

(c) The probability that exactly 1 laptop is defective out of the 8 is given by
8
P[ K = 1] = p K (1)] =  (0.1) (0.9) = 8(0.1)(0.9 ) = 0.3826
1 7 7

1 
(d) The expected number of defective laptops out of the 8 is given by E[ K ] = 8 p = 0.8.

4.9 On the average, 25% of the products manufactured by a certain company are found to be
defective. If we select four of these products at random and denote the number of the four
products that are defective by the random variable X, determine the mean and variance of X.

Solution: The probability that a product is defective is p = 0.25. Thus, the PMF of X is

4  4
p X ( x) =   p x (1 − p ) 4− x =  (0.25) (0.75)
4− x
x = 0, 1, 2, 3, 4
x

 x  x

Thus, the mean and variance of X are given by

E[ X ] = 4 p = 1
σ X2 = 4 p (1 − p ) = 0.75

4.10 Five fair coins are tossed. Assuming that the outcomes are independent, find the PMF of
the number of heads obtained in the experiment.

Solution: Let N (5) be a random variable that denotes the number of heads in the 5 tosses. Then
N (5) is a binomial random variable and its PMF is given by

n 5− n 5
5  n 5− n  5  1   1   5  1 
p N ( 5) (n) =   p (1 − p ) =      =   
      n = 0, 1, 2, 3, 4, 5
n  n  2   2   n  2 

4.11 A company makes gadgets that it sells in packages of eight. It has been found that the
probability that a gadget made by the company is defective is 0.1 independently of other
gadgets. If the company offers a money-back guarantee for any package that contains more
than one defective gadget, what is probability that the person that bought a given package
will be refunded?

5
Solution: Let K (8) be a random variable that denotes the number of gadgets in a package of 8
that are defective. Since the probability that a gadget is defective is 0.1 independently of other
gadgets, K (8) is a binomial random variable and its PMF is given by

8  8 
p K (8) (k ) =   p k (1 − p) 8− k =  (0.1) (0.9 )
8− k
k = 0, 1, Κ , 8
k

k  k 

Let A denote the event that the person that bought a given gadget will be refunded. Thus, we
have that

{
P[ A] = P[ K (8) > 1] = 1 − P[ K (8) ≤ 1] = 1 − {p K (8) (0) + p K (8) (1)} = 1 − (0.9) 8 + 8(0.1)(0.9) 7 }
= 0.1869

4.12 At least 10 of 12 people in a jury are required to find a person guilty before the person can
be convicted. Assume that each juror acts independently of other jurors and each juror has
a probability of 0.7 of finding a person guilty. What is the probability that a person is
convicted?

Solution: Let N (12) be a random variable that denotes the number of jurors among the 12
people in the jury that find the person guilty. Since each juror acts independently of other jurors
and each juror has a probability p = 0.7 of finding a person guilty, N (12) is a binomial random
variable and its PMF is given by

12  12 
p N (12) (n) =   p n (1 − p)12−n =  (0.7 ) (0.3)
12 − n
n = 0, 1, Κ , 12
n

n n

Let B denote the event that a person is convicted. Then the probability of event B is given by

P[ B] = P[ N (12) ≥ 10] = p N (12) (10) + p N (12) (11) + p N (12) (12)

12!
= (0.7 )10 (0.3)2 + 12! (0.7 )11 (0.3)1 + 12! (0.7 )12 (0.3)0
10!2! 11!1! 12!0!
= 66(0.7 ) (0.3) + 12(0.7 ) (0.3) + (0.7 ) = 0.2528
10 2 11 12

4.13 A radar system has a probability of 0.1 of detecting a certain target during a single scan.
Find the probability that the target will be detected
(a) at least two times in four consecutive scans.
(b) at least once in 20 consecutive scans.

Solution: Let K (n) be a random variable that denotes the number of target detections in n
consecutive scans. If we assume that the outcomes of the scans are independent of each other and

6
the probability of a target detection in each scan is p = 0.1, then K (n) is a binomial random
variable and its PMF is given by
n n
p K ( n ) (k ) =   p k (1 − p ) n − k =  (0.1) (0.9 )
n−k
k = 0, 1, Κ , n
k

 
k  
k
(a) The probability that the target will be detected at least 2 times in 4 consecutive scans is
given by
P[ K (4) ≥ 2] = p K ( 4) (2) + p K ( 4) (3) + p K ( 4) (4)
4!
= (0.1)2 (0.9)2 + 4! (0.1)3 (0.9)1 + 4! (0.1)4 (0.9)0
2!2! 3!1! 4!0!
= 6(0.1) (0.9 ) + 4(0.1) (0.9 ) + (0.1) = 0.04906
2 2 3 4

(b) The probability that the target will be detected at least once in 20 consecutive scans is
given by
P[ K (20) ≥ 1] = 1 − P[ K (20) < 1] = 1 − P[ K (20) = 0] = 1 − p K ( 20) (0) = 1 − (0.9) 20 = 0.8784

4.14 A machine makes errors in a certain operation with probability p. There are two types of
errors: type A error and type B error. The fraction of errors of type A is a, and the fraction
of errors of type B is 1 − a.
(a) What is the probability of k errors in n operations?
(b) What is the probability of k A type A errors in n operations?
(c) What is the probability of k B type B errors in n operations?
(d) What is the probability of k A type A errors and k B type B errors in n operations?

Solution: Since the probability that the machine makes errors in a certain operation with
probability p and the fraction of errors of type A is a, the probability of type A error is p A = pa,
and the probability of a type B error is p B = p (1 − a ). Let K (n) denote the number of errors in n
operations, K A (n) the number of type A errors in n operations, and K B (n) the number of type B
errors in n operations. Then the PMFs of K (n), K A (n) and K B (n) have the binomial
distribution.

(a) The probability of k errors in n operations is given by

n
p K ( n ) (k ) =   p k (1 − p) n −k k = 0, 1, Κ , n
k 

(b) The probability of k A type A errors in n operations is given by

n  n 
p K A ( n ) (k A ) =   p Ak A (1 − p A ) n − k A =  (ap ) k A (1 − ap) n − k A k A = 0, 1, Κ , n
kA  kA 

7
 (c) The probability of k B type B errors in n operations is given by

n  n 
p K B ( n ) (k B ) =   p Bk B (1 − p B ) n − k B =  {p(1 − a )} B {1 − p(1 − a )} B
n− k
k B = 0, 1, Κ , n
k

kA   kB 

(d) The probability of k A type A errors and k B type B errors in n operations is given by

 n 
P[ K A (n) = k A , K B (n) = k B ] =  (ap )k A {p (1 − a )}k B (1 − p )n −k A −k B n − k A − kB ≥ 0
kA kB 

4.15 Studies indicate that 40% of marriages end in divorce, where it is assumed that divorces are
independent of each other. Out of 10 married couples, determine the following
probabilities:
(a) That only the Arthurs and the Martins will stay married
(b) That exactly 2 of the 10 couples will stay married.
Solution:
(a) The event that only the Arthurs and the Martins will stay married is a specific event
whose probability of occurrence is the probability that these two couples remain married
while the other 8 couples get divorced. Thus, the probability of this event is

p = (0.4 ) × (0.6 ) = 0.00024

8 2

(b) If N (10) denotes the number of married couples that stay married, then the probability
that exactly 2 of the 10 couples will stay married is given by

10  10!
P[ N (10) = 2] = p N (10) (2) =  (0.6) 2 (0.4) 8 = (0.6)2 (0.4)8 = 0.01062
2 8!2!

4.16 A car has five traffic lights on its route. Independently of other traffic lights, each traffic
light turns red as the car approaches the light (and thus forces the car to stop at the light)
with a probability of 0.4.
(a) Let K be a random variable that denotes the number of lights at which the car stops.
What is the PMF of K?
(b) What is the probability that the car stops at exactly two lights?
(c) What is the probability that the car stops at more than two lights?
(d) What is the expected value of K?
Solution:
(a) K is a binomial B (5, 0.4) random variable whose PMF is given by

8
 5  5 
p K (k ) =   p k (1 − p ) 5−k =  (0.4) k (0.6) 5−k k = 0, 1, Κ , 5
k  k 
(b) The probability that the car stops at exactly two lights is given by
5
p K (2) =  (0.4) 2 (0.6) 3 = 10(0.4) 2 (0.6) 3 = 0.3456
 2
(c) The probability that the car stops at more than two lights is given by

P[ K > 2] = p K (3) + p K (4) + p K (5) = 10(0.4) 3 (0.6) 2 + 5(0.4) 4 (0.6) + (0.4) 5 = 0.31744

(d) The expected value of K is E[ K ] = 5 p = 5(0.4) = 2.0.

4.17 In a class of 18 boys and 12 girls, boys have a probability of 1/3 of knowing the answer to a
typical question that the teacher asks and girls have a probability of 1/2 of knowing the
answer to the question. Assume that each student acts independently. Let K be a random
variable that denotes the number of students who know the answer to a question that the
teacher asks in class. Determine the following:
(a) The PMF of K
(b) The mean of K
(c) The variance of K.
Solution: Since the total number of students is 30, the probability that a randomly selected
student is a boy is p B = 18
30 = 0.6,
and the probability that a randomly selected student is a girl is
pG = 1 − p B = 0.4. Thus, the probability p that a randomly selected knows the answer is
1 1
p = × (0.6) + × (0.4) = 0.4
3 2
Thus, K is a binomially distributed random variable; that is, K ~ B (30, 0.4).
(a) The PMF of K is
 30   30 
p K (k ) =   p k (1 − p ) 30−k =  (0.4) k (0.6) 30− k k = 0, 1, Κ , 30
k  k 
(b) The mean of K is E[ K ] = 30 p = 30(0.4) = 12

(c) The variance of K is σ K2 = 30 p (1 − p) = 30(0.4)(0.4) = 7.2.

4.18 A bag contains 2 red balls and 6 green balls. A ball is randomly selected from the bag, its
color is noted and the ball is put back into the bag, which is then thoroughly mixed.
Determine the probability that in 10 such selections a red ball is selected exactly 4 times
using
(a) the binomial distribution
(b) the Poisson approximation to the binomial distribution

9
Solution: Since the balls are drawn with replacement, the probability that a red ball is drawn is
given by p R = 82 = 0.25 and the probability that a green ball is drawn is 0.75. Let K denote the
number of times a red ball is drawn in 10 trials. Then K ~ B(10, 0.25).
(a) Using the binomial distribution we have that

10  10!
P[ K = 4] =  (0.25) 4 (0.75) 6 = (0.25) 4 (0.75) 6 = 0.1460
4 4!6!
(b) Using the Poisson approximation to the binomial distribution we have that

λ4
P[ K = 4] = e −λ
4!

where λ = 10 p B = 2.5. Thus, we obtain

(2.5) 4 −2.5 (2.5) 4 −2.5

P[ K = 4] = e = e = 0.1336
4! 24

4.19 Ten balls are randomly tossed into 5 boxes labeled B1 , B2 , Κ , B5 . Determine the
following probabilities:
(a) Each box gets 2 balls
(b) Box B3 is empty
(c) Box B2 has 6 balls.

Solution: The probability that a ball lands in box Bi , i = 1, 2, Κ , 5, is p i = 1 / 5 = 0.2.

(a) The probability that each box gets 2 balls is given by



10  2
 p1 ( ) 5
=
10!
(0.2)10 = 0.0116
 2 2 2 2 2 (2!) 5

(b) The probability that B3 is empty is given by

10  0 10 
  p 3 (1 − p3 )10 =  (0.2) 0 (0.8)10 = (0.8)10 = 0.1074
0 0

(c) Box B2 has 6 balls is given by

10  6 10  10!

  p 2 (1 − p 2 ) 4 =  (0.2) 6 (0.8) 4 = (0.2) 6 (0.8) 4 = 0.0055
6
  6
  6! 4!

10
Section 4.4: Geometric Distribution
4.20 A fair die is rolled repeatedly until a 6 appears.
(a) What is the probability that the experiment stops at the fourth roll?
(b) Given that the experiment stops at the third roll, what is the probability the sum of all
the three rolls is at least 12?
Solution: Let K be the random variable that denotes the number of times that a fair die is rolled
repeatedly until a 6 appears. The probability that a 6 appears on any roll is p = 16 . If we assume
that the outcome of each roll is independent of the outcomes of the other rolls, then K is a
geometric random variable and its PMF is

k −1
k −1  1  5 
p K (k ) = p (1 − p ) =    k = 1, 2, Κ
 6  6 
(a) The probability that the experiment stops at the fourth roll is given by

3
k −1  1  5  125
P[ K = 4] = p K (4) = p (1 − p) =    = = 0.0964
 6  6  1296

(b) Let A denote the event that the experiment stops at the third roll and B the event that the
sum of the three rolls is at least 12. Then
P[ A ∩ B ]
P[ B | A] =
P[ A]
The event A ∩ B is the event that the sum of the first two rolls is at least 6 with neither one
being a 6 and the third roll is a 6. That is, the event space of A ∩ B is as follows:

(1, 5, 6), (2, 4, 6) , (2, 5, 6), (3, 3, 6), (3, 4, 6) 

 
A ∩ B = (3, 5, 6), (4, 2, 6) , (4, 3, 6), (4, 4, 6), (4, 5, 6)
(5,1, 6), (5, 2, 6) , (5, 3, 6), (5, 4, 6), (5, 5, 6) 
 
Since the sample space of an experiment that consists of rolling a die three times contains
6 × 6 × 6 = 216 equally likely sample points, we have that P[ A ∩ B] = 216 15
. Also,
2
 1  5  25
P[ A] = p (1 − p ) =    =
3

 6  6  216
Thus,
P[ A ∩ B] 15
15
P[ B | A] = = 216
25
= = 0.6
P[ A] 216
25

4.21 A certain door can be opened by exactly one of six keys. If you try the keys one after
another, what is the expected number of keys you will have to try before the door is
opened?

11
Solution: The probability that a key opens the door on any trial is p = 16 . Let K be a random
variable that denotes the number of trials until the door is opened. The PMF of K is given by

p K (k ) = Ap (1 − p) k −1 k = 1, 2, Κ , 6

where A is a normalizing factor required to make the PMF sum to 1. Specifically, we have that

1 − (1 − p) 6 
{ }
6 5

∑ Ap (1 − p ) k −1
= Ap ∑ Ap (1 − p ) m
= 1 = Ap 
1 − (1 − p )
 = A 1 − (1 − p )
6

k =1 m =0  
1
⇒ A=
1 − (1 − p ) 6

Thus, we have the following truncated geometric distribution:

p(1 − p ) k −1
p K (k ) = Ap (1 − p ) k −1 = k = 1, 2, Κ , 6
1 − (1 − p ) 6

The expected number of keys that we have to try before the door is opened is given by

6 6
p
E[ K ] = ∑ kp K (k ) = ∑ k (1 − p) k −1

k =1 1 − (1 − p) 6 k =1

 16   5 5
2
5
3
5
4
 5  
5

=  1 + 2  + 3  + 4  + 5  + 6   = 2.8535
1 − ( 6 )  
5 6
6 6 6 6  6  

4.22 A box contains R red balls and B blue balls. An experiment is conducted with the balls as
follows. A ball is randomly selected from the box, its color is noted and the ball is put back
into the box. The process is repeated until a blue ball is selected.
(a) What is the probability that the experiment stops after exactly n trials?
(b) What is the probability that the experiment requires at least k trials before it stops?
Solution: Since the selection is done with replacement, the probability of success in any trial is
B
p=
B+R
Let X be a random variable that denotes the number of trials until a blue ball is selected. Then the
PMF of X is given by
p X ( x) = p(1 − p ) x −1 x = 1, 2, Κ
(a) The probability that the experiment stops after exactly n trials is
n −1
n −1  B  R 
P[ X = n] = p X (n) = p (1 − p ) =   n = 1, 2, Κ
 B + R  B + R 
12
 (b) The probability that the experiment requires at least k trials before it stops is

k −1 k −2
P[ X ≥ k ] = 1 − P[ X < k ] = 1 − ∑ p (1 − p) n −1 = 1 − p ∑ (1 − p ) m = 1 −
[
p 1 − (1 − p ) k −1 ]
n =1 m=0 1 − (1 − p )
k −1
 R 
= (1 − p) k −1 =  
B+ R

4.23 Twenty percent of the population of a particular city wear glasses. If you randomly stop
people from that city, determine the following probabilities:
(a) It takes exactly 10 tries to get a person who wears glasses
(b) It takes at least 10 tries to get a person who wears glasses.
Solution: Let K denote the number of trials until we find a person who wears glasses. Since the
probability of success is p = 0.2, the PMF of K is given by
p K (k ) = p (1 − p ) k −1 = (0.2)(0.8) k −1 k = 1, 2, Κ

(a) The probability that it takes exactly 10 tries to get a person who wears glasses is

P[ K = 10] = p K (10) = 0.2(0.8) 9 = 0.02684

(b) The probability that it takes at least 10 tries to get a person who wears glasses is
9
P[ K ≥ 10] = 1 − P[ K < 10] = 1 − ∑ p(1 − p ) k −1 = (1 − p ) 9 = (0.8) 9 = 0.1342
k =1

4.24 A student is planning to take the scholastic aptitude test (SAT) exam to gain admission to a
top college. She hopes to keep taking the exam until she gets a score of at least 2000 and
then she will stop. Her score in any of the exams is uniformly distributed between 800 and
2200, and her score in one exam is independent of her score in any other exam.
(a) What is the probability that she reaches her goal of scoring at least 2000 points in any
exam?
(b) What is the PMF of the number of times she will take the exam before reaching her
goal?
(c) What is the expected number of times she will take the exam?
Solution: Since her score in any exam is uniformly distributed between 800 and 2200, the PDF
of X, her score in any exam, is given by

 1400
1
800 ≤ x ≤ 2200
f X ( x) = 
0 otherwise

13
 (a) The probability that she reaches her goal of scoring at least 2000 points in any exam is
given by
2200
2200 1 2200  x  200 1
p = P[ X ≥ 2000] = ∫ f X ( x)dx = ∫ dx =   = = = 0.1428
x = 2000 =
1400 x 2000
1400  2000 1400 7

(b) Let K denote the number of times that she will take the exam before reaching her goal.
Then K is a geometrically distributed random variable whose PMF is given by
k −1
k −1 16
p K (k ) = p (1 − p ) =   k = 1, 2, Κ
77

(c) The expected number of times she will take the exam is E[ K ] = 1
p = 7.

Section 4.5: Pascal Distribution

4.25 Sam is fond of taking long-distance trips. During each trip his car has a tire failure in each
100-mile stretch with a probability of 0.05. He recently embarked on an 800-mile trip and
took two spare tires with him on the trip.
(a) What is the probability that the first change of tire occurred 300 miles from his starting
point?
(b) What is the probability that his second change of tire occurred 500 miles from his
starting point?
(c) What is the probability that he completed the trip without having to change tires?
Solution: Let K be a random variable that denotes the number of 100-mile units Sam travels
before a tire fails. Then the PMF of K is given by

p K (k ) = p (1 − p ) k −1 = 0.05(0.95)
k −1
k = 1, 2, Κ

(a) The probability that the first change of tire occurred 300 miles (or the 3rd 100-mile unit)
from his starting point is given by

P[ K = 3] = p K (3) = 0.05(0.95) = 0.04512

2

(b) Let K r denote the number of 100-mile units he travels until the rth tire change. Then K r
is an rth-order Pascal random variable with PMF

 k − 1
p K r (k ) =  (0.05) r (0.95)k −r k = r , r + 1, r + 2, Κ ; r = 1, 2, Κ , k
 r − 1

Thus, the probability that his second change of tire occurred 500 miles (or the 5th 100-
mile unit) from his starting point is given by
14
  5 − 1  4
P[ K 2 = 5] = p K 2 (5) =  (0.05) 2 (0.95)3 =  (0.05) 2 (0.95)3 = 4(0.05) 2 (0.95)3 = 0.00857
 2 − 1 1 

(c) The probability that he completed the trip without having to change tires is given by

P[ K > 8] = 1 − P[ K ≤ 8] = 1 − 0.05∑ (0.95)

8
k −1
=1−
{ }
0.05 1 − (0.95) 8
= (0.95) 8 = 0.6634
k =1 1 − 0.95

4.26 Six applicants for a job were all found to be qualified by the company. The company then
ranked these applicants in a priority order. Three positions are open. Past experience has
shown that 20% of applicants who are offered this kind of position by the company do not
accept the offer. What is the probability that the sixth-ranked applicant will be offered one
of the positions?

Solution: The probability of success, which is the probability that an applicant who is offered a
job actually accepts it, is p = 0.2. Let K r be a random variable that denotes the number of
candidates offered a job up to an including the rth candidate who accepts the job. Then K r is the
rth-order Pascal random variable with PMF

 k − 1
p K r (k ) =  (0.2) r (0.8)k − r k = r , r + 1, r + 2, Κ ; r = 1, 2, Κ , k
 r −1

The probability that the sixth-ranked applicant will be offered one of the 3 positions is the
probability that the 6th candidate is either the first or second or third applicant to accept the job.
This probability, Q, is given by

Q = P[(K1 = 6) ∪ (K 2 = 6 ) ∪ (K 3 = 6))] = P[ K1 = 6] + P[ K 2 = 6] + P[ K 3 = 6]
 5 5
= (0.2)(0.8) +  (0.2) 2 (0.8) +  (0.2) 3 (0.8)
5 4 3

1   2
= (0.2)(0.8) + 4(0.2) 2 (0.8) + 10(0.2) 3 (0.8) = 0.0463
5 4 3

4.27 Twenty percent of the population of a particular city wear glasses. If you randomly stop
people from that city, determine the following probabilities:
(a) It takes exactly 10 tries to get the third person who wears glasses
(b) It takes at least 10 tries to get the third person who wears glasses.

Solution: Let K r be a random variable that denotes the number of trials up to and including the
trial that results in the rth person who wears glasses. Since the probability of successes on any
trial is p = 0.2, the PMF of K r is given by

15
  k − 1
p K r (k ) =  (0.2) r (0.8)k − r k = r , r + 1, r + 2, Κ ; r = 1, 2, Κ , k
 r − 1 

(a) The probability that it takes exactly 10 tries to get the third person who wears glasses is
given by

10 − 1 9
P[ K 3 = 10] = p K 3 (10) =  (0.2) 3 (0.8)7 =  (0.2) 3 (0.8)7 = 36(0.2) 3 (0.8)7 = 0.0604
 3 −1   2

(b) The probability that it takes at least 10 tries to get the third person who wears glasses is
given by

9
 k − 1 9
 k − 1
P[ K 3 ≥ 10] = 1 − P[ K 3 < 10] = 1 − ∑  (0.2) 3 (0.8) = 1 − ∑  (0.2) 3 (0.8)
k −3 k −3

k =3  3 − 1  k =3  2 
 2  3  4 5 3
 (0.2) (0.8) +  (0.2) (0.8) +  (0.2) (0.8) +  (0.2) (0.8) 
3 0 3 1 3 2 3

 2   2  2  2 
=1−  
 6 7 8  6
+  (0.2) (0.8) +  (0.2) (0.8) +  (0.2) (0.8)
3 4 3 5 3
  2  2  2 
 
{
= 1 − (0.2) 3 1 + 3(0.8) + 6(0.8) 2 + 10(0.8) 3 + 15(0.8) 4 + 21(0.8) 5 + 28(0.8) 6 }
= 1 − (0.2) 3 (32.7253) = 0.7382

4.28 The probability of getting a head in a single toss of a biased coin is q. In an experiment that
consists of repeated tosses of the coin, what is the probability that the 18th head occurs on
the 30th toss?

Solution: Let N k be a random variable that denotes the number of tosses up to and including the
toss that results in the kth head. Then N k is a kth-order Pascal random variable with the PMF

 n − 1 k
p N k (n) =  q (1 − q )n− k n = k , k + 1, k + 2, Κ ; k = 1, 2, Κ , n
 k − 1 

The probability that the 18th head occurs on the 30th toss is given by

 30 − 1 18  29 
P[ N 18 = 30] = p N18 (30) =  q (1 − q )12 =  q 18 (1 − q )12
18 − 1  17 

4.29 Pete makes house calls to give away free books to families with children. He gives away
books to those families that open the door for him when he rings their doorbell and have
children living at home. He gives exactly one book to a qualified family. Studies indicate

16
 that the probability that the door is opened when Pete rings the doorbell is 0.75, and the
probability that a family has children living at home is 0.5. If the events “door opened” and
“family has children” are independent, determine the following:
(a) The probability that Pete gives away his first book at the third house he visits
(b) The probability that he gives away his second book to the fifth family he visits
(c) The conditional probability that he gives away the fifth book to the eleventh family he
visits, given that he has given away exactly four books to the first eight families he
visited.
(d) Given that he did not give away the second book at the second house, what is the
probability that he will give it out at the fifth house?
Solution: If we define success as the situation in which Pete gives away a book, then the
probability of success at each door that Pete visits is p = 0.75 × 0.5 = 0.375. Let X k be a random
variable that denotes the number of doors that Pete visits up to and including the door where he
has his kth success. Then X k is a kth-order Pascal random variable with PMF

 x − 1 k
p X k ( x) =   p (1 − p )x −k x = k , k + 1, k + 2, Κ ; k = 1, 2, Κ , x
 k − 1
(a) The probability that Pete gives away his first book at the third house he visits is given by

 3 − 1 1
P[ X 1 = 3] = p X1 (3) =   p (1 − p )2 = p (1 − p ) 2 = (0.375)(0.625) 2 = 0.1465
1 − 1 
(b) The probability that he gives away his second book to the fifth family he visits is given
by

 5 − 1 2  4
P[ X 2 = 5] = p X 2 (5) =   p (1 − p )3 =  (0.375) 2 (0.625) 3 = 4(0.375) 2 (0.625) 3
 2 − 1 1 
= 0.1373

(c) Since the outcomes of the visits are independent, the event that he gives away the fifth
book to the eleventh family he visits, given that he has given away exactly four books to
the first eight families he visited, is equivalent to the event that he encounters failures at
the 9th and 10th doors but success at the 11th door. Thus, the probability of this event is
given by
q = p (1 − p ) 2 = (0.375)(0.625) 2 = 0.1465

(d) Given that he did not give away the second book at the second house, the probability that
he will give it out at the fifth house is given by

17
 P[( X 2 = 5) ∩ ( X 2 > 2 )] P[ X 2 = 5] p X 2 (5)
P[ X 2 = 5 | X 2 > 2] = = =
P[ X 2 > 2] P[ X 2 > 2] 1 − P[ X 2 ≤ 2]
 5 − 1 2
 p (1 − p )
3

2 − 1 4 p 2 (1 − p ) 4 p 2 (1 − p )
3 2
p X 2 (5) 
= = = =
1 − p X 2 ( 2) 1 − p 2 (1 − p) 0 1− p2 1+ p
4(0.375) 2 (0.625) 2
= = 0.1598
1.375
4.30 The Carter family owns a bookstore and their son, who just completed an introductory
course in probability at a local college, has determined that the probability that anyone who
comes to the store actually buys a book is 0.3. If the family gives a coupon to the local ice
cream place to every customer who buys a book from the store, what is the probability that
on a particular day the third coupon was given to the eighth customer?

Solution: Let X k be a random variable that denotes the number of customers up to and
including the customer that received the kth coupon. If p = 0.3 is the probability that a customer
receives a coupon, then X k is a kth-order Pascal random variable with PMF

 x − 1 k
p X k ( x) =   p (1 − p )x −k x = k , k + 1, k + 2, Κ ; k = 1, 2, Κ , x
 k − 1
Thus, the probability that on a particular day the third coupon was given to the eighth customer is
given by

 8 − 1 3 7
P[ X 3 = 8] = p X 3 (8) =   p (1 − p )5 =  (0.3) 3 (0.7 )5 = 21(0.3) 3 (0.7 )5 = 0.0953
 3 − 1 2

4.31 A telemarketer is paid $1 for each sale she makes. The probability that any call she makes
results in a sale is 0.6.
(a) What is the probability that she earned her third dollar on the sixth call she made?
(b) If she made 6 calls per hour, what is the probability that she earned $8 in two hours?

Solution: The probability of success is p = 0.6. Let X k be a random variable that denotes the
number of calls up to and including the call that results in the kth success. Then X k is a kth-
order Pascal random variable with PMF

 x − 1 k
p X k ( x) =   p (1 − p )x −k x = k , k + 1, k + 2, Κ ; k = 1, 2, Κ , x
 k − 1 
(a) The probability that she earned her third dollar on the sixth call she made is given by

18
  6 − 1 3 5
P[ X 3 = 6] = p X 3 (6) =   p (1 − p )3 =  (0.6) 3 (0.4 )3 = 10(0.24) 3 = 0.13824
 3 − 1  2
(b) If she made 6 calls per hour, then in 2 hours she made 12 calls. Therefore, the
probability that she earned $8 in two hours is given by the binomial distribution of 8
successes in 12 trials, which is
12  8 12 
  p (1 − p )4 =  (0.6) 8 (0.4 )4 = 495(0.6) 8 (0.4 )4 = 0.2128
8  8

Section 4.6: Hypergeometric Distribution

4.32 A list contains the names of 4 girls and 6 boys. If 5 students are randomly selected from the
list, what is the probability that those selected will consist of 2 girls and 3 boys?

Solution: The probability p that those selected will consist of 2 girls and 3 boys is given by

 4  6 
  
2 3
p =    = 0.4762
10 
 
5

4.33 There are 50 states in the United States, and each state is represented by 2 senators in the
U.S. Senate. A group of 20 U.S. senators is chosen randomly from the 100 senators in the
U.S. Senate to visit a troubled part of the world.
(a) What is the probability that the two Massachusetts senators are among those chosen?
(b) What is the probability that neither of the two Massachusetts senators is among those
selected?
Solution: Let M denote the number of Massachusetts senators among the group of 20 senators
randomly chosen. To see M as a hypergeometric random variable, we imagine the senators
partitioned into two groups: one group of 2 senators from Massachusetts, and one group of 98
from other states. Thus, we have that

(a) The probability that the two Massachusetts senators are among those chosen is

 2  98 
  
2 18
P[ M = 2] =    = 0.03838
100 
 
 20 

(b) The probability that neither of the two Massachusetts senators is among those selected is

19
  2  98 
  
0 20
P[ M = 0] =    = 0.63838
100 
 
 20 

4.34 A professor provides 12 review problems for an exam and tells the students that the actual
exam will consist of 6 problems chosen randomly from the 12 review problems. Alex
decides to memorize the solutions to 8 of the 12 problems. If Alex cannot solve any of the
other 4 problems that he did not memorize, what is the probability that he is able to solve 4
or more problems correctly in the exam?

Solution: By memorizing only 8 of the 12 problems Alex has partitioned the problems into two
sets: the set of problems that he knows and the set of problems that he does not know. Let K be a
random variable that denotes the number of problems that Alex gets correctly. Then K is a
hypergeometric random variable with PMF

 8  4 
  
 k  6 − k 
p K (k ) = P[ K = k ] = k = 0, 1, Κ , 6
12 
 
6
Thus, the probability that he is able to solve 4 or more problems correctly in the exam is given by

 8  4   8  4   8  4 
   +    +   
4 2 5 1 6 0 420 + 224 + 28
P[ K ≥ 4] = p K (4) + p K (5) + p K (6) =          = = 0.7273
12  924
 
6
4.35 A class has 18 boys and 12 girls. If the teacher randomly selects a group of 15 students to
represent the class in a competition, determine the following:
(a) The probability that 8 members of the group are girls
(b) The expected number of boys in the group
Solution: The total number of students is 30. Let N be a random variable that denotes the number
of girls among a group of 15 students randomly selected to represent the class in a competition.

(a) The probability that 8 members of the group are girls is given by

12 18 
  
8 7
P[ N = 8] =    = 0.10155
 30 
 
15 

20
 (b) The probability that a randomly selected student is a boy is p = 18 30 = 0.6.
Thus, the
expected number of boys in the group is 15 p = 15(0.6) = 9.
4.36 A drawer contains 10 left gloves and 12 right gloves. If you randomly pull out a set of 4
gloves, what is the probability that the set consists of 2 right gloves and 2 left gloves?
Solution: The probability p that the set consists of 2 right gloves and 2 left gloves is given by

10 12 
  
2 2
p =    = 0.4060
 22 
 
4

Section 4.7: Poisson Distribution

4.37 The number of cars that arrive at a gas station is a Poisson random variable with a mean of
50 cars per hour. The station has only one attendant, and each car requires exactly one
minute to fill up. If we define a waiting line as the condition in which two or more cars are
found at the same time at the gas station, what is the probability that a waiting line will
occur at the station?

Solution: Let N denote the number of cars that arrive at the gas station over a 1-minute interval.
Since N is a Poisson random variable with mean λ = 5060 = 6 cars per minute, its PMF is
5

(5 / 6) n −5 / 6
p N ( n) = e n = 0, 1, 2, Κ
n!

Since it takes exactly 1 minute to service a car, a waiting line occurs when at least 1 other car
arrives within the 1-minute interval that it takes to finish servicing the current receiving service.
Thus, the probability that a waiting line will occur at the station is given by

PN > 0] = 1 − P[ N = 0] = 1 − p N (0) = 1 − e −5 / 6 = 0.5654

4.38 The number of traffic tickets that a certain traffic officer gives out on any day has been
shown to have a Poisson distribution with a mean of 7.
(a) What is the probability that on one particular day the officer gave out no ticket?
(b) What is the probability that she gives out fewer than 4 tickets in one day?
Solution: Let K denote the number of traffic tickets that the traffic officer gives out on any day.
Then the PMF of K is given by

7 k −7
p K (k ) = e k = 0, 1, 2, Κ
k!
(a) The probability that on one particular day the officer gave out no ticket is given by
21
 7 0 −7
P[ K = 0] = p K (0) = e = e −7 = 0.0009
0!
(b) The probability that she gives out fewer than 4 tickets in one day is given by
 49 343  538 −7
P[ K < 4] = p K (0) + p K (1) + p K (2) + p K (3) = e −7 1 + 7 + + = e = 0.0818
 2 6  6

4.39 A Geiger counter counts the particles emitted by radioactive material. If the number of
particles emitted per second by a particular radioactive material has a Poisson distribution
with a mean of 10 particles, determine the following:
(a) The probability of at most 3 particles in one second
(b) The probability of more than 1 particle in one second.
Solution: Let M denote the number of particles emitted per second. Since M has a Poisson
distribution with a mean of 10, its PMF is given by
10 m −10
p M ( m) = e m = 0, 1, 2, Κ
m!
(a) The probability of at most 3 particles in one second is given by

 100 1000  683 −10

P[ M ≤ 3] = p M (0) + p M (1) + p M (2) + p M (3) = e −10 1 + 10 + + = e = 0.01034
 2 6  3
(b) The probability of more than 1 particle in one second is given by

P[ M > 1] = 1 − P[ M ≤ 1] = 1 − p M (0) − p M (1) = 1 − e −10 {1 + 10} = 1 − 11e −10 = 0.9995

4.40 The number of cars that arrive at a drive-in window of a certain bank over a 20-minute
period is a Poisson random variable with a mean of four cars. What is the probability that
more than three cars will arrive during any 20-minute period?

Solution: Let K denote the number of cars that arrive at the window over a 20-minute period.
Since K has a Poisson distribution with a mean of 4, its PMF is given by
4 k −4
p K (k ) = e k = 0, 1, 2, Κ
k!
The probability that more than three cars will arrive during any 20-minute period is given by

P[ K > 3] = 1 − P[ K ≤ 3] = 1 − {p K (0) + p K (1) + p K (2) + p K (3)}

 16 64  71
= 1 − e −4 1 + 4 + +  = 1 − e −4 = 0.5665
 2 6 3

4.41 The number of phone calls that arrive at a secretary’s desk has a Poisson distribution with a
mean of 4 per hour.
(a) What is the probability that no phone calls arrive in a given hour?
(b) What is the probability that more than 2 calls arrive within a given hour?
22
Solution: Let N denote the number of phone calls that arrive in 1 hour. Since N is a Poisson
random variable with a mean of 4, its PMF is

4 n −4
p N ( n) = e n = 0, 1, 2, Κ
n!
(a) The probability that no phone calls arrive in a given hour is given by
4 0 −4
P[ N = 0] = p N (0) = e = e −4 = 0.0183
0!

(b) The probability that more than 2 calls arrive within a given hour is given by

P[ N > 2] = 1 − P[ K ≤ 2] = 1 − {p N (0) + p N (1) + p NK (2)}

 16 
= 1 − e − 4 1 + 4 +  = 1 − 13e −4 = 0.7619
 2

4.42 The number of typing mistakes that Ann makes on a given page has a Poisson distribution
with a mean of 3 mistakes.
(a) What is the probability that she makes exactly 7 mistakes on a given page?
(b) What is the probability that she makes fewer than 4 mistakes on a given page?
(c) What is the probability that Ann makes no mistake on a given page?
Solution: Let K denote the number of typing mistakes on a given page. Since K has a Poisson
distribution with a mean of 3, its PMF is given by

3 k −3
p K (k ) = e k = 0, 1, 2, Κ
k!
(a) The probability that she makes exactly 7 mistakes on a given page is given by

3 7 −3
P[ K = 7] = p K (7) = e = 0.0216
7!
(b) The probability that she makes fewer than 4 mistakes on a given page is given by

 9 27 
P[ K < 4] = p K (0) + p K (1) + p K (2) + p K (3) = e −3 1 + 3 + +  = 13e −3 = 0.6472
 2 6

(c) The probability that Ann makes no mistake on a given page is given by

3 0 −3
P[ K = 0] = p K (0) = e = e −3 = 0.0498
0!

23
Section 4.8: Exponential Distribution
4.43 The PDF of a certain random variable T is given by

f T (t ) = ke −4t t≥0

(a) What is the value of k?

(b) What is the expected value of T?
(c) Find P[T < 1].

Solution:
(a) To find the value of k, we know that
∞
∞ ∞
− 4t  e −4t  k
∫ −∞
f T (t )dt = 1 = k ∫ e
0
dt = k −
 4 0 4
 = ⇒k=4

(b) The expected value of T is E[T ] = 14 = 0.25

1
[
(c) P[T < 1] = ∫ 4e − 4t dt = − e − 4t
0
]
1
0 = 1 − e − 4 = 0.9817

4.44 The lifetime X of a system in weeks is given by the following PDF:

f X ( x) = 0.25e −0.25 x x≥0

(a) What is the expected value of X?

(b) What is the CDF of X?
(c) What is the variance of X?
(d) What is the probability that the system will not fail within two weeks?
(e) Given that the system has not failed by the end of the fourth week, what is the prob-
ability that it will fail between the fourth and sixth weeks?
Solution:
(a) The expected value is E[ X ] = 1
0.25 =4
(b) The CDF is FX ( x ) = 1 − e −0.25 x , x ≥ 0
(c) The variance is σ X2 = 1
( 0.25 ) 2
= 16
(d) The probability that the system will not fail within two weeks is given by

{ }
P[ X > 2] = 1 − P[ X ≤ 2] = 1 − FX (2) = 1 − 1 − e −0.25( 2) = e −0.5 = 0.6065
(e) Because of the forgetfulness property of the exponential distribution, the probability that
the system will fail between the fourth and sixth weeks, given that it has not failed by the
end of the fourth week, is simply the probability that it will fail within 2 weeks, which is

24
 P[4 < X ≤ 6 | X > 4] = P[ X ≤ 2] = FX (2) = 1 − e −0.25( 2) = 1 − e −0.5 = 0.3935

4.45 The time T in hours between bus arrivals at a bus station in downtown Lowell is a random
variable with the following PDF:

f T (t ) = 2e −2t t≥0

(a) What is the expected value of T?

(b) What is the variance of T?
(c) What is P[T > 1] ?
Solution: We know that FT (t ) = 1 − e −2t
(a) E[T ] = 12 = 0.5
(b) σ T2 = 1
22
= 14 = 0.25

{ }
(c) P[T > 1] = 1 − P[T ≤ 1] = 1 − FT (1) = 1 − 1 − e −2 (1) = e −2 = 0.1353

4.46 The PDF of the times between successive bus arrivals at a suburban bus stop is given by

f T (t ) = 0.1e −0.1t t≥0

where T is in minutes. A turtle that requires exactly 15 minutes to cross the street starts
crossing the street at the bus station immediately after a bus has left the station. What is the
probability that the turtle will not be on the road when the next bus arrives?

Solution: The probability p that the turtle will not be on the road when the next bus arrives is the
probability that no bus arrives within the time it takes the turtle to cross the road, which is given
by
{ }
p = P[T > 15] = 1 − P[T ≤ 15] = 1 − FT (15) = 1 − 1 − e −0.1(15) = e −1.5 = 0.2231

4.47 The PDF of the times between successive bus arrivals at a suburban bus stop is given by

f T (t ) = 0.2e −0.2t t≥0

where T is in minutes. An ant that requires exactly 10 minutes to cross the street starts
crossing the street at the bus station immediately after a bus has left the station. Given that
no bus has arrived in the past 8 minutes since the ant started its journey across the street,
determine the following:

(a) The probability that the ant will completely cross the road before the next bus arrives.
(b) The expected time until the next bus arrives.
Solution:
25
 (a) The probability p that the ant will completely cross the road before the next bus arrives is
the probability that no bus arrives within the remaining 2 minutes of the ant’s trip.
Because of the forgetfulness property of the exponential distribution, the PDF of the time
until the next bus arrival is still exponentially distributed with the same mean as T. Thus,
p is given by

p = P[T > 2] = 1 − P[T ≤ 2] = 1 − FT (2) = e −0.2 ( 2 ) = e −0.4 = 0.6703

(b) Because of the forgetfulness property of the exponential distribution, the expected time
until the next bus arrival is is the expected value of T, which is E[T ] = 01.2 = 5.

4.48 The times between telephone calls that arrive at a switchboard are exponentially distributed
with a mean of 30 minutes. Given that a call has just arrived, what is the probability that it
takes at least 2 hours before the next call arrives?

Solution: Let X denote the time between call arrivals at the switchboard. Then the PDF of X is
given by

1 − x / 30
f X ( x) = e x≥0
30

Given that a call has just arrived, the probability that it takes at least 2 hours (or 120 minutes)
before the next call arrives is given by

P[ X > 120] = 1 − P[ X ≤ 120] = 1 − FX (120) = e −120 / 30 = e −4 = 0.0183

4.49 The durations of calls to a radio talk show are known to be exponentially distributed with a
mean of 3 minutes.
(a) What is the probability that a call will last less than 2 minutes?
(b) What is the probability that a call will last longer than 4 minutes?
(c) Given that a call has already lasted 4 minutes, what is the probability that it will last at
least another 4 minutes?
(d) Given that a call has already lasted 4 minutes, what is the expected remaining time until
it ends?

Solution: Let X denote the duration of a call to the talk show. Then the PDF of X is given by

1
f X ( x) = e − x / 3 x≥0
3

(a) The probability that a call will last less than 2 minutes is given by

P[ X < 2] = P[ X ≤ 2] = FX (2) =1 − e −2 / 3 = 0.4866

26
 (b) The probability that a call will last longer than 4 minutes

P[ X > 4] = 1 − P[ X ≤ 4] = 1 − FX (4) = e −4 / 3 = 0.2636

(c) Because of the forgetfulness property of the exponential distribution, the probability that
a call will last at least another 4 minutes, given that it has already lasted 4 minutes, is the
probability that a call will last at least 4 minutes and is given by

P[ X ≥ 8 | X > 4] = P[ X ≥ 4] = 1 − FX (4) = e −4 / 3 = 0.2636

(d) Because of the forgetfulness property of the exponential distribution, the expected
remaining time until a call ends, given that it has already lasted 4 minutes, is the mean
duration of a call, which is E[ X ] = 3 minutes.

4.50 The life of a particular brand of batteries is exponentially distributed with a mean of 4
weeks. You just replaced the battery in your gadget with the particular brand.
(a) What is the probability that the battery life exceeds 2 weeks?
(b) Given that the battery has lasted 6 weeks, what is the probability that it will last at least
another 5 weeks?

Solution: Let X denote the life of a battery in weeks. Then the PDF of X is given by

1 −x / 4
f X ( x) = e x≥0
4

(a) The probability that the battery life exceeds 2 weeks is given by

P[ X > 2] = 1 − P[ X ≤ 2] = 1 − FX (2) = e −2 / 4 = e −0.5 = 0.6065

(b) Because of the forgetfulness property of the exponential distribution, the probability that
the battery will last at least another 5 weeks, given that it has already lasted 6 weeks, is
the probability that a battery will last at least 5 weeks and is given by

P[ X ≥ 11 | X > 6] = P[ X ≥ 5] = 1 − FX (5) = e −5 / 4 = 0.2865

4.51 The PDF of the times T in weeks between employee strikes at a certain company is given
by
f T (t ) = 0.02e −0.02t t≥0
(a) What is the expected time between strikes at the company?
(b) Find P[T ≤ t | T < 40] for all t.
(c) Find P[40 < T < 60].

Solution:

27
 1
(a) The expected time between strikes at the company is E[T ] = = 50 weeks.
0.02
(b) P[T ≤ t | T < 40] is given by

P[(T ≤ t ) ∩ (T < 40)] P[T ≤ t ] F (t )

P[T ≤ t | T < 40] = = = T 0 ≤ t < 40
P[T < 40] P[T < 40] FT (40)
1 − e −0.02t 1 − e −0.02t
= − 0.02 ( 40)
= − 0.8
(
= 1.8160 1 − e −0.02t ) 0 ≤ t < 40
1− e 1− e

(c) P[40 < T < 60] = FT (60) − FT (40) = e −0.8 − e −1.2 = 0.1481

4.52 Find the PDF of the random variable X whose hazard function is given by h X ( x) = 0.05.

Solution: We have that

h X ( x) = 0.05 =
f X ( x)
1 − FX ( x )
{ } { }
⇒ 1 − FX ( x) = exp − ∫ h X (u )du = exp − [0.05t ]0 = e −0.05 x
0
x x

This implies that FX ( x) = 1 − e −0.05 x and the PDF is given by

f X ( x) = 0.05{1 − FX ( x)} = 0.05e −0.05 x x≥0

This means that X is an exponentially distributed random variable. Note that we can also obtain
f X (x) as follows:
d
f X ( x) = FX ( x) = 0.05e −0.05 x x≥0
dx
Section 4.9: Erlang Distribution
4.53 A communication channel fades in a random manner. The duration X of each fade is
exponentially distributed with a mean of 1 / λ. The duration T of the interval between the
end of one fade and the beginning of the next fade is an Erlang random variable with PDF
µ 4 t 3 e − µt
f T (t ) = t≥0
3!

If we observe the channel at a randomly selected instant, what is the probability that it is in
the fade state?

Solution: The cycle of events can be represented as follows:

28
 Λ

The PDF of X is given by

f X ( x ) = λ e − λx x≥0

We observe that T is a 4th-order Erlang random variable. Thus, its expected value is

4
E[T ] =
µ
Let Y denote the duration of one cycle of fade-no fade condition on the channel. Then
Y = X + T , and E[Y ] = E[ X ] + E[T ]. Thus, the probability p that the channel is in the fade state
at a randomly selected instant is given by

E[ X ] E[ X ] 1
λ µ
p= = = =
E[Y ] E[ X ] + E[T ] 1
λ + 4
µ µ + 4λ

4.54 The random variable X, which denotes the interval between two consecutive events, has the
PDF
f X ( x) = 4 x 2 e −2 x x≥0

If we assume that intervals between events are independent, determine the following:

(a) The expected value of X.

(b) The expected value of the interval between the 11th and 13th events
(c) The probability that X ≤ 6.
Solution: X is a 3rd-order Erlang random variable and the parameter of the underlying
exponential distribution is λ = 2. Thus,
(a) The expected value of X is E[ X ] = 3 / λ = 3 / 2 = 1.5.
(b) The expected value of the interval between the 11th and 13th events is the length of 2
inter-arrival times, which is 2 E[ X ] = 3.
(c) The probability that X ≤ 6 is given by
2
(2 x) k e −2 x  (12) 2 
P[ X ≤ 6] = FX (6) = 1 − ∑ = 1 − e −12 1 + 12 + −12
 = 1 − 85e = 0.9995
k =0 k! x =6  2 

4.55 The students in the electrical and computer engineering department of a certain college
arrive at the departmental lounge to view the video of the lecture for a particular course
according to a Poisson process with a rate of 5 students per hour. The person in charge of
29
 the operating the VCR will not turn the machine on until there are at least 5 students in the
lounge.
(a) Given that there is currently no student in the lounge, what is the expected waiting time
until the VCR is turned on?
(b) Given that there is currently no student in the lounge, what is the probability that the
VCR is not turned on within one hour from now?
Solution: Let N denote the number of students who arrive in one hour. Then the PMF of N is
given by

5 n −5
p N ( n) = e n = 0, 1, 2, Κ
n!

(a) Since the number of arrivals is a Poisson random variable with mean λ = 5, the interval
X between arrivals has the exponential distribution with PDF

f X ( x) = 5e −5 x x≥0

where E[ X ] = 1
5 hours or 12 minutes. Thus, given that there is currently no student in the
lounge, the expected waiting time until the VCR is turned on is the mean time between 5
arrivals, which is the mean of a 5th-order Erlang random variable X 5 and is given by
E[ X 5 ] = 5 E[ X ] = 1 hour.

(b) Given that there is currently no student in the lounge, the probability that the VCR is not
turned on within one hour from now

4
(5 x) k e −5 x 4
5 k e −5
P[ X 5 > 1] = 1 − P[ X 5 ≤ 1] = 1 − FX 5 (1) = ∑ =∑
k =0 k! x =1 k =0 k!
 25 125 625  1569 −5
= e −5 1 + 5 + + + = e = 0.4405
 2 6 24  24

Section 4.10: Uniform Distribution

4.56 Jack is the only employee of an auto repair shop that specializes in installing mufflers. The
time T minutes that it takes Jack to install a muffler has the PDF
0.05 10 ≤ t ≤ 30
f T (t ) = 
0 otherwise

(a) What is the expected time it takes Jack to install a muffler?

(b) What is the variance of the time it takes Jack to install a muffler?
Solution: Since T is a uniformly distributed random variable, we have that
30
 10 + 30
(a) The expected value of T is given by E[T ] = = 20
2
(30 − 10) 2 100
(b) The variance of T is given by σ T2 = = = 33.33
12 3

4.57 A random variable X is uniformly distributed between 0 and 10. Find the probability that X
lies between the standard deviation σ X and the mean E[X ].

Solution: The PDF of X is given by

0.1 0 ≤ x ≤ 10
f X ( x) = 
0 otherwise

Its mean, variance and standard deviation are given by

10 + 0
E[ X ] = =5
2
(10 − 0) 2 100
σ X2 = =
12 12
100
σX = = 2.8867
12

Thus, the probability that X lies between σ X and E[ X ] is given by

5 5
P[σ X < X < E[ X ]] = P[2.8867 < X < 5] = ∫ f X ( x)dx = 0.1∫ dx
2.8867 2.8867

= 0.1[x ]2.8867 = 0.2113

5

4.58 A random variable X is uniformly distributed between 3 and 15. Find the following
parameters:
(a) The expected value of X
(b) The variance of X
(c) The probability that X lies between 5 and 10
(d) The probability that X is less than 6.

Solution: The PDF of X is given by

1
 3 ≤ x ≤ 15
f X ( x) = 12
0 otherwise

31
 15 + 3
(a) The expected value of X is given by E[ X ] = =9
2

(15 − 3) 2 144
(b) The variance of X is given by σ X2 = = = 12
12 12
(c) The probability that X lies between 5 and 10 is given by
10
10 10 1 x 5
P[5 < X < 10] = ∫ f X ( x)dx = ∫ dx =   =
5 5 12 12  5 12

(d) The probability that X is less than 6 is given by

6
6 6 1 x 3 1
P[ X < 6] = ∫ f X ( x)dx = ∫ dx =   = =
3 3 12 12  3 12 4

4.59 Starting at 7 am, buses arrive at a particular bus stop in a college campus at intervals of 15
minutes (that is, at 7 am, 7:15, 7:30, etc.). Joe is a frequent passenger on this route, and the
time he arrives each morning to catch a bus is known to be uniformly distributed between 7
am and 7:30 am.
(a) What is the probability that Joe waits less than 5 minutes for a bus?
(b) What is the probability that he waits more than 10 minutes for a bus?
Solution: Let T be a random variable that denotes the time that Joe arrives at the bus stop. The
following figure shows the PDF of T as well as part of the bus arrival times.
fT (t)

32
 (a) To wait less than 5 minutes for the bus, Joe must arrive between 7:10 am and 7:15 am or
between 7:25 am and 7:30 am. Thus, if A is the event that Joe waits less than 5 minutes
for the bus, then the probability of this event is given by
P[ A] = P[(10 < T < 15) ∪ (25 < T < 30] = P[10 < T < 15] + P[25 < T < 30]
15 30
15 30 1 15 30 1 t   t  10 1
= ∫ f T (t )dt + ∫ f T (t )dt = ∫ dt + ∫ dt =   +   = =
10 25 10 30 25 30
 30  10  30  25 30 3
(b) To wait more than 10 minutes for the bus, Joe must arrive between 7:00 am and 7:05 am
or between 7:15 am and 7:20 am. Thus, if B is the event that Joe waits more than 10
minutes for the bus, then the probability of this event is given by
P[ B] = P[(0 < T < 5) ∪ (15 < T < 20] = P[0 < T < 5] + P[15 < T < 20]
5 20
5 20 5 1 20 1  t   t  10 1
= ∫ f T (t )dt + ∫ f T (t )dt = ∫ dt + ∫ dt =   +   = =
0 15 0 30 15 30  30  0  30  15 30 3

4.60 The time it takes a bank teller to serve a customer is uniformly distributed between 2 and 6
minutes. A customer has just stepped up to the window, and you are next in line.
(a) What is the expected time you will wait before it is your turn to be served?
(b) What is the probability that you wait less than 1 minute before being served?
(c) What is the probability that you wait between 3 and 5 minutes before being served?

Solution: Let X be a random variable that denotes the time it takes a teller to serve a customer.
Then the PDF of X is given by

1
 2≤x≤6
f X ( x) =  4
0 otherwise

Given that a customer has just stepped up to the window and you are next in line,
(a) The expected time you will wait before it is your turn to be served is the mean service
time of a customer, which is E[ X ] = 6+2 2 = 4 minutes

(b) The probability that you wait less than 1 minute before being served is the probability
that it takes less than 1 minute to serve a customer, which is 0 since the service time lies
between 2 and 6 minutes.
(c) The probability that you wait between 3 and 5 minutes before being served is given by
5
5 5 1 x 2 1
P[3 < X < 5] = ∫ f X ( x)dx = ∫ dx =   = =
3 3 4  43 4 2

33
Section 4.11: Normal Distribution
4.61 The mean weight of 200 students in a certain college is 140 lbs, and the standard deviation
is 10 lbs. If we assume that the weights are normally distributed, evaluate the following:
(a) The expected number of students that weigh between 110 and 145 lbs
(b) The expected number of students that weigh less than 120 lbs
(c) The expected number of students that weigh more than 170 lbs.
Solution: Let X be a random variable that denotes the weight of the students. Since the number
of observations N = 200, we know that X is approximately normally distributed. Given that
µ X = 140 and σ X = 10, we have that

(a) The fraction of students that weigh between 110 lbs and 145 lbs is given by

 145 − 140   110 − 140 

P[110 < X < 145] = FX (145) − FX (110) = Φ  − Φ  = Φ (0.50) − Φ (− 3.00 )
 10   10 
= Φ (0.50 ) − {1 − Φ (3.00)} = Φ (0.50) + Φ (3.00) − 1
= 0.6915 + 0.9987 − 1 = 0.6902

Thus, expected number of students that weigh between 110 and 145 lbs is given by
0.6902 × 200 = 138.04.
(b) The fraction of students that weigh less than 120 lbs is given by

 120 − 140 
P[ X < 120] = FX (120) = Φ  = Φ(− 2.00) = 1 − Φ(2.00) = 1 − 0.9772 = 0.0228
 10 

Thus, expected number of students that weigh less than 120 lbs is given by
0.0228 × 200 = 4.56.
(c) The fraction of students that weigh more than 170 lbs is given by

 170 − 140 
P[ X > 170] = 1 − P[ X ≤ 170] = 1 − FX (170) = 1 − Φ  = 1 − Φ(3.00)
 10 
= 1 − 0.9987 = 0.0013

Thus, expected number of students that weigh more than 170 lbs is given by
0.0013 × 200 = 0.26.
4.62 The weights of parcels that are dropped off at a local shipping center can be represented by
a random variable X that is normally distributed with mean µ X = 70 and standard
deviation σ X = 10. Determine the following:
(a) P[ X > 50]
(b) P[60 < X ]

34
 (c) P[60 < X < 90].

Solution:
(a) The probability that X is greater than 50 is given by

 50 − 70 
P[ X > 50] = 1 − P[ X ≤ 50] = 1 − FX (50) = 1 − Φ  = 1 − Φ (− 2.00)
 10 
= 1 − {1 − Φ (2.00)} = Φ (2.00) = 0.9772

(b) The probability that X is less than 60 is given by

 60 − 70 
P[ X < 60] = FX (60) = Φ  = Φ (− 1.00 ) = 1 − Φ (1.00) = 1 − 0.8413 = 0.1587
 10 

(c) The probability that X lies between 60 and 90 is given by

 90 − 70   60 − 70 
P[60 < X < 90] = FX (90) − FX (60) = Φ  − Φ  = Φ(2.00 ) − Φ(− 1.00)
 10   10 
= Φ(2.00) − {1 − Φ (1.00)} = Φ(2.00) + Φ (1.00) − 1 = 0.9772 + 0.8413 − 1 = 0.8185

4.63 Find the probability of getting between 4 and 8 heads in 12 tosses of a fair coin by using
(a) the binomial distribution
(b) the normal approximation to the binomial distribution
Solution: Let K be a random variable that denotes the number of heads in 12 tosses of a fair coin.
Then the probability of success is p = 12 and the PMF of K is given by
k 12− k 12
12  1   1  12  1 
p K (k ) =      =    k = 0, 1, Κ , 12
 k  2   2   k  2 
(a) Using the direct method, the probability of getting between 4 and 8 heads is given by
 1  12  12  12  12  12  3498
12 12
8
12  1 
P[4 ≤ K ≤ 8] = ∑    =     +   +   +   +   = = 0.8540
k = 4  k  2   2   4   5   6   7   8  4096
(b) Using the normal approximation to the binomial distribution, we have that
n = 12
p = 0.5
E[ K ] = np = 6
σ K2 = np (1 − p ) = 3
Therefore, the approximate value of the probability of getting between 4 and 8 heads is
given by

35
 8−6 4−6  2   2 
P[4 ≤ K ≤ 8] = Fk (8) − FK (4) = Φ  − Φ  = Φ  − Φ − 
 3   3   3  3
 2    2   2 
= Φ  − 1 − Φ  = 2Φ  − 1 = 2Φ (1.15) − 1
 3   3   3
= 2(0.9394) − 1 = 0.8788

4.64 The test scores X of a certain subject were found to be approximately normally distributed
with mean µ X and standard deviation σ X . So the professor decided to assign grades
according to Table 4.3. What fraction of the class gets A, B, C, D, and F?

Table 4.3: Table for Grading Score

Test Score Letter Grade

X < µ X − 2σ X F

µ X − 2σ X ≤ X < µ X − σ X D

µX −σ X ≤ X < µX C

µX ≤ X < µX +σ X B

X ≥ µX +σ X A

Solution:
(a) The fraction of the class with the letter grade A is given by
 µ +σ X − µX 
P[ A] = P[ X ≥ µ X + σ X ] = 1 − P[ X < µ X + σ X ] = 1 − FX ( µ X + σ X ) = 1 − Φ X 
 σX 
= 1 − Φ(1) = 1 − 0.8413 = 0.1587

(b) The fraction of the class with the letter grade B is given by
 µ +σ X − µX   µ − µX 
P[ B] = P[ µ X ≤ X < µ X + σ X ] = FX ( µ X + σ X ) − FX ( µ X ) = Φ X  − Φ X 
 σX   σX 
= Φ (1) − Φ(0) = 0.8413 − 0.5000 = 0.3413

(c) The fraction of the class with the letter grade C is given by

36
  µ − µX   µ −σ X − µX 
P[C ] = P[ µ X − σ X ≤ X < µ X ] = FX ( µ X ) − FX ( µ X − σ X ) = Φ X  − Φ X 
 σX   σX 
= Φ (0 ) − Φ (− 1) = Φ(0 ) − {1 − Φ(1)} = Φ(0 ) + Φ(1) − 1 = 0.5000 + 0.8413 − 1 = 0.3413

(d) The fraction of the class with the letter grade D is given by

P[ D] = P[ µ X − 2σ X ≤ X < µ X − σ X ] = FX ( µ X − σ X ) − FX ( µ X − 2σ X )
 µ −σ X − µX   µ − 2σ X − µ X 
= Φ X  − Φ X  = Φ(− 1) − Φ (− 2 ) = {1 − Φ (1)} − {1 − Φ(2)}
 σ X   σ X 
( ) ( )
= Φ 2 − Φ 1 = 0.9772 − 0.8413 − 1 = 0.1359

(e) The fraction of the class with the letter grade F is given by
 µ − 2σ X − µ X 
P[ F ] = P[ X < µ X − 2σ X ] = FX ( µ X − 2σ X ) = Φ X  = Φ(− 2 ) = 1 − Φ(2 )
 σX 
= 1 − 0.9772 = 0.0228

4.65 The random variable X is a normal random variable with zero mean and standard deviation
σ X . Compute the probability P[| X |≤ 2σ X ].
Solution: The probability P[| X |≤ 2σ X ] is given by

 2σ − 0   − 2σ X − 0 
P[| X |≤ 2σ X ] = P[−2σ X < X < 2σ X ] = FX (2σ X ) − FX (−2σ X ) = Φ X  − Φ 
 σX   σX 
= Φ(2) − Φ (− 2 ) = Φ(2) − {1 − Φ(2)} = 2Φ (2 ) − 1 = 2(0.9772) − 1 = 0.9544

4.66 The annual rainfall in inches in a certain region has a normal distribution with a mean of 40
and variance of 16. What is the probability that the rainfall in a given year is between 30
and 48 inches?
Solution: The probability that the rainfall in a given year is between 30 and 48 inches is given by

 48 − 40   30 − 40 
P[30 < X < 48] = FX (48) − FX (30) = Φ  − Φ  = Φ(2.00) − Φ (− 2.50)
 4   4 
= Φ(2.00 ) − {1 − Φ(2.50)} = Φ (2.00) + Φ(2.50) − 1 = 0.9772 + 0.9938 − 1
= 0.9710

37
Chapter 5: Multiple Random Variables

Chapter Summary: This chapter deals with multiple random variables including the joint
cumulative distribution function of bivariate random variables, conditional distributions,
covariance, correlation coefficient, functions of multivariate random variables, and multinomial
distributions.

Section 5.3: Discrete Bivariate Random Variables

5.1 The joint PMF of two discrete random variables X and Y is given by
kxy x = 1, 2, 3; y = 1, 2, 3
p XY ( x, y ) = 
0 otherwise

(a) Determine the value of the constant k.

(b) Find the marginal PMFs of X and Y.
(c) Find P[1 ≤ X ≤ 2, Y ≤ 2].
Solution:
(a) To determine the value of k we have that
3 3 3 3

∑∑ p XY ( x, y ) = 1 = k ∑∑ xy = k ∑ x{1 + 2 + 3} = 6k ∑ x
x y x =1 y =1 x =1 x =1

1
= 6k {1 + 2 + 3} = 36k ⇒ k =
36
(b) The marginal PMFs of X and Y are given by
3
x 3 6x x
y = {1 + 2 + 3} =
x
p X ( x) = ∑ p XY ( x, y ) = ∑ = x = 1, 2, 3
y =1 36 y =1 36 36 6
3
y 3 6y y
x = {1 + 2 + 3} =
y
p y ( y ) = ∑ p XY ( x, y ) = ∑ = y = 1, 2, 3
x =1 36 x =1 36 36 6

(c) Observe that p XY ( x, y ) = p X ( x) pY ( y ), which implies that X and Y are independent

random variables. Thus
P[1 ≤ X ≤ 2, Y ≤ 2] = P[1 ≤ X ≤ 2, ]P[ Y ≤ 2] = {p X (1) + p X (2)}× {pY (1) + pY (2)}
1
= {1 + 2}{1 + 2} = 9 = 1
36 36 4

5.2 A fair coin is tossed three times. Let the random variable X denote the number of heads in
the first two tosses, and let the random variable Y denote the number of heads in the third
toss. Determine the joint PMF p XY ( x, y ) of X and Y.

1
Solution: Let Ω denote the sample space of the experiment. Then Ω , X and Y are given as
follows:

Ω X Y
HHH 2 1
HHT 2 0
HTH 1 1
HTT 1 0
THH 1 1
THT 1 0
TTH 0 1
TTT 0 0

Thus, the joint PMF of X and Y, p XY ( x, y ), is given by

 18 x = 0, y = 0
1
8 x = 0, y = 1
1 x = 1, y = 0
 1
4

p XY ( x, y ) =  4 x = 1, y = 1
1 x = 2, y = 0
8
 18 x = 2, y = 1

0 otherwise

5.3 The joint PMF of two random variables X and Y is given by

0.10 x = 1, y = 1
0.35 x = 2, y = 2

p XY ( x, y ) = 0.05 x = 3, y = 3
0.50 x = 4, y = 4

0 otherwise

(a) Determine the joint CDF FXY ( x, y )

(b) Find P[1 ≤ X ≤ 2, Y ≤ 2].

Solution:
(a) The joint CDF FXY ( x, y ) can be obtained as follows:

2
FXY ( x, y ) = P[ X ≤ x, Y ≤ y ] = ∑∑ p XY (u , v)
u ≤ x v≤ y

FXY (1, 1) = ∑∑ p XY (u , v) = p XY (1, 1) = 0.10

u ≤1 v ≤1

FXY (1, 2) = ∑∑ p XY (u , v) = p XY (1, 1) + p XY (1, 2) = p XY (1, 1) = 0.10

u ≤1 v ≤ 2

FXY (1, 3) = ∑∑ p XY (u , v) = p XY (1, 1) + p XY (1, 2) + p XY (1, 3) = p XY (1, 1) = 0.10

u ≤1 v ≤3

FXY (1, 4) = ∑∑ p XY (u , v) = p XY (1, 1) + p XY (1, 2) + p XY (1, 3) + p XY (1, 3) = p XY (1, 1) = 0.10

u ≤1 v ≤ 4

FXY (2, 1) = ∑∑ p XY (u , v) = p XY (1, 1) + p XY (2, 1) = p XY (1, 1) = 0.10

u ≤ 2 v ≤1

FXY (2, 2) = ∑∑ p XY (u, v) = p XY (1, 1) + p XY (1, 2) + p XY (2, 1) + p XY (2, 2)

u ≤2 v≤ 2

= p XY (1, 1) + p XY (2, 2) = 0.45

FXY (2, 3) = ∑∑ p XY (u, v) = p XY (1, 1) + p XY (1, 2) + p XY (1, 3) + p XY (2, 1) + p XY (2, 2) + p XY (2, 3)
u ≤ 2 v ≤3

= p XY (1, 1) + p XY (2, 2) = 0.45

FXY (2, 4) = ∑∑ p XY (u, v) = p XY (1, 1) + p XY (1, 2) + p XY (1, 3) + p XY (1, 4) + p XY (2, 1) + p XY (2, 2)
u ≤2 v≤ 4

+ p XY (2, 3) + p XY (2, 4) = p XY (1, 1) + p XY (2, 2) = 0.45

FXY (3, 1) = ∑∑ p XY (u , v) = p XY (1, 1) + p XY (2, 1) + p XY (3, 1) = p XY (1, 1) = 0.10
u ≤ 3 v ≤1

FXY (3, 2) = ∑∑ p XY (u, v) = p XY (1, 1) + p XY (1, 2) + p XY (2, 1) + p XY (2, 2) + p XY (3, 1) + p XY (3, 2)

u ≤3 v ≤ 2

= p XY (1, 1) + p XY (2, 2) = 0.45

FXY (3, 3) = ∑∑ p XY (u , v) = p XY (1, 1) + p XY (1, 2) + p XY (1, 3) + p XY (2, 1) + p XY (2, 2) + p XY (2, 3)
u ≤ 3 v ≤3

+ p XY (3, 1) + p XY (3, 2) + p XY (3, 3)

= p XY (1, 1) + p XY (2, 2) + p XY (3, 3) = 0.50
FXY (3, 4) = ∑∑ p XY (u, v) = p XY (1, 1) + p XY (1, 2) + p XY (1, 3) + p XY (1, 4) + p XY (2, 1) + p XY (2, 2)
u ≤3 v ≤ 4

+ p XY (2, 3) + p XY (2, 4) + p XY (3, 1) + p XY (3, 2) + p XY (3, 3) + p XY (3, 4)

= p XY (1, 1) + p XY (2, 2) + p XY (3, 3) = 0.50

3
FXY (4, 1) = ∑∑ p XY (u , v) = p XY (1, 1) + p XY (2, 1) + p XY (3, 1) + p (4, 1) = p XY (1, 1) = 0.10
u ≤ 4 v ≤1

FXY (4, 2) = ∑∑ p XY (u, v) = p XY (1, 1) + p XY (1, 2) + p XY (2, 1) + p XY (2, 2) + p XY (3, 1) + p XY (3, 2)

u ≤4 v≤ 2

+ p XY (4, 1) + p XY (4, 2)
= p XY (1, 1) + p XY (2, 2) = 0.45
FXY (4, 3) = ∑∑ p XY (u, v) = p XY (1, 1) + p XY (1, 2) + p XY (1, 3) + p XY (2, 1) + p XY (2, 2) + p XY (2, 3)
u ≤ 4 v ≤3

+ p XY (3, 1) + p XY (3, 2) + p XY (3, 3) + p XY (4, 1) + p XY (4, 2) + p XY (4, 3)

= p XY (1, 1) + p XY (2, 2) + p XY (3, 3) = 0.50
FXY (4, 4) = 1

Thus, the joint CDF of X and Y is given by

0.00 x < 1, y < 1

0.10 x = 1, y = 1

0.10 x = 1, y = 2

0.10 x = 1, y = 3
0.10 x = 1, y = 4

0.10 x = 2, y = 1
0.45 x = 2, y = 2

0.45 x = 2, y = 3

FXY ( x, y ) = 0.45 x = 2, y = 4
0.10 x = 3, y = 1

0.45 x = 3, y = 2
0.50 x = 3, y = 3

0.50 x = 3, y = 4

0.10 x = 4, y = 1
0.45 x = 4, y = 2

0.50 x = 4, y = 3
1.00 x = 4, y = 4


(b) P[1 ≤ X ≤ 2, Y ≤ 2] is given by

P[1 ≤ X ≤ 2, Y ≤ 2] = p XY (1, 1) + p XY (1, 2) + p XY (2, 1) + p XY (2, 2)

= p XY (1, 1) + p XY (2, 2) = 0.45

5.4 Two discrete random variables X and Y have the joint CDF defined as follows:
4
  121 x = 0, y = 0
1
3 x = 0, y = 1
 2 x = 0, y = 2
FXY ( x, y ) =  13
6 x = 1, y = 0
7 x = 1, y = 1
 12
1 x = 1, y = 2

Determine the following:

(a) P[0 < X < 2, 0 < Y < 2]
(b) the marginal CDFs of X and Y (i.e., FX (x) and FY ( y ))
(c) P[ X = 1, Y = 1]

Solution: We first find the joint PMF of X and Y as follows:

1 1
FXY (0, 0) = = ∑∑ p XY (u , v) = p XY (0, 0) ⇒ p XY (0, 0) =
12 u ≤0 v≤0 12
1 1
FXY (0, 1) = = ∑∑ p XY (u , v) = p XY (0, 0) + p XY (0, 1) = + p XY (0, 1)
3 u ≤0 v≤1 12
1 1 1
⇒ p XY (0, 1) = − =
3 12 4
2 1
FXY (0, 2) = = ∑∑ p XY (u , v) = p XY (0, 0) + p XY (0, 1) + p XY (0, 2) = + p XY (0, 2)
3 u π 0 v≤ 2 3
2 1 1
⇒ p XY (0, 2) = − =
3 3 3
1 1
FXY (1, 0) = = ∑∑ p XY (u, v) = p XY (0, 0) + p XY (1, 0) = + p XY (1, 0)
6 u ≤1 v≤0 12
1 1 1
⇒ p XY (1, 0) = − =
6 12 12
7 5
FXY (1, 1) = = ∑∑ p XY (u, v) = p XY (0, 0) + p XY (0, 1) + p XY (1, 0) + p XY (1, 1) = + p XY (1, 1)
12 u ≤1 v≤1 12
7 5 1
⇒ p XY (1, 1) = − =
12 12 6
FXY (1, 2) = 1 = ∑∑ p XY (u , v) = p XY (0, 0) + p XY (0, 1) + p XY (0, 2) + p XY (1, 0) + p XY (1, 1) + p XY (1, 2)
u ≤1 v ≤ 2

11 11 1
= + p XY (1, 2) ⇒ p XY (1, 2) = 1 − =
12 12 12

5
 1
(a) P[0 < X < 2, 0 < Y < 2] = p XY (1, 1) =
6
(b) To obtain the marginal CDFs of X and Y, we first obtain their marginal PMFs:

 p (0, 0) + p XY (0, 1) + p XY (0, 2) x=0

p x ( x) = ∑ p XY ( x, y ) =  XY
y  p XY (1, 0) + p XY (1, 1) + p XY (1, 2) x =1
2
 3 x=0
=
1 x =1
 3
 p XY (0, 0) + p XY (1, 0) y=0

pY ( x) = ∑ p XY ( x, y ) =  p XY (0, 1) + p XY (1, 1) y =1
x  p (0, 2) + p (1, 2) y=2
 XY XY

1
6 y=0

5
= y =1
12
5
12 y=2


Thus, the marginal CDFs are given by

0 x<0

Fx ( x) = ∑ p X (u ) =  23 0 ≤ x <1
u≤ x 1 x ≥1

0 y<0
1 0 ≤ y <1
6
FY ( y ) = ∑ pY (v) =  7
v≤ y  12 1≤ y < 2
1 y≥2


1
(c) P[ X = 1, Y = 1] = p XY (1, 1) =
6
5.5 Two discrete random variables X and Y have the joint PMF defined as follows:

6
  121 x = 1, y = 1
1
6 x = 1, y = 2
1 x = 1, y = 3
 12
 16 x = 2, y = 1

p XY ( x, y ) =  14 x = 2, y = 2
1 x = 2, y = 3
 12
 121 x = 3, y = 1
1
 12 x = 3, y = 2
0 otherwise


Determine the following:

(a) the marginal PMFs of X and Y (i.e., p X (x) and pY ( y ))

(b) P[ X < 2.5]
(c) the probability that Y is odd
Solution:
(a) The marginal PMFs of X and Y are given by

 p XY (1, 1) + p XY (1, 2) + p XY (1, 3) x =1


p x ( x) = ∑ p XY ( x, y ) =  p XY (2, 1) + p XY (2, 2) + p XY (2, 3) x=2
y  p (3, 1) + p (3, 2) + p (3, 3) x=3
 XY XY XY

1
3 x =1

1
= x=2
2
1
6 x=3

 p XY (1, 1) + p XY (2, 1) + p XY (3, 1) y =1

pY ( x) = ∑ p XY ( x, y ) =  p XY (1, 2) + p XY (2, 2) + p XY (3, 2) y=2
x  p (1, 3) + p (2, 3) + p (3, 3) y=3
 XY XY XY

1
3 y =1

1
= y=2
2
1
6 y=3

7
 (b) P[ X < 2.5] = p X (1) + p X (2) = 5
6

(c) The probability that Y is odd is pY (1) + p X (3) = 1

2

5.6 Two discrete random variables X and Y have the joint PMF given by
0.2 x = 1, y = 1
0.1 x = 1, y = 2

0.1 x = 2, y = 1

p XY ( x, y ) = 0.2 x = 2, y = 2
0.1 x = 3, y = 1

0.3 x = 3, y = 2
0 otherwise


Determine the following:

(a) the marginal PMFs of X and Y (i.e., p X (x) and pY ( y ))
(b) the conditional PMF of X given Y, p X |Y ( x | y )
(c) whether X and Y are independent.
Solution:
(a) The marginal PMFs of X and Y are given by

 p XY (1, 1) + p XY (1, 2) x =1

p x ( x) = ∑ p XY ( x, y ) =  p XY (2, 1) + p XY (2, 2) x=2
y  p (3, 1) + p (3, 2) x=3
 XY XY

0.3 x =1

= 0.3 x=2
0.4 x=3

 p (1, 1) + p XY (2, 1) + p XY (3, 1) y =1
pY ( x) = ∑ p XY ( x, y ) =  XY
x  p XY (1, 2) + p XY (2, 2) + p XY (3, 2) y=2
0.4 y =1
=
0.6 y=2
(b) The conditional PMF of X given Y is given by
p XY ( x, y )
p X |Y ( x | y ) =
pY ( y )
(c) To test whether X and Y are independent, we proceed as follows:

8
  00..24 x =1
p ( x, 1) p XY ( x, 1)  0.1
P[ X | Y = 1] = p X |Y ( x | 1) = XY = =  0.4 x=2
pY (1) 0.4  0.1
 0.4 x=3
 12 x =1
1
= 4 x=2
1 x=3
4
 00..61 x =1
p ( x, 2) p XY ( x, 2)  0.2
P[ X | Y = 2] = p X |Y ( x | 2) = XY = =  0.6 x=2
pY ( 2) 0.6  0.3
 0.6 x=3
61
x =1
1
= 3 x=2
1 x=3
2

Since p X |Y ( x | 1) ≠ p X |Y ( x | 2), we conclude that X and Y are not independent. Note also that we
could have tested for independence by observing that p X ( x) pY ( y ) ≠ p XY ( x, y ), which also
shows that the random variables are not independent.

Section 5.4: Continuous Bivariate Random Variables

5.7 The joint PDF of two continuous random variables X and Y is given by
kx 0 < y ≤ x <1
f XY ( x, y ) = 
0 otherwise

(a) Determine the value of the constant k.

(b) Find the marginal PDFs of X and Y (i.e., f X (x) and f Y ( y ))
(c) Find P[0 < X < 12 , 0 < Y < 14 ]

Solution:
(a) To determine the value of the constant k, we must carefully define the ranges of the
values of X and Y as follows:

9
 Y
Y=X

Y> X

Y< X

X
1
∞ ∞ 1  x3 
x k 1
1 = ∫ ∫ f XY ( x, y )dxdy = ∫ ∫ kxdydx = ∫ kx dx = k   = 2
−∞ − ∞ x =0 y = 0 x =0
 3 0 3
which implies that k = 3. Note that we can also obtain k by reversing the order of
integration as follows:
1 1
1 1  x2  k 1 1 k y3 
1 = ∫ ∫ kxdxdy = ∫ k   dy = ∫ 1 − y 2 dy =  y −  =
y =0 x = y x =0 2 x =0
k
[ ]
 2 y 2 3 0 3
which gives the same result; that is, k = 3.

(b) The marginal PDFs of X and Y are given as follows:

∞ x
f X ( x) = ∫ f XY ( x, y )dy = ∫ kxdy = kx 2 = 3 x 2 0 ≤ x ≤1
y = −∞ y =0
1

f y ( y) = ∫
∞

x = −∞
1  x2  3
f XY ( x, y )dx = ∫ kxdx = k   = 1 − y 2
x= y
[ ] 0 < y ≤1
 2 y 2
(c) To evaluate P[0 < X < 12 , 0 < Y < 14 ], we need to find the region of integration as
follows:

10
 Y

1
4

1 1 1
X
0
4 2

Thus, we have that

1
x 1 1 1 1
kx
P[0 < X < 12 , 0 < Y < 14 ] = ∫ 4 ∫ kxdydx + ∫ 2 ∫
4
kxdydx = ∫ 4 kx 2 dx + ∫ 2 dx
x =0 y =0 x = 14 y =0 x =0 x = 14 4
 x 3  4  x 2  2 
1 1

   1  1  1  1 1  11
= k   +    = k    +  −  =
 3  0  8  14   3  64  8  4 16  128

5.8 The joint CDF of two continuous random variables X and Y is given by
1 − e − ax − e −by + e − ( ax +by ) x ≥ 0, y ≥ 0
FXY ( x, y ) = 
0 otherwise

(a) Find the marginal PDFs of X and Y

(b) Carefully show why or why not X and Y are independent.
Solution:
(a) We consider two methods of obtaining the marginal PDFs:
Method 1: We first obtain the joint PDF and then obtain the marginal PDFs as follows:

∂ 2 FXY ( x, y )
f XY ( x, y ) = = abe −( ax +by ) x ≥ 0, y ≥ 0
∂x∂y
∞ ∞
f X ( x) = ∫ f XY ( x, y )dy = abe −ax ∫ e −by dy = ae −ax x≥0
−∞ 0
∞ ∞
f Y ( y) = ∫ f XY ( x, y )dx = abe −by ∫ e − ax dx = be −by y≥0
−∞ 0

Method 2: We can first obtain the marginal CDFs and then differentiate then to obtain the
marginal PDFs as follows:

11
 FX ( x) = FXY ( x, ∞) = 1 − e − ax x≥0
FY ( y ) = FXY (∞, y ) = 1 − e −by y≥0
Thus,
dFX ( x)
f X ( x) = = ae − ax x≥0
dx
dF ( y )
f Y ( y) = Y = be −by y≥0
dy
{ }{ }
(b) We observe that f X ( x) f Y ( y ) = ae − ax be −by = abe − ( ax +by ) = f XY ( x, y ). That is, the
product of the marginal PDFs is equal to the joint PDF. Therefore, we conclude that X
and Y are independent random variables.

5.9 Two random variables X and Y have the joint PDF given by

ke − ( 2 x +3 y ) x ≥ 0, y ≥ 0
f XY ( x, y ) = 
0 otherwise

Determine the following:

(a) the value of the constant k that makes f XY ( x, y ) a true joint PDF
(b) the marginal PDFs of X and Y
(c) P[ X < 1, Y < 0.5]
Solution: A simple way to solve parts (a) and (b) jointly is by noting that the joint PDF is of the
form f XY ( x, y ) = k × a ( x) × b( y ) in the rectangular region 0 ≤ x < ∞, 0 ≤ y < ∞, where a(x) is
the x-factor and b( y ) is the y-factor. Therefore, X and Y are independent and we have that
f X ( x) = Ae −2 x x≥0
f Y ( y ) = Be −3 y y≥0
k = A× B
Since the PDFs are of the form f U (U ) = λe − λu , u ≥ 0, which is the PDF of an exponentially
distributed random variable U with parameter λ , we conclude that A = 2, B = 3 and k = 6.

A more formal solution is as follows:

∞ ∞
−2 x
∞∞
−3 y  1  1  k
(a) ∫− ∞ ∫−∞ f XY ( x, y )dxdy = 1 = k ∫
x =0
e dx ∫ y =e0 dy = k  2  3  = 6 ⇒ k = 6
(b) The marginal PDFs are given by
∞ ∞
f X ( x) = ∫ f XY ( x, y )dy = 6e − 2 x ∫ e −3 y dy = 2e − 2 x x≥0
−∞ 0
∞ ∞
f Y ( y) = ∫ f XY ( x, y )dx = 6e −3 y ∫ e − 2 x dx = 3e −3 y y≥0
−∞ 0

12
 (c) Because X and Y are independent, P[ X < 1, Y < 0.5] is given by
P[ X < 1, Y < 0.5] = P[ X < 1]P[Y < 0.5] = FX (1) FY (0.5) = 1 − e −2(1) 1 − e −3( 0.5) { }{ }
{
= 1− e −2
}{1 − e } = 0.6717
−1.5

5.10 Two random variables X and Y have the joint PDF given by
k (1 − x 2 y ) 0 ≤ x ≤ 1; 0 ≤ y ≤ 1
f XY ( x, y ) = 
0 otherwise

Determine the following:

(a) the value of the constant k that makes f XY ( x, y ) a true joint PDF
(b) the conditional PDFs of X given Y, f X |Y ( x | y ), and Y given X, f Y | X ( y | x)
Solution:
(a) To determine the value of k we know that
1

∫
∞

y = −∞ ∫
∞

x = −∞
1 1
{ } 
f XY ( x, y )dxdy = 1 = k ∫ ∫ 1 − x y dxdy = k ∫  x −
y =0 x = 0
2
y =0
x3 y 
3 0
1
 dy

1
1  y  y2   1  5k
= k ∫ 1 −  dy = k  y −  = k 1 −  =
y =0
 3  6 0  6 6
Thus, k = 1.2.
(b) To find the conditional PDFs, we must first find the marginal PDFs, which are given by

1
∞ 1  x2 y2   x2 
f X ( x) = ∫ f XY ( x, y )dy = k ∫ (1 − x y )dy = k  y −
2
 = 1. 2 1 −  0 ≤ x ≤ 1
y = −∞ y =0 2 0 2
 
1
∞ 1  x3 y   y
f Y ( y) = ∫ f XY ( x, y )dx = k ∫ (1 − x y )dx = k  x −
2
 = 1.21 −  0 ≤ y ≤1
x = −∞ y =0
 3 0  3
Thus, the conditional PDFs of X given Y, f X |Y ( x | y ), and Y given X, f Y | X ( y | x) are
given by

f XY ( x, y ) 1.2(1 − x 2 y ) 3(1 − x 2 y )
f X |Y ( x | y ) = = = 0 ≤ x ≤1
f Y ( y)  y 3− y
1.21 − 
 3
f XY ( x, y ) 1.2(1 − x 2 y ) 2(1 − x 2 y )
f Y | X ( y | x) = = = 0 ≤ y ≤1
f X ( x)  x2  2 − x2
1.21 − 
 2

13
5.11 The joint PDF of two random variables X and Y is given by
6 xy 
f XY ( x, y ) =  x 2 +  0 < x < 1, 0 < y < 2
7 2

(a) What is the CDF, FX (x), of X?

(b) Find P[ X > Y ]
(c) Find P[Y > 12 | X < 12 ]
Solution:
(a) To find the CDF, FX (x), of X, we first obtain its PDF as follows:
2
∞
f X ( x) = ∫
y = −∞
6 2  xy  6
f XY ( x, y )dy = ∫  x 2 + dy =  x 2 y +
7 y =0  2 7
xy 2 

4 0 7
6
(
= 2x 2 + x ) 0 < x <1

Thus, the CDF is given by

x

FX ( x ) = ∫
x

y = −∞
6 x
( ) 6  2u 3 u 2 
f X (u )du = ∫ 2u 2 + u du = 
7 y =0 7 3
+ 
2 0
0 x<0

 6  2x 3 x 2 
=   +  0 ≤ x <1
7  3 2 
1 otherwise


(b) To find P[ X > Y ], we proceed as follows. First, we need to define the region of
integration through the following figure.
Y

Y=X

X >Y

X
Thus,

14
 x
1 x 6 1 x  xy  6 1  xy 2 
P[ X > Y ] = ∫ ∫ f XY ( x, y )dydx = ∫ ∫  x 2 + dydx = ∫  x 2 y +  dx
x =0 y =0 7 x =0 y =0  2  7 x =0  4 0
1
6 1  x3  6 5 1  6 5  x 
4
15
= ∫  x 3 +  dx =  ×  ∫ x 3 dx =  ×    =
7 x =0  4  7 4  x =0  7 4   4  0 56

(c) P[Y > 12 | X < 12 ] is given by

6 12 2  2 xy 
P[Y > 12 , X < 12 ] P[Y > 12 , X < 12 ] 7 ∫ x =0 ∫ y = 12 
 x + dydx
2
P[Y > 2 | X < 2 ] =
1 1
= = = 0.8625
P[ X < 2 ]
1 1
FX ( 2 ) 1
FX ( 2 )

5.12 Two random variables X and Y have the joint PDF given by
ke − ( x + y ) x ≥ 0, y ≥ x
f XY ( x, y ) = 
0 otherwise

Determine the following:

(a) the value of the constant k that makes f XY ( x, y ) a true joint PDF
(b) P[Y < 2 X ]

Solution:
(a) The value of the constant k that makes f XY ( x, y ) a true joint PDF can be found as
follows. First, we determine the region of integration via the following figure:
Y

Y=X

Y> X

X>Y

X
Thus,

15
 ∞
∫ ∫
−∞
∞

−∞
f XY ( x, y )dxdy = 1 = k ∫
∞

y =0 ∫
y

x =0
e −( x + y ) dxdy = k ∫
∞

y =0
[ ] y
e − y − e − x 0 dy = k ∫
∞

y =0
[ ]
e − y 1 − e − y dy
∞

= k∫
∞

y =0
[e −y
−e −2 y
] 
dy = k − e − y +
e −2 y   1 k
 = k 1 −  = ⇒ k = 2
 2 0  2 2
(b) To evaluate P[Y < 2 X ], we graph the region of integration as shown in the following
figure:
Y

Y =2X Y =X

Y < 2X
Y >2X

X
0 1 2

Thus,
P[Y < 2 X ] = k ∫
∞

x =0 ∫
2x

y=x
e −( x + y ) dydx = k ∫
∞

x =0
[
e −x − e− y ]
2x
x dx = k ∫
∞

y =0
[ ]
e − x e − x − e − 2 x dx
∞

= k∫
∞

x =0
[e −2 x
−e −3 x
]  e − 2 x e −3 x 
dy = k − +
1 1 1 1
 = k  −  = 2  =
 2 3 0  2 3 6 3

5.13 The joint PDF of two random variables X and Y is given by

6x
f XY ( x, y ) = 1 ≤ x + y ≤ 2, x ≥ 0, y ≥ 0
7

(a) Without actually performing the integration, obtain the integral that expresses the
P[Y > X 2 ]. (That is, just give the exact limits of the integration.)
(b) In a very convincing way, obtain the exact value of P[ X > Y ].
16
Solution:
(a) To obtain the integral that expresses the probability P[Y > X 2 ], we must show the region
of integration, as illustrated in the figure below.
Y

X +Y =1
Y =X2

X +Y = 2

X
−1+ 5
2
From the figure we see that the region of integration is the hatched region, which has been
partitioned into two new areas labeled A and B. Area A is bounded by the line x = 0,
which is the y-axis; the line x = (−1 + 5 ) / 2, which is the feasible solution to the
simultaneous equations x + y = 1 and y = x 2 ; the line x + y = 1; and the line x + y = 2.
Similarly, area B is bounded by the curve y = x 2 , the line x + y = 2, and the line
x = (−1 + 5 ) / 2. Thus, the desired result is given by
P[Y > X 2 ] = ∫ f XY ( x, y )dxdy + ∫ f XY ( x, y )dxdy
A B

6
x dydx
( −1+ 5 ) / 2 2− x 1 2− x

7 ∫ x = 0 ∫
=  x dydx + ∫ ∫
y =1− x  x = ( −1+ 5 ) / 2 y= x2

(b) To obtain the exact value of P[ X > Y ], we note that the new region of integration is as
shown below:

17
 Y

X +Y = 2
X +Y =1
X=Y

Thus, we obtain the three areas labeled A, B, and C; and the desired result is as follows:
P[ X > Y ] = ∫ f XY ( x, y )dxdy + ∫ f XY ( x, y )dxdy + ∫ f XY ( x, y )dxdy

{∫ }
A B C

6 1 x 1 0.5 2 2− x
=
7 x = 0.5 ∫ y = 0.5
x dydx + ∫
x = 0.5 ∫ y =1− x
x dydx + ∫
x =1 ∫ y =0
x dydx

=
6
7
{∫ 1

x = 0.5
x[ x − 0.5]dx + ∫
1

x = 0.5
x[ x − 0.5]dx + ∫
2

x =1
x[2 − x]dx}

6  x 3 0.5 x 2   2 x 3   6  5
1 1 2
 x 3 0.5 x 2  5 2
=  −  + −  + x −   =  + + 
7  3 2  0.5  3 2  0.5  3  1  7  48 48 3 
 
6 7  3
=  =
7 8  4

5.14 Two random variables X and Y have the joint PDF given by
 1 −2 y
 e 0 ≤ x ≤ 4, y ≥ 0
f XY ( x, y ) =  2
0 otherwise

Determine the marginal PDFs of X and Y.

Solution: The marginal PDFs of X and Y are given by

18
 ∞
∞ 1 ∞ 1  e −2 y  1
f X ( x) = ∫ f XY ( x, y )dy = ∫ e −2 y dy = −  = 0≤ x≤4
−∞ 2 0 2  2 0 4
∞ 1 4
f Y ( y ) = ∫ f XY ( x, y )dx = e −2 y ∫ dx = 2e − 2 y y≥0
−∞ 2 0

Note that the joint PDF is completely separable in the form f XY ( x, y ) = k × a( x) × b( y ) in the
rectangular region 0 ≤ x ≤ 4, 0 ≤ y ≤ ∞, where k is a constant, a(x) is the x-factor and b( y ) is
the y-factor. Therefore, we must have that

f X ( x) = A 0≤x≤4
−2 y
f Y ( y ) = Be y≥0
1
= AB
2
To find the values of A and B we note that
∞ 4 1
∫ −∞
f X ( x)dx = 1 = A∫ dx = 4 A ⇒ A =
0 4
1
Therefore, B = 2
1
= 2, as before. Also, because Y has an exponential distribution, we have that
4

f Y ( y ) = λ e − λ y ⇒ λ = B = 2.

Section 5.6: Conditional Distributions

5.15 A box contains 3 red balls and 2 green balls. One ball is randomly selected from the box,
its color is observed and it is put back into the box. A second ball is then selected and goes
through the same process. Let the random variables X and Y be defined as follows: X = 0
if the first ball is green, and X = 1 if the first ball is red; Y = 0 if the second ball is green
and Y = 1 if the second ball is red.
(a) Find the joint PMF of X and Y.
(b) Find the conditional PMF of X given Y.
Solution: Let Ω denote the sample space of the experiment, R the event that a red ball is drawn,
and G the event that a green ball is drawn. Since the experiment is performed with replacement,
the outcomes of the two draws are independent and the probability of a sample point in Ω is the
product of the probabilities of those events. More importantly, since P[ R] = 3 / 5 = 0.6 and
P[G ] = 2 / 5 = 0.4, we obtain the following values for the probabilities of the sample points in Ω
with their corresponding values of X and Y:

19
 w∈Ω P[w] X Y
RR 0.36 1 1
RG 0.24 1 0
GR 0.24 0 1
GG 0.16 0 0

(a) The joint PMF p XY ( x, y ) of X and Y is given by

0.16 x = 0, y = 0
0.24 x = 0, y = 1

p XY ( x, y ) = 0.24 x = 1, y = 0
0.36 x = 1, y = 1

0.00 otherwise

(b) The conditional PMF of X given Y is given by

p XY ( x, y )
p X |Y ( x | y ) =
pY ( y )
But the marginal PMF of Y is given by
0.16 + 0.24 0.4 y=0
pY ( y ) = ∑ p XY ( x, y ) =  =
x 0.24 + 0.36 0.6 y =1
Thus, we obtain the following result:
0.4 x=0
p X |Y ( x | 0) = 
0.6 x =1
0.4 x=0
p X |Y ( x | 1) = 
0.6 x =1
5.16 Let the random variables X and Y have the joint PDF f XY ( x, y ) = 2e − ( x + 2 y ) , x ≥ 0, y ≥ 0.
Find the conditional expectation of
(a) X given Y
(b) Y given X

Solution: We know that

∞
E[ X | Y ] = ∫ xf X |Y ( x | y )dx
−∞
∞
E[Y | X ] = ∫ yf Y | X ( y | x)dy
−∞

{ }{ }
Observe that the region of interest is rectangular and f XY ( x, y ) = 2e − ( x + 2 y ) = e − x 2e −2 y ; that is,
the joint PDF is the product of two valid PDFs: f X ( x) = e − x and f Y ( y ) = 2e −2 y . Therefore, X
and Y are independent and we obtain
20
 ∞ ∞ ∞
E[ X | Y ] = ∫ xf X |Y ( x | y )dx = ∫ xf X ( x)dx = ∫ xe − x dx = E[ X ] = 1
−∞ 0 0

∞ ∞ ∞ 1
E[Y | X ] = ∫ yf Y | X ( y | x)dxy = ∫ yf Y ( y )dy = ∫ 2 ye − 2 x dy = E[Y ] =
−∞ 0 0 2
5.17 A fair coin is tossed four times. Let X denote the number of heads obtained in the first two
tosses, and let Y denote the number of heads obtained in the last two tosses.
(a) Find the joint PMF of X and Y
(b) Show that X and Y are independent random variables.
Solution: Let Ω denote the sample space of the experiment, H the event that a toss came up
heads, and T the event that a toss came up tails. Since the outcome of each toss is independent of
the outcomes of the other tosses, the probability of a sample point in Ω is the product of the
probabilities of the outcomes that constitute the event represented by the sample point. Thus, Ω ,
X and Y are given as follows:

w∈Ω P[w] X Y
HHHH 1
16
2 2
HHHT 1
16
2 1
HHTH 1
16
2 1
HHTT 1
16
2 0
HTHH 1
16
1 2
HTHT 1
16
1 1
HTTH 1
16
1 1
HTTT 1
16
1 0
THHH 1
16
1 2
THHT 1
16
1 1
THTH 1
16
1 1
TTHH 1
16
0 2
THTT 1
16
1 0
TTHT 1
16
0 1
TTTH 1
16
0 1
TTTT 1
16
0 0

(a) The joint PMF p XY ( x, y ) of X and Y is given by

21
  161 x = 0, y = 0
1
8 x = 0, y = 1
1 x = 0, y = 2
 16
 18 x = 1, y = 0
1
 x = 1, y = 1
p XY ( x, y ) =  14
8 x = 1, y = 2
1 x = 2, y = 0
 16
 18 x = 2, y = 1
1
 16 x = 2, y = 2
0 otherwise


(b) To show that X and Y are independent random variable we proceed as follows:
 161 + 18 + 161 = 14 x=0
1 1 1 1
 + + = x =1
p X ( x) = ∑ p XY ( x, y ) =  81 41 8 1 2 1
y  16 + 8 + 16 = 4 x=2
0 otherwise

 161 + 18 + 161 = 14 y=0
1 1 1 1
 + + = y =1
pY ( y ) = ∑ p XY ( x, y ) =  81 41 8 1 2 1
x  16 + 8 + 16 = 4 y=2
0 otherwise

Now,
 161 x = 0, y = 0
1
8 x = 0, y = 1
1 x = 0, y = 2
 16
 18 x = 1, y = 0
1
 x = 1, y = 1
p X ( x) pY ( y ) =  14
8 x = 1, y = 2
1 x = 2, y = 0
 16
 18 x = 2, y = 1
1
 16 x = 2, y = 2
0 otherwise

Since p X ( x) pY ( y ) = p XY ( x, y ), we conclude that X and Y are independent.

22
5.18 Two random variables X and Y have the joint PDF

f XY ( x, y ) = xye − y
2
/4
0 ≤ x ≤ 1, y≥0

(a) Find the marginal PDFs of X and Y.

(b) Determine if X and Y are independent.
Solution: There are two ways to solve the problem: brute force method and smart method. We
consider both methods.
Brute Force Method: The marginal PDFs are obtained as follows:
∞ ∞
f X ( x) = ∫ f XY ( x, y )dy = x ∫ ye − y
2
/4
dy
−∞ 0

Let u = y 2 / 4 ⇒ du = ydy / 2 ⇒ ydy = 2du. Thus,

∞
f X ( x) = x ∫ ye − y
0
2
/4
∞
dy = x ∫ 2e −u du = 2 x − e −u
0
[ ]
∞
0 = 2x 0 ≤ x ≤1

Similarly,
1
∞
− y2 / 4
1
− y2 / 4  x2  1 − y2 / 4
f Y ( y) = ∫ f XY ( x, y )dx = ye ∫ xdx = ye   = ye y≥0
−∞ 0
 2 0 2
To determine if X and Y are independent, we proceed as follows:

1 
f X ( x) f Y ( y ) = {2 x} ye − y / 4  = xye − y / 4 = f XY ( x, y )
2 2

2 
Thus, we conclude that X and Y are independent.

{ }
Smart Method: We observe that f XY ( x, y ) = {x} ye − y / 4 ; that is, it can be separated into an x-
2

function and a y-function. Since the region is rectangular, we conclude that X and Y are
independent. Thus,
f X ( x) = Ax 0 ≤ x ≤1
f Y ( y ) = Bye − y
2
/4
y≥0
1 = AB
The easiest approach is to evaluate A from the relationship:
1
∞  x2 1A
∫ −∞
f X ( x)dx = 1 = A∫ xdx = A  = ⇒ A = 2
0
 2 0 2
Thus, B = 1 / A = 1 / 2, as we obtained earlier.

5.19 Two random variables X and Y have the joint PDF

23
 6x
f XY ( x, y ) = 1 ≤ x + y ≤ 2, x ≥ 0, y ≥ 0
7

(a) Find the marginal PDFs of X and Y.

(b) Determine if X and Y are independent.
Solution: The region of interest is as shown in the following figure:

X +Y = 2
X +Y =1

(a) The marginal PDFs of X and Y are given by

 6 2− x
∞  7 ∫ y =1− xxdy 0 ≤ x <1
f X ( x) = ∫ f XY ( x, y )dy = 
−∞
 6 2− x xdy
 7 ∫ y =0
1≤ x < 2

 67 x 0 ≤ x <1
= 6
 7 x( 2 − x) 1≤ x < 2

24
  6 2− y
∞  7 ∫ x =1− yxdx 0 ≤ y <1
f Y ( y ) = ∫ f XY ( x, y )dx = 
−∞
 6 2− y xdx
 7 ∫ x =0
1≤ y < 2

 73 (3 − 2 y ) 0 ≤ y <1
= 3
 7 (2 − y ) 2 1≤ y < 2

(b) From the above results we observe that f X ( x) f Y ( y ) ≠ f XY ( x, y ). Therefore, we conclude

that X and Y are not independent.

Section 5.7: Covariance and Correlation Coefficient

5.20 Two discrete random variables X and Y have the joint PMF given by
0 x = −1, y = 0
1 x = −1, y = 1
3
 1 x = 0, y = 0
p XY ( x, y ) =  3
0 x = 0, y = 1
0 x = 1, y = 0
1
 3 x = 1, y = 1

(a) Are X and Y independent?

(b) What is the covariance of X and Y?
Solution:
(a) To determine if X and Y are independent we find their marginal PMFs as follows:
 13 x = −1
1
 x=0
p X ( x) = ∑ p XY ( x, y ) =  13
y 3 x =1
0 otherwise

 13 y=0

pY ( y ) = ∑ p XY ( x, y ) =  23 y =1
x 0 otherwise

From the above results we observe that p X ( x) pY ( y ) ≠ p XY ( x, y ). Therefore, X and Y are
not independent.
(b) The covariance of X and Y can be obtained as follows:

25
 1
E[ X ] = {− 1 + 0 + 1} = 0
3
1 2 2
E[Y ] = (0) + (1) =
3 3 3
1
E[ XY ] = ∑∑ xyp XY ( x, y ) = {(−1)(1) + (0)(0) + (1)(1)} = 0
x y 3
Cov ( x, y ) = σ XY = E[ XY ] − E[ X ] X [Y ] = 0

5.21 Two events A and B are such that P[ A] = 14 , P[ B | A] = 12 and P[ A | B] = 14 . Let the
random variable X be defined such that X = 1 if event A occurs and X = 0 if event A does
not occur. Similarly, let the random variable Y be defined such that Y = 1 if event B occurs
and Y = 0 if event B does not occur.
(a) Find E[ X ] and the variance of X.
(b) Find E[Y ] and the variance of Y.
(c) Find ρ XY and determine whether or not X and Y are uncorrelated.

Solution: First, we find P[B] and the PMFs of X and Y:

P[ AB] 1
P[ B | A] = ⇒ P[ AB ] = P[ B | A]P[ A] =
P[ A] 8
1
P[ AB] P[ AB] 1
P[ A | B] = ⇒ P[ B] = = 8
1
=
P[ B] P[ A | B ] 4 2
Note that because P[ A | B] = P[ A] and P[ B | A] = P[ B] events A and B are independent. Thus,
the random variables X and Y are independent. The PMFs of X and Y are given by
1 − P[ A] = 34 x=0

p X ( x) =  P[ A] = 14 x =1
0 otherwise

1 − P[ B] = 12 y=0

pY ( y ) =  P[ B ] = 12 y =1
0 otherwise

(a) The mean and variance of X are given by
1
E[ X ] = 34 (0) + 14 (1) =
4
1
E[ X 2 ] = 34 (0 2 ) + 14 (12 ) =
4
1 1 3
σ X2 = E[ X 2 ] − (E[ X ])2 = − =
4 16 16

26
 (b) The mean and variance of Y are given by
1
E[Y ] = 12 (0) + 12 (1) =
2
1
E[ X 2 ] = 12 (0 2 ) + 12 (12 ) =
2
1 1 1
σ X2 = E[ X 2 ] − (E[ X ])2 = − =
2 4 4

(c) As stated earlier, X and Y are independent because events A and B are independent. Thus,
σ XY = 0 and ρ XY = 0, which means that X and Y are uncorrelated.

5.22 A fair die is tossed three times. Let X be the random variable that denotes the number of 1’s
and let Y be the random variable that denotes the number of 3’s. Find the correlation
coefficient of X and Y.
Solution: Let A denote the event that an outcome of the toss is neither 1 nor 3. Then the sample
space of the experiment and the values of X and Y are shown in the following table.

w∈Ω P[w] X Y
1 111 ( ) = 2161
1 3
6
3 0
2 113 ( ) = 2161
1 3
6
2 1
3 11A (16 )2 ( 23 ) = 1082 = 541 2 0
4 1A1 ( 16 )2 ( 23 ) = 1082 = 541 2 0
5 131 ( 16 )3 = 2161 2 1
6 1AA ( 16 )( 23 )2 = 272 1 0
7 1A3 (16 )2 ( 23 ) = 1082 = 541 1 1
8 133 ( 16 )3 = 2161 1 2
9 13A (16 )2 ( 23 ) = 1082 = 541 1 1
10 333 ( 16 )3 = 2161 0 3
11 33A ( 16 )2 ( 23 ) = 1082 = 541 0 2
12 331 ( 16 )3 = 2161 1 2
13 3A3 ( 16 )2 ( 23 ) = 1082 = 541 0 2
14 313 ( 16 )3 = 2161 1 2
15 3AA ( 16 )( 23 )2 = 272 0 1
16 3A1 ( 16 )2 ( 23 ) = 1082 = 541 1 1
17 311 ( 16 )3 = 2161 2 1

27
 18 31A ( 16 )2 ( 23 ) = 1082 = 541 1 1
19 AAA ( 23 )3 = 278 0 0
20 AA1 ( 16 )( 23 )2 = 272 1 0
21 AA3 ( 16 )( 23 )2 = 272 0 1
22 A1A ( 16 )( 23 )2 = 272 1 0
23 A3A ( 16 )( 23 )2 = 272 0 1
24 A11 ( 16 )2 ( 23 ) = 1082 = 541 2 0
25 A13 ( 16 )2 ( 23 ) = 1082 = 541 1 1
26 A33 ( 16 )2 ( 23 ) = 1082 = 541 0 2
27 A31 (16 )2 ( 23 ) = 1082 = 541 1 1

The PMFs of X and Y are given by

 216
1
+ 3( 541 ) + 3( 272 ) + 278 = 125
216 x=0
 1
3( 216 ) + 6( 54 ) + 3( 27 ) = x =1
1 2 75
p X ( x) =  1
216

3( 216 ) + 3( 54 ) = x=2

1 15
216
1 = 1
x=3
 216 216

 216
1
+ 3( 541 ) + 3( 272 ) + 278 = 125
216 y=0
 1
3( 216 ) + 6( 54 ) + 3( 27 ) = y =1
1 2 75
pY ( y ) =  1
216

3( 216 ) + 3( 54 ) = y=2

1 15
216
1 = 1
y=3
 216 216

Finally, the joint PMF of X and Y is given by

28
  216
64
x = 0, y = 0
 48
 216 x = 0, y = 1
 12 x = 0, y = 2
 216
 216
1
x = 0, y = 3
 48
 216 x = 1, y = 0
 24
p XY ( x, y ) =  216 x = 1, y = 1
3 x = 1, y = 2
 216
 216
12
x = 2, y = 0
3
 216 x = 2, y = 1
1 x = 3, y = 0
 216
0.00 otherwise

Thus, the correlation coefficient of X and Y, ρ XY , can be obtained as follows:

0(125) + 1(75) + 2(15) + 3(1) 1

E[ X ] = E[Y ] = =
216 2
0 (125) + 1 (75) + 2 (15) + 3 (1) 2
2 2 2 2
E[ X 2 ] = E[Y 2 ] = =
216 3
2 1 5
σ X2 = σ Y2 = E[ X 2 ] − (E[ X ])2 = − =
3 4 12
(1)(1)(24) + (1)(2)(3) + (2)(1)(3) 36 1
E[ XY ] = = =
216 216 6
1 1 1
σ XY = E[ XY ] − E[ X ]E[Y ] = − = −
6 4 12
σ − 1
1
ρ XY = XY = 512 = − = −0.2
σ XσY 12 5

Section 5.9: Multinomial Distributions

5.23 A box contains 10 chips from supplier A, 16 chips from supplier B, and 14 chips from
supplier C. Assume that we perform the following experiment 20 times: We draw one chip
from the box, note the supplier from where it came and put it back into the box. What is the
probability that a chip from vendor B is drawn 9 times?

Solution: Let p A denote the probability that a chip is from supplier A, p B the probability that a
chip is from supplier B and pC the probability that a chip is from supplier C. Then,

29
 10
pA = = 0.25
40
16
pB = = 0.40
40
14
pC = = 0.35
40

Let K be a random variable that denotes the number of time that a chip from supplier B is drawn
in 20 trials. Then K is a binomially distributed random variable with the PMF

 20 
p K (k ) =   p Bk (1 − p B ) 20−k k = 0, 1, Κ , 20
k 

Therefore, the probability that a chip from supplier B is drawn 9 times in 20 trials is given by

 20 
P[ K = 9] = p K (9) =   (0.4) (0.6)11 = 0.1597
9

9 

5.24 With reference to the previous problem, what is the probability that a chip from vendor A is
drawn 5 times and a chip from vendor C is drawn 6 times?
Solution: If a chip from supplier A is drawn 5 times and a chip from supplier C is drawn 6 times
in 20 trials, then a chip from supplier B is drawn 20 − (5 + 6) = 9 times in those 20 trials. Thus,
the probability of this event is given by

 20  5 9 6 20!
p =   p A p B pC = (0.25)5 (0.4)9 (0.35)6 = 0.0365
5 9 6 5! 9! 6!

5.25 The students in one college have the following rating system for their professors: excellent,
good, fair, and bad. In a recent poll of the students, it was found that they believe that 20%
of the professors are excellent, 50% are good, 20% are fair, and 10% are bad. Assume that
12 professors are randomly selected from the college.
(a) What is the probability that 6 are excellent, 4 are good, 1 is fair, and 1 is bad?
(b) What is the probability that 6 are excellent, 4 are good, and 2 are fair?
(c) What is the probability that 6 are excellent and 6 are good?
(d) What is the probability that 4 are excellent and 3 are good?
(e) What is the probability that 4 are bad?
(f) What is the probability that none is bad?

30
Solution: Let p E denote the probability that a professor is rated excellent, pG the probability
that a professor is rated good, p F the probability that a professor is rated fair, and p B the
probability that a professor is rated bad. Then we are given that
p E = 0.2
pG = 0.5
p F = 0.2
PB = 0.1
Given that 12 professors are randomly selected from the college, then
(a) The probability that 6 are excellent, 4 are good, 1 is fair, and 1 is bad is given by

 12  6 4 1 1 12!
p1 =   p E pG p F p B = (0.2)6 (0.5)4 (0.2)1 (0.1)1 = 0.0222
 6 4 1 1 6! 4! 1!1!

(b) The probability that 6 are excellent, 4 are good, and 2 are fair is the probability that 6 are
excellent, 4 are good, 2 are fair and none is bad and is given by

 12  6 4 2 0 12!
p 2 =   p E pG p F p B = (0.2)6 (0.5)4 (0.2)2 (0.1)0 = 0.0044
 6 4 2 0 6! 4! 1!1!
(c) The probability that 6 are excellent and 6 are good is the probability that 6 are excellent,
6 are good and none is either fair or bad and is given by

 12  6 6 0 0 12!
p3 =   p E p G p F p B = (0.2)6 (0.5)6 (0.2)0 (0.1)0 = 0.000924
 6 6 0 0 6! 6! 0!0!

(d) The probability that 4 are excellent and 3 are good is the probability that 4 are excellent,
3 are good and 5 are either fair or bad and is given by

 12  4 3 12!
p 4 =   p E pG ( p F + p B )5 = (0.2)4 (0.5)3 (0.3)5 = 0.0135
 4 3 5 4! 3! 5!

(e) The probability that 4 are bad is the probability that 4 are bad and 8 are not bad and is
given by

12  12!
p5 =   p B4 (1 − p B ) = (0.1)4 (0.9)8 = 0.0213
8

4  4! 8!

(f) The probability that none is bad is the probability that the 12 are not bad and is given by

12  12!
p 6 =   p B0 (1 − p B ) = (0.1)0 (0.9)12 = (0.9)12 = 0.2824
12

0  0! 12!

31
5.26 Studies on the toasters made by a company indicate that 50% are good, 35% are fair, 10%
burn the toast, and 5% catch fire. If a store has 40 of these toasters in stock, determine the
following probabilities:
(a) 30 are good, 5 are fair, 3 burn the toast, and 2 catch fire
(b) 30 are good and 4 are fair
(c) None catches fire
(d) None burns the toast and none catches fire

Solution: Let p G denote the probability that a toaster is good, p F the probability that a toaster is
rated fair, p B the probability that a toaster burns the toast, and pC the probability that a toaster
can catch fire. Then we are given that
pG = 0.50
p F = 0.35
p B = 0.10
PC = 0.05
Given that a store has 40 toasters in stock, then
(a) The probability that 30 are good, 5 are fair, 3 burn the toast, and 2 catch fire is given by

 40  30 5 3 2 40!
p1 =   pG p F p B pC = (0.50)30 (0.35)5 (0.10)3 (0.05)2 = 0.000028
 30 5 3 2  30! 5! 3!2!

(b) The probability that 30 are good and 4 are fair is the probability that 30 are good, 4 are
fair and 6 are either bad or can catch fire, which is given by

 40  30 4 40!
p 2 =   pG p F ( p B + pC )6 = (0.50)30 (0.35)4 (0.15)6 = 0.00034
 30 4 6  30! 5! 3!2!

(c) The probability that none catches fire is given by

 40  40!
p3 =   pC0 (1 − pC ) = (0.05)0 (0.95)40 = (0.95)40 = 0.1285
40

0  0! 40!

(d) The probability that none burns the toast and none catches fire is given by

 40  40!
p 4 =  ( p B + pC ) (1 − p B − pC ) = (0.15)0 (0.85)40 = (0.85)40 = 0.0015
0 40

0  0! 40!

5.27 Ten pieces of candy are given out at random to a group that consists of 8 boys, 7 girls, and
5 adults. If anyone can get more than one piece of candy, find the following probabilities:
(a) 4 pieces go to the girls and 2 go to the adults
(b) 5 pieces go to the boys

32
Solution: Let p B denote the probability that a candy goes to a boy, p G the probability that a
candy goes to a girl, and p A the probability that a candy goes to an adult. Then,

8
pB = = 0.40
20
7
pG = = 0.35
20
5
pA = = 0.25
20

Given that 10 pieces of candy are given out at random to the group, we have that
(a) The probability that 4 pieces go to the girls and 2 go to the adults is the probability that 4
pieces go to the girls, 2 go to the adults and 4 go to the boys, which is given by

 10  4 4 2 10!
p1 =   p B pG p A = (0.40)4 (0.35)4 (0.25)2 = 0.0756
 4 4 2 4! 4! 2!

(b) The probability that 5 pieces go to the boys is the probability that 5 pieces go to the boys
and 5 pieces go to others (i.e., either the girls or the adults), and is given by

10  10!
p 2 =   p B5 (1 − p B ) = (0.40)5 (0.60)5 = 0.20066
5

5  5! 5!

33
Chapter 6: Functions of Random Variables

Chapter Summary: This chapter deals with functions of random variables including linear and
power functions of one random variable, moments of functions of one random variable, sums of
independent random variables, the maximum and minimum of two independent random
variables, two functions of two random variables, laws of large numbers, the central limit
theorem, and order statistics.

Section 6.2: Functions of One Random Variable

6.1 Suppose X is a random variable and Y = aX − b, where a and b are constants. Find the
PDF, expected value and variance of Y.
Solution: The PDF of Y can be obtained as follows:

 y + b  y+b
FY ( y ) = P[Y ≤ y ] = P[aX − b ≤ y ] = P  X ≤  = FX  
 a   a 
d d  y+b 1  y+b
fY ( y) = FY ( y ) = FX  = fX  
dy dy  a  | a |  a 

The expected value and variance of Y are given by

E[Y ] = E[aX − b] = E[aX ] − E[b] = aE[ X ] − b

[ ] [ ] [
σ Y2 = E (Y − E[Y ])2 = E (aX − b − aE[ X ] + b )2 = E (aX − aE[ X ])2 ]
= E [a ( X − E[ X ]) ] = a E [( X − E[ X ]) ]
2 2 2 2

= a 2σ X2

6.2 If Y = aX 2 + b, where a > 0 is a constant, b is a constant and the PDF of X, f X (x), is

known, find
(a) the PDF of Y
(b) the PDF of Y when f X (x) is an even function

Solution:
(a) The PDF of Y can be obtained as follows:

1
  y − b  y −b
FY ( y ) = P[Y ≤ y ] = P[aX 2 + b ≤ y ] = P  X 2 ≤ = P X ≤ 
 a   a 
 y −b y −b  y −b   y −b 
= P − <X <  = FX   − FX  − 
 a a   a  
 a 

d d   y − b   y − b 
f Y ( y) = FY ( y ) =  FX   − FX  − 
dy dy   a  
 a 

1   y − b   y − b 
=  f X   + fX −  y>0
2 ay   a  
 a 

(b) When f X (x) is an even function, f X ( x) = f X (− x) and the PDF of Y becomes

2   y − b  1  y −b 
f Y ( y) =  f X   =
 f X   y>0
2 ay   a  ay  a 

6.3 If Y = aX 2 , where a > 0 is a constant and the mean and other moments of X are known,
determine the following in terms of the moments of X:
(a) the mean of Y
(b) the variance of Y.
Solution:
{
(a) E[Y ] = E[aX 2 ] = aE[ X 2 ] = a σ X2 + (E[ X ])
2
}
{[
(b) σ Y2 = E[Y 2 ] − (E[Y ]) = E[a 2 X 4 ] − a σ X2 + (E[ X ])
2 2
]}
2
{ [
= a 2 E[ X 4 ] − σ X2 + (E[ X ]) ]}
2 2

6.4 If Y = | X | and the PDF of X, f X (x), is known, find the PDF of Y in terms of f X (x).

Solution: The PDF of Y can be obtained as follows:

FY ( y ) = P[Y ≤ y ] = P[| X |≤ y ] = P[− y ≤ X ≤ y ] = FX ( y ) − FX (− y )

d
f Y ( y) = FY ( y ) = f X ( y ) + f X (− y )
dy

6.5 The random variable X has the following PDF:

1 −1 < x < 2
f X ( x) =  3
0 otherwise

If we define Y = 2 X + 3, what is the PDF of Y?

 y −3
Solution: We know that f Y ( y ) = 1
2 fX  . with f X (x) as defined, we obtain
 2 
2
 1
 1< y < 7
f Y ( y) =  6
0 otherwise

where the range of Y follows from the fact that 2(−1) + 3 = 1 and 2(2) + 3 = 7.

6.6 Assume that Y = a X , where a > 0 is a constant and the PDF of X, f X (x), is known.
(a) Determine the PDF of Y in terms of the PDF of X
(b) Find the PDF of Y for the special case where Y = e X and the PDF of X is given by
1 0 < x <1
f X ( x) = 
0 otherwise

Solution:
(a) The PDF of Y can be obtained as follows:

 ln( y )   ln( y ) 
FY ( y ) = P[Y ≤ y ] = P[a X ≤ y ] = X ln(a ) ≤ ln( y )] = P  X ≤  = FX  
 ln(a )   ln(a ) 
d 1  ln( y ) 
f Y ( y) = FY ( y ) = f X   y>0
dy y ln(a )  ln(a ) 

(b) When a = e, ln(a) = ln(e) = 1 and the PDF of Y becomes

1
f X (ln( y ) )
f Y ( y) = y>0
y
When f X (x) is as defined, we obtain
1
 0< y<e
f Y ( y) =  y
0 otherwise

where the limits follow from the solution to the equation ln( y ) = 0 ⇒ y = e 0 = 1 and the
equation ln( y ) = 1 ⇒ y = e1 = e.
6.7 Assume that Y = ln X , where the PDF of X, f X (x), is known. Find the PDF of Y in terms
of the PDF of X.
Solution: The PDF of Y can be obtained as follows:
FY ( y ) = P[Y ≤ y ] = P[ln( X ) ≤ y ] = P[ X ≤ e y ] = FX (e y )
d
f Y ( y) = FY ( y ) = e y f X (e y )
dy

3
Section 6.4: Sums of Random Variables
6.8 A random variable X has the PDF f X ( x) = 2 x, 0 ≤ x ≤ 1, and 0 elsewhere. Independently of
X, a random variable Y is uniformly distributed between − 1 and 1. Let the random
variable W = X + Y . Determine the PDF of W.

Solution: We have that

f X ( x) = 2 x 0 ≤ x ≤1
1 −1 ≤ y ≤ 1
f Y ( y) =  2
0 otherwise
Since X and Y are independent, we have that
∞
f W ( w) = f X ( w) ∗ f Y ( w) = ∫ f X ( w − y ) f Y ( y )dy
−∞

To evaluate the integral we carry out the following convolution operations:

fX (x) f Y ( y)

1
2

Case 1: In the range − 1 ≤ w ≤ − 34 , we have the convolution integral as the shaded area:

4
 f Y ( y)

f X (w− y)

1
2

h
-1 w y
-2 0 1

Thus, in this case, from similar triangles properties we have that h = 2( w + 1); thus, we obtain

1
f W ( w) = h{w − (−1)} = ( w + 1) 2
2

Case 2: In the range − 34 ≤ w ≤ 0, we have the convolution integral as the shaded area:

f Y ( y)

f X (w− y)

1
2

w − 14

In this case we have that

1 1  1  1  1  7 w
f W ( w) = A + B = w − − (−1) + w −  w −   = +
2 4  2  4  2  16 2
5
Case 3: In the range 0 ≤ w ≤ 1, we have the convolution integral as the shaded area:

f Y ( y)

f X (w− y)

A 1 B
2

y
-2 -1 w −1 0 w 1
w − 14

In this case we have that

1 1  1  1  1  7
f W ( w) = A + B = w − − ( w − 1) + w −  w −   =
2 4  2  4  2  16

Case 4: In the range 1 ≤ w ≤ 54 , we have the convolution integral as the areas A and B. In this
case area B is trapezoidal.

f Y ( y)

f X (w− y)

A
1
B
2

2( w − 1)
y
-2 -1 0 1w

w −1 w − 14
6
Since B is trapezoidal, we have that

1 1  1 1   1  7
f W ( w) = A + B = w − − ( w − 1) +  + 2( w − 1)1 −  w −  = − ( w − 1) 2
2 4  2 2   4  16

Case 5: In the range 5

4 ≤ w ≤ 2, we have the convolution integral as the shaded area:

f Y ( y)

f X (w− y)

1 A
2

y
-2 -1 0 w −1 1 w 2

In this case, we have that

1 1  w
f W ( w) = A = w − − ( w − 1) = 1 −
2 4  2

Thus, the PDF of W is given by

( w + 1) 2 − 1 ≤ w ≤ − 34

7 +w − 34 ≤ w ≤ 0
16 2
7
 0 ≤ w ≤1

f W ( w) = 16
 7 − ( w − 1) 2 1≤ w ≤ 5
16 4


1 − w 5
≤w≤2
 2 4

0 otherwise


7
6.9 X and Y are two independent random variables with PDFs

f X ( x) = 4e −4 x x≥0
−2 y
f Y ( y ) = 2e y≥0

If we define the random variable U = X + Y , find

(a) the PDF of U

(b) P[U > 0.2]
Solution:
(a) Since X and Y are independent, we can obtain the PDF of U as follows:
∞
f U (u ) = f X (u ) ∗ f Y (u ) = ∫ f X (u − y ) f Y ( y )dy
−∞

Since f X ( x) = 0 for x < 0, we must have that u − y ≥ 0 ⇒ y ≤ u. Thus

f U (u ) = ∫
∞

−∞
f X (u − y ) f Y ( y )dy = ∫
y =0
u
{4e − 4(u − y )
}{2e }dy = 8e ∫
−2 y −4u
u

y =0
e 2 y dy
u

= 8e −4u e2y 

2
−4u
{−2u
 = 4e e − 1 = 4 e − e
2u −4u
} { } u≥0
 0
(b) The probability that U is greater than 0.2 is given by
∞
 e − 2u e − 4 u 
P[U > 0.2] = ∫
∞

u = 0.2
f U (u )du = 4∫
∞

u = 0.2
{e −2u
−e −4u
}
du = 4 − + 
 2 4  0.2
= 2e −0.4 − e −0.8 = 0.8913

6.10 Suppose we roll two dice. Let the random variables X and Y denote the numbers that appear
on the dice. What is the expected value of X + Y ?

Solution: The PMFs and expected values of X and Y are given by

1
p X ( x) = x = 1, 2, Κ , 6
6
1
pY ( y ) = y = 1, 2, Κ , 6
6
1 21 7
E[ X ] = E[Y ] = {1 + 2 + 3 + 4 + 5 + 6} = =
6 6 2

Let U = X + Y . Then the expected value of U is E[U ] = E[ X ] + E[Y ] = 7.

6.11 Assume that X is a random variable that denotes the sum of the outcomes of two tosses of a
fair coin. Denote the outcome “heads” by 1 and the outcome “tails” by 0. What is the
expected value of X?
8
Solution: The sample space of the experiment is as follows:

w∈Ω P[w] X
HH 0.25 2
HT 0.25 1
TH 0.25 1
TT 0.25 0

Thus, the PMF of X is as follows:

0.25 x=0
0.50 x =1

p X ( x) = 
0.25 x=2
0.00 otherwise

Therefore, the expected value of X is E[ X ] = 0(0.25) + 1(0.50) + 2(0.25) = 1.

6.12 Suppose we select 4 students at random from a class of 10 boys and 12 girls. Let the
random variable X denote the number of boys selected, and let the random variable Y
denote the number of girls selected. What is E[ X − Y ] ?

Solution:

The PMF of X, which is the probability of selecting x boys and hence 4 − x girls, is given by

10  12 
  
 x  4 − x 
p X ( x) = x = 0, 1, 2, 3, 4
 22 
 
4 

The sample space for the experiment and the values of X and Y are shown in the following table:

X Y P[ X ] X −Y
0 4 0.0677 -4
1 3 0.3007 -2
2 2 0.4060 0
3 1 0.1969 2
4 0 0.0287 4

Let U = X − Y . Then from the table above, the PMF of U is given by

9
 0.0677 u = −4
0.3007 u = −2

0.4060 u=0
pU (u ) = 
0.1969 u=2
0.0287 u=4

0.00 otherwise

Thus, the expected value of U is

E[U ] = −4(0.0677) − 2(0.3007) + 0(0.4060) + 2(0.1969) + 4(0.0287) = −0.3636

Note that we can also obtain the same result by noting that E[U ] = E[ X ] − E[Y ]. The PMF of Y
is simply given by pY ( y ) = p X (4 − x). Thus,

E[ X ] = 0 p X (0) + 1 p X (1) + 2 p X (2) + 3 p X (3) + 4 p X (4)

= 0(0.0677) + 1(0.3007) + 2(0.4060) + 3(0.1969) + 4(0.0287) = 1.8182
E[Y ] = 0 pY (0) + 1 pY (1) + 2 pY (2) + 3 pY (3) + 4 pY (4)
= 0(0.0287) + 1(0.1969) + 2(0.4060) + 3(0.3007) + 4(0.0677) = 2.1818
E[U ] = E[ X ] − E[Y ] = −0.3636

6.13 Suppose we put 8 balls randomly into 5 boxes. What is the expected number of empty
boxes?

Solution: Let p denote the probability that a ball is put in a tagged box. Then p = 15 = 0.2. Let
X k be a random variable that has a value 1 if the kth box contains no ball and 0 otherwise. Then
the PMF of X k is given by

(1 − p )8 x =1 (0.8)8 x =1
p X k ( x) =  ⇒ p X k ( x) = 
1 − (1 − p )8 x=0 1 − (0.8)8 x=0

Thus, E[ X k ] = 1(0.8) 8 + 0{1 − (0.8) 8 } = (0.8) 8 = 0.1678. The number of empty boxes is given by
X = X 1 + X 2 + X 3 + X 4 + X 5 . Therefore, the expected value of X is

E[ X ] = E[ X 1 + X 2 + X 3 + X 4 + X 5 ] = 5 E[ X 1 ] = 5(0.1678) = 0.8388

6.14 Two coins A and B are used in an experiment. Coin A is a biased coin that has a
probability of heads equal to 1/4 and a probability of tails equal to 3/4. Coin B is a fair
coin. Each coin is tossed four times. Let X be the random variable that denotes the number

10
 of heads resulting from coin A, and let Y be the random variable that denotes the number of
heads resulting from coin B. Determine the following:
(a) The probability that X = Y
(b) The probability that X > Y
(c) The probability that X + Y ≤ 4

Solution: Let p A denote the probability that coin A comes up heads and p B the probability that
coin B comes up heads. Then we have that p A = 14 and p B = 12 . Since X denotes the number of
heads resulting from 4 tosses of coin A and Y denotes the number of heads in 4 tosses of coin B,
the PMFs of X and Y are given by
0.3164 x=0
0.4219 x =1
4 
p X ( x) =   p Ax (1 − p A ) 4− x = 0.2109 x=2
 x 0.0469 x=3

0.0039 x=4
0.0625 y=0
0.2500 y =1
4  
pY ( y ) =   p Ay (1 − p A ) 4− y = 0.3750 y=2
 y 0.2500 y=3

0.0625 y=4
Since X and Y are independent, their joint PMF p XY ( x, y ) = p X ( x) pY ( y ). Thus,
(a) The probability that X = Y is given by
P[ X = Y ] = P[ X = 0, Y = 0] + P[ X = 1, Y = 1] + P[ X = 2, Y = 2] + P[ X = 3, Y = 3]
+ P[ X = 4, Y = 4]
= p X (0) pY (0) + p X (1) pY (1) + p X (2) pY (2) + p X (3) pY (3) + p X (4) pY (4) = 0.2163

(b) The event {X > Y } is defined by

{X > Y } = {X = 1, Y = 0} ∪ {X = 2, Y = 0} ∪ {X = 2, Y = 1} ∪ {X = 3, Y = 0} ∪ {X = 3, Y = 1} ∪
{X = 3, Y = 2} ∪ {X = 4, Y = 0} ∪ {X = 4, Y = 1} ∪ {X = 4, Y = 2} ∪ {X = 4, Y = 3}
Since these events are mutually exclusive, the probability of the event {X > Y } is the sum
of the probabilities of these events. Since the CDF of Y is defined by
y
FY ( y ) = P[Y ≤ y ] = ∑ pY ( y )
k =0

we have that

11
 FY (0) = pY (0) = 0.0625
FY (1) = pY (0) + pY (1) = 0.3125
FY (2) = pY (0) + pY (1) + pY (2) = 0.6875
FY (3) = pY (0) + pY (1) + pY (2) + pY (3) = 0.9375
Thus,
P[ X > Y ] = p X (1) FY (0) + p X (2) FY (1) + p X (3) FY (2) + p X (4) FY (3) = 0.1282

(c) The event X + Y ≤ 4 is defined by

{X + Y ≤ 4} = {X = 0, Y = 0} ∪ {X = 0, Y = 1} ∪ {X = 0, Y = 2} ∪ {X = 0, Y = 3} ∪ {X = 0, Y = 4} ∪
{X = 1, Y = 0} ∪ {X = 1 Y = 1} ∪ {X = 2, Y = 2} ∪ {X = 1, Y = 3} ∪
{X = 2, Y = 0} ∪ {X = 2, Y = 1} ∪ {X = 2, Y = 2} ∪ {X = 3, Y = 0} ∪ {X = 3, Y = 1}
{X = 4, Y = 0}

Since these events are mutually exclusive, the probability of the event {X + Y ≤ 4} is the
sum of the probabilities of these events. That is,

P[ X + Y ≤ 4] = p X (0) + p X (1){1 − pY (4)} + p X (2){1 − pY (3) − p( Y 4)}

+ p X (3){pY (0) + pY (1)} + p X (4) pY (0) = 0.8718

6.15 Two random variables X and Y have the joint PDF given by

f XY ( x, y ) = 4 xy 0 < x < 1, 0 < y < 1

If we define the random variable U = X + Y , find the PDF of U.

Solution: Since the joint PDF is separable and the region is rectangular, we conclude that X and
Y are independent and their marginal PDFs are given by

f X ( x) = Ax 0 < x <1
f Y ( y ) = By 0 < y <1
4 = AB
We obtain the parameter A as follows:
1
1 1  x2  A
∫ x =0 X
f ( x )dx = 1 = ∫ x =0
Axdx = A   = ⇒ A=2
 2 0 2
Thus, B = 2 and the marginal PDFs become

12
 f X ( x) = 2 x 0 < x <1
f Y ( y) = 2 y 0 < y <1

Note that the marginal PDFs can also be obtained by the traditional method of integrating the
joint PDF over x and y and the independence of X and Y can be established by showing that the
product of the marginal PDFs is equal to the joint PDF. Since the random variables are
independent, the PDF of U = X + Y is given by
∞
f U (u ) = f X (u ) ∗ f Y (u ) = ∫ f X (u − y ) f Y ( y)dy
−∞

Since f X (x) is defined to be nonzero only over the range 0 < x < 1, we have that 0 < u − y < 1,
which implies that 0 < y < u and u − 1 < y < 1. Thus, we obtain

 u4(u − y ) ydy
∫ 0
0 < u <1
f U (u ) =  2
∫ 4(u − y ) ydy 1< u < 2
 u −1
 2u 3
 0 < u <1
= 3
(
 2 6u − u 3 − 4
 3
) 1< u < 2

Sections 6.4 and 6.5: Maximum and Minimum of Independent Random Variables

6.16 Suppose we roll two dice. Let the random variables X and Y denote the numbers that appear
on the dice. What is the expected value of
(a) max ( X , Y )
(b) min ( X , Y ) ?

Solution: Let X denote the number that appears on the first die and let Y the number that appears
on the second die. Then the sample space of the experiment is given as follows:

13
 Y

Y>X

6 ∗ ∗ ∗ ∗ ∗ ∗
5
∗ ∗ ∗ ∗ ∗ ∗
4 ∗ ∗ ∗ ∗ ∗ ∗
3 ∗ ∗ ∗ ∗ ∗ ∗ X>Y

2 ∗ ∗ ∗ ∗ ∗ ∗
1 ∗ ∗ ∗ ∗ ∗ ∗
X
1 2 3 4 5 6
(a) Let W = max ( X , Y ). Then the PMF of W is given by
 361 w =1
3
 36 w=2
5 w=3
 36
pW ( w) =  367 w=4
9 w=5
 36
 11
36
w=6

0 otherwise
Thus, the expected value of W is given by

1
E[W ] = {1(1) + 2(3) + 3(5) + 4(7) + 5(9) + 6(11)} = 161 = 4.4722
36 36

(b) Let V = min ( X , Y ). Then the PMF of V is given by

14
  11
36
v =1
9
 36 v=2
7 v=3
 36
pV (v) =  365 v=4
3 v=5
 36
 361 v=6

0 otherwise
Thus, the expected value of V is given by

1
E[V ] = {1(11) + 2(9) + 3(7) + 4(5) + 5(3) + 6(1)} = 91 = 2.5278
36 36

6.17 A system consists of two components A and B that are connected in series. If the lifetime
of A is exponentially distributed with a mean 200 hours and the lifetime of B is
exponentially distributed with a mean of 400 hours, what is the PDF of X, the time until the
system fails?

Solution: The configuration of the system is shown below.

λ µ

Let the random variable U denote the lifetime of A and let the random variable V denote the
lifetime of B. Then the PDFs and CDFs of U and V are given by

1 1
f U (u ) = λe −λu , u ≥ 0, FU (u ) = 1 − e −λu , E[U ] = = 200 ⇒ λ =
λ 200
1 1
f V (v) = µe − µv , v ≥ 0, FV (v) = 1 − e − µv , E[V ] = = 400 ⇒ µ =
µ 400

The time, X, until the system fails is given by X = min (U , V ). If we assume that A and B fail
independently, then U and V are independent random variables and the PDF of X can be obtained
as follows:

15
 FX ( x) = P[ X ≤ x] = P[min (U , V ) ≤ x] = P[(U ≤ x ) ∪ (V ≤ x )]
= P[U ≤ x ] + P[V ≤ x ] − P[U ≤ x, V ≤ x ] = FU ( x) + FV ( x) − FUV ( x, x)
= FU ( x) + FV ( x) − FU ( x) FV ( x)
d
f X ( x) = FX ( x) = f U ( x) + f V ( x) − f U ( x) FV ( x) − FU ( x) f V ( x)
dx
= f U ( x){1 − FV ( x)} + f V ( x){1 − FU ( x)}
= λe −λx e − µx + µe − µx e −λx = λe −(λ + µ )x + µe −(λ + µ )x = (λ + µ )e −(λ + µ ) x
 1 1 
 exp{− ( 200 + )x} = 3 −( 3 x / 400)
= + 1 1
400 e x≥0
 200 400  400

6.18 A system consists of two components A and B that are connected in parallel. If the lifetime
of A is exponentially distributed with a mean 200 hours and the lifetime of B is
exponentially distributed with a mean of 400 hours, what is the PDF of Y, the time until the
system fails?

Solution: The system configuration is shown below.

λ

Let the random variable U denote the lifetime of A and let the random variable V denote the
lifetime of B. Then the PDFs and CDFs of U and V are given by
1 1
f U (u ) = λe −λu , u ≥ 0, FU (u ) = 1 − e −λu , E[U ] = = 200 ⇒ λ =
λ 200
1 1
f V (v) = µe − µv , v ≥ 0, FV (v) = 1 − e − µv , E[V ] = = 400 ⇒ µ =
µ 400
The time, Y, until the system fails is given by Y = max (U , V ). If we assume that A and B fail
independently, then U and V are independent random variables and the PDF of Y can be obtained
as follows:

FY ( y ) = P[Y ≤ y ] = P[max (U , V ) ≤ y ] = P[(U ≤ y ) ∩ (V ≤ y )] = FUV ( y, y ) = FU ( y ) FV ( y )

FY ( y ) = f U ( y ) FV ( y ) + FU ( y ) f V ( y ) = λe −λy {1 − e − µy }+ µe − µy {1 − e −λy }
d
f Y ( y) =
dy
= λe −λy + µe − µy − (λ + µ )e −(λ + µ ) y
1 − y / 200 1 − y / 400 3 −( 3 y / 400)
= e + e − e y≥0
200 400 400
16
6.19 The random variables X 1 , X 2 , X 3 , X 4 , X 5 are independent and identically distributed
exponential random variables with parameter λ. Find the following probability:
P[max( X 1 , X 2 , X 3 , X 4 , X 5 ) ≤ a].

Solution: The PDF of X k is given by

f X k ( x) = λe − λx , x ≥ 0; k = 1, 2, Κ , 5
Since random variables X 1 , X 2 , X 3 , X 4 , X 5 are independent, we have that

P[max( X 1 , X 2 , X 3 , X 4 , X 5 ) ≤ a ] = P[ X 1 ≤ a, X 2 ≤ a, X 3 ≤ a, X 4 ≤ a, X 5 ≤ a]
= FX1 X 2 X 3 X 4 X 5 (a, a, a, a, a ) = FX1 (a ) FX 2 (a ) FX 3 (a ) FX 4 (a) FX 5 (a)
[
= 1 − e − λa ]
5

6.20 A system consists of three independent components X, Y, and Z whose lifetimes are
exponentially distributed with means 1 / λ X , 1 / λY and 1 / λ Z , respectively. Determine the
PDF and expected value of W, the time until the system fails (or the lifetime of the system),
under the following system configurations:

(a) The components are connected in series

(b) The components are connected in parallel
(c) The components are connected in a backup mode with X used first, then Y, and then Z.
Solution: The PDFs of the lifetimes are given by
f X ( x) = λ X e − λX x x≥0
− λY y
f Y ( y ) = λY e y≥0
− λZ z
f Z ( z ) = λZ e z≥0
(a) When the components are connected in series, the time W until the system fails is given
by W = min ( X , Y , Z ), and the arrangement is as shown in the following figure.
λX λY λZ

The PDF of W can be obtained as follows:

17
 FW ( w) = P[W ≤ w] = P[min ( X , Y , Z ) ≤ w] = P[( X ≤ w) ∪ (Y ≤ w) ∪ (Z ≤ w)]
= P[ X ≤ w] + P[Y ≤ w] + P[Z ≤ w] − P[ X ≤ w, Y ≤ w] − P[X ≤ w, Z ≤ w] − P[Y ≤ w, Z ≤ w]
+ P[X ≤ w, Y ≤ w, Z ≤ w]
= FX ( w) + FY ( w) + FZ ( w) − FX ( w) FY ( w) − FX ( w) FZ ( w) − FY ( w) FZ ( w)
+ FX ( w) FY ( w) FZ ( w)

FW ( w) = f X ( w){1 − FY ( w) − FZ ( w) + FY ( w) FZ ( w)}
d
f W ( w) =
dw
+ f Y ( w){1 − FX ( w) − FZ ( w) + FX ( w) FZ ( w)}
+ f Z ( w){1 − FX ( w) − FY ( w) + FX ( w) FY ( w)}
= (λ X + λY + λ Z )e −(λX +λY +λZ )w w≥0

(b) When the components are connected in parallel, the time W until the system fails is given
by W = max( X , Y , Z ), and the arrangement is as shown in the following figure
λX

λY
Y

λZ
Z

The PDF of W can be obtained as follows:

FW ( w) = P[W ≤ w] = P[max ( X , Y , Z ) ≤ w] = P[( X ≤ w) ∩ (Y ≤ w) ∩ (Z ≤ w)] = FX ( w) FY ( w) FZ ( w)

d
f W ( w) = FW ( w) = f X ( w) FY ( w) FZ ( w) + f Y ( w) FX ( w) FZ ( w) + f Z ( w) FX ( w) FY ( w)
dw
{ } { } {
= λ X e −λX w 1 − e −λY w − e −λZ w + λY e −λY w 1 − e −λX w − e −λZ w + λZ e −λZ w 1 − e −λ X w − e −λY w }
+ (λ X + λY + λZ )e −(λX +λY +λZ )w w≥0

(c) When the components are connected in a backup mode with X used first, then Y, and then
Z, the time until the system fails is given by W = X + Y + Z , and the PDF of W is given
by f W ( w) = f X ( w) ∗ f Y ( w) ∗ f Z ( w). Let S = X + Y . Then f S ( s ) = f X ( s ) ∗ f Y ( s ),
W = S + Z , and f W ( w) = f S ( w) ∗ f Z ( w). Now,
18
 s s
f S (s) = f X ( s) ∗ fY ( s) = ∫ f X ( s − y ) f Y ( y )dy = ∫ λ X e −λ X ( s− y)
λY e −λ y dy
Y

y =0 y =0

λ X λY
=
λY − λ X
{e −λ s − e −λ s }
X Y
s≥0

Thus, the PDF of W becomes

λ X λY λ Z
∫ {e }
w w
λ
f W ( w) = f S ( w) ∗ f Z ( s ) = ∫ f S (w − z ) f Z ( z )dz = − X ( w− z )
− e −λY ( w− z ) e −λZ z dz
z =0 λY − λ X z =0

λ X λY λZ
=
(λ X − λY )(λY − λZ )(λZ − λ X )
{λ X (e −λ w − e −λ Y Zw
) + λ (e
Y
− λZ w
) ( )}
− e −λ X w + λ Z e −λX w − e −λY w , w ≥ 0

Section 6.8: Two Functions of Two Random Variables

6.21 Two independent random variables X and Y have variances σ X2 = 9 and σ Y2 = 25,
respectively. If we define two new random variables U and V as follows:
U = 2 X + 3Y
V = 4 X − 2Y

(a) Find the variances of U and V

(b) Find the correlation coefficient of U and V
(c) Find the joint PDF of U and V in terms of f XY ( x, y ).

Solution: The mean of U and the mean V are given by

E[U ] = µU = 2µ X + 3µ Y
E[V ] = µV = 4 µ X − 2 µY
Consider a random variable K that is the sum of two random variables L and M with
known means and variances. Specifically, let K = aL + bM , where a and b are constants.
The mean of K is E[ K ] = µ K = aE[ L] + bE[ M ] = aµ L + bµ M . Thus, the variance of K is
given by
σ K2 = E[( K − µ K ) 2 ] = E[(aL + bM − aµ L − bµ M ) 2 ] = E[{a ( L − µ L ) + b( M − µ M )}2 ]
= E[a 2 ( L − µ L ) 2 + 2ab( L − µ L )( M − µ M ) + b 2 ( M − µ M ) 2 ]
= a 2 E[( L − µ L ) 2 ] + 2abE[( L − µ L )( M − µ M )} + b 2 E[( M − µ M ) 2 ]
= a 2σ L2 + 2ab σ LM + b 2σ M2
where σ LM is the covariance of L and M. If L and M are independent, then the covariance
σ LM = 0 and σ K2 = a 2σ L2 + b 2σ M2 .
(a) Since X and Y are independent, the variances of U and V are given by

19
 σ U2 = 4σ X2 + 9σ Y2 = 4(9) + 9(25) = 261
σ V2 = 16σ X2 + 4σ Y2 = 16(9) + 4(25) = 244

(b) The correlation coefficient of U and V can be obtained as follows:

σ UV = E[UV ] − µU µV = E[UV ] − {2µ X + 3µ Y }{4 µ X − 2 µ Y } = E[UV ] − 8µ X2 − 8µ X µ Y + 6µ Y2

But
E[UV ] = E[(2 X + 3Y )(4 X − 2Y )] = E[8 X 2 + 8 XY − 6Y 2 ] = 8 E[ X 2 ] + 8E[ X ]E[Y ] − 6 E[Y 2 ]
{ } {
= 8 σ X2 + µ X2 + 8µ X µY − 6 σ Y2 + µY2 }
Thus,
σ UV = E[UV ] − 8µ X2 − 8µ X µY + 6 µY2
{ } { }
= 8 σ X2 + µ X2 + 8µ X µY − 6 σ Y2 + µY2 − 8µ X2 − 8µ X µ Y + 6 µY2 = 8σ X2 − 6σ Y2
σ UV 8σ − 6σ 2
8(9) − 6(25)
2
− 78
ρUV = = X
= Y
= = −0.3091
σUσV (261)(244) (261)(244) (261)(244)

(c) The joint PDF of U and V can be obtained in terms of f XY ( x, y ) as follows. First, the
solution to the equations U = 2 X + 3Y and V = 4 X − 2Y is
2U + 3V 2U − V
X= , Y=
16 8
The Jacobian of the transformation is given by
∂u ∂u
∂x ∂y 2 3
J ( x, y ) = ∂v ∂v
= = −16
∂x ∂y 4 −2
Thus, we obtain

f XY ( 2u16+3v , 2u8−v ) 1  2u + 3v 2u − v  1  2u + 3v   2u − v 
f UV (u , v) = = f XY  , = fX   fY  
| −16 | 16  16 8  16  16   8 
where the last equality follows from the fact that X and Y are independent.

6.22 Two random variables X and Y have variances σ X2 = 16 and σ Y2 = 36. If their correlation
coefficient is 0.5, determine the following:
(a) The variance of the sum of X and Y
(b) The variance of the difference of X and Y.
Solution:
(a) Let U = X + Y ⇒ µU = 0. Thus, the variance of U is σ U2 = E[U 2 ]. Now, the second
moment of U is given by

20
 E[U 2 ] = E[( X + Y ) 2 ] = E[ X 2 + 2 XY + Y 2 ] = E[ X 2 ] + 2 E[ XY ] + E[Y 2 ] = σ X2 + 2 E[ XY ] + σ Y2

Since E[ XY ] = σ XY = ρ XY σ X σ Y , the variance of U becomes

σ U2 = E[U 2 ] = σ X2 + 2 ρ XY σ X σ Y + σ Y2 = 16 + 2(0.5)(4)(6) + 36 = 76

(b) V = X − Y ⇒ µV = 0. Thus, the variance of V is σ V2 = E[V 2 ]. Now, the second moment

of V is given by

E[V 2 ] = E[( X − Y ) 2 ] = E[ X 2 − 2 XY + Y 2 ] = E[ X 2 ] − 2 E[ XY ] + E[Y 2 ] = σ X2 − 2 E[ XY ] + σ Y2

Thus,
σ V2 = E[V 2 ] = σ X2 − 2 ρ XY σ X σ Y + σ Y2 = 16 − 2(0.5)(4)(6) + 36 = 28

6.23 The joint PDF of two continuous random variables X and Y is given by

e − ( x + y ) x ≥ 0, y ≥ 0
f XY ( x, y ) = 
0 otherwise

If we define the random variable W = X / Y , find the PDF of W.

Solution: To find the PDF of W, we first find the PDFs of X and Y. We note that the joint PDF is
separable into an x-factor and a y-factor, and the region over which the PDF is defined is
rectangular. Therefore, X and Y are independent and the marginal PDFs are given by

f X ( x) = Ae − x x≥0
−y
f Y ( y ) = Be y≥0
1 = AB

Now,

∫
∞

−∞
∞
[
f X ( x)dx = 1 = ∫ Ae − x dx = A − e − x
0
]
∞
0 =A

Thus, A = B = 1. The CDF of W is given by

X   X  X 
FW ( w) = P[W ≤ w] = P  ≤ w = P  ≤ w, Y > 0  ∪  ≤ w, Y < 0 
Y   Y  Y 
X  X 
= P  ≤ w, Y > 0 + P  ≤ w, Y < 0
Y  Y 
= P[ X ≤ wY , Y > 0] + P[ X ≥ wY , Y < 0]

21
Since f XY ( x, y ) = 0 when y < 0, we obtain the following region of integration:
Y

X = wY

X ≤ wY

Thus,
∞ wy ∞ wy
FW ( w) = P[ X ≤ wY , Y > 0] = ∫ ∫ f XY ( x, y )dxdy = ∫ e−y ∫ e − x dxdy
y =0 x =0 y =0 x =0
∞

[− e ] [ ]  e − y ( w+1)  1
∞ wy ∞
−y −x −y − wy w
=∫ e dy = ∫ e 1− e dy = − e − y +  = 1− =
y =0 0 y =0
 w +1 0 w +1 w +1
d 1
f W ( w) = FW ( w) = 0<w<∞
dw (w + 1)2

6.24 Let X and Y be two independent random variables that are uniformly distributed between 0
and 1. If we define Z = XY , find the PDF of Z.

Solution: The PDF of X and Y is given by

1 0 ≤ x ≤1
f X ( x) = f Y ( x) = 
0 otherwise

Since X and Y are independent, their joint PDF is f XY ( x, y ) = f X ( x) f Y ( y ). Given that Z = XY ,

we define an auxiliary variable W = X . Thus, the solution to the equations

Z = XY
W=X

is X = W , Y = Z / W . The Jacobian of the transformation is

∂z ∂z
∂x ∂y y x
J ( x, y ) = ∂w ∂w
= = − x = −w
∂x ∂y 1 0
Thus, the joint PDF of Z and W is given by

22
 1
f XY ( x, y ) 1 1  0 < z ≤ w ≤1
f ZW ( z , w) = = f XY ( w, z / w) = f X ( w) f Y ( z / w) =  w
J ( x, y ) | w| | w| 0 otherwise
Finally, the PDF of Z is given by
1 1 1 − ln( z ) 0 < z <1
f Z ( z ) = ∫ f ZW ( z , w)dw = ∫ dw = 
w= z w= z w
0 otherwise

6.25 Suppose X and Y are independent and identically distributed geometric random variables
with success parameter p. Find the PMF of S = X + Y .

Solution: The PMFs of X and Y are given by

p X ( x) = p (1 − p ) x −1 x ≥1
y −1
pY ( y ) = p (1 − p) y ≥1
Since X and Y are independent, the PMF of S is the convolution of the PMFs of X and Y. That is,

∞
p S ( s ) = p X ( s ) ∗ pY ( s ) = ∑p
y = −∞
X ( s − y ) pY ( y )

To find the limits of the summation, we note that x ≥ 1 ⇒ s − y ≥ 1 ⇒ y ≤ s − 1. And since we are
given that y ≥ 1, the PMF of S becomes
s −1 s −1 s −1
p S ( s ) = ∑ p X ( s − y ) pY ( y) = ∑ p 2 (1 − p ) s − y −1 (1 − p ) y −1 = p 2 ∑ (1 − p) s −2
y =1 y =1 y =1
s −2
= ( s − 1) p (1 − p )
2
s≥2
Note the S is a second-order Pascal random variable.

6.26 Three independent continuous random variables X, Y, and Z are uniformly distributed
between 0 and 1. If the random variable S = X + Y + Z , determine the PDF of S.

Solution: Let the random variable W be the sum of X and Y; that is, W = X + Y , and thus
S = W + Z . The PDFs of X, Y, and Z are given by
1 0 ≤ x ≤1
f X ( x ) = f Y ( x ) = f Z ( x) = 
0 otherwise
The PDF of W is given by
∞
f W ( w) = f X ( w) ∗ f Y ( w) = ∫ f X ( w − y ) f Y ( y )dy
y = −∞

The limits of integration can be computed with the aid of the figures below.

23
 fY ( y ) f X ( w − y) fY ( y ) f X (w − y)

1 1

0 w 1 0 w −1 1 w

From Figure (a) we see that when 0 ≤ w ≤ 1, we have that

∞ w
f W ( w) = ∫ f X ( w − y ) f Y ( y )dy = ∫ dy = w
y = −∞ y =0

Similarly, when 1 ≤ w ≤ 2, as in Figure (b), we have that

∞ 1
f W ( w) = ∫ f X ( w − y ) f Y ( y )dy = ∫ dy = 2 − w
y = −∞ y = w−1

Thus,
w 0 ≤ w ≤1

f W ( w) = 2 − w 1≤ w ≤ 2
0 otherwise

From this we have that the PDF of S is given by
∞ ∞
f S ( s) = ∫ f W ( s − z ) f Z ( z )dz = ∫ fW ( w) f Z ( s − w)dw
z = −∞ w= −∞

The figure below shows f W (w) and f Z ( s − w) when s = 0. Note that when s < 0 there is no
overlap between the two PDFs.

fZ (0−w)

f W (w)

When 0 ≤ s ≤ 1, the area overlapped by the two PDFs is shown below

24
 f Z ( s − w)

fW (w)

Thus,
s 1 2
f S ( s) = ∫ fW ( w) f Z ( s − w)dw =
s 0 ≤ s ≤1
w=0 2
Similarly, when 1 ≤ s ≤ 2, we have the situation illustrated in the following figure.

f Z ( s − w)

s −1
2−s f W (w)

s −1

Since the area of interest is the sum of the areas of 2 trapezoids labeled A and B, and the area of
a trapezoid is given by 12 (sum of parallel sides) × (height), we have that
1
{1 + (s − 1)}{1 − ( s − 1)} + 1 {1 + (2 − s)}{s − 1}
s
f S (s) = ∫ f W ( w) f Z ( s − w)dw =
w= s −1 2 2
1
=
2
{6 s − 2 s 2 − 3} 1≤ s ≤ 2
Finally, when 2 ≤ s ≤ 3, we have the following situation:

f Z ( s − w)
1
f W (w)
2 − (s − 1)

w
-1 0 1 s −1 2 s

25
For this case, we have that
1
{2 − (s − 1)}{2 − (s − 1)} = 1 (3 − s) 2
2
f S ( s) = ∫ f W ( w) f Z ( s − w)dw = 2≤s≤3
w = s −1 2 2
Thus, we obtain the PDF of S as
 12 s 2 0 ≤ s ≤1
1
 2 (6 s − 2 s − 3) 1≤ s ≤ 2
2
f S (s) = 
 12 (3 − s ) 2≤s≤3
2

0 othewise


6.27 Suppose X and Y are two continuous random variables with the joint PDF f XY ( x, y ). Let
the functions U and W be defined as follows: U = 2 X + 3Y , and W = X + 2Y . Find the
joint PDF f UW (u , w).

Solution: The solution to the two equations is X = 2U − 3W , Y = 2W − U . The Jacobian of the

transformation is given by
∂u ∂u
∂x ∂y 2 3
J ( x, y ) = ∂w ∂w = =1
∂x ∂y 1 2
Thus, the joint PDF f UW (u , w) is given by
f (2u − 3w, 2w − u )
f UW (u, w) = XY = f XY (2u − 3w, 2 w − u )
1

6.28 Find f UW (u, w) in terms of f XY ( x, y ) if U = X 2 + Y 2 and W = X 2 − Y 2 .

Solution: The solutions to the equations are

U +W
X =±
2
U −W
Y =±
2

The Jacobian of the transformation is given by

∂u ∂u
∂x ∂y 2x 2y
J ( x, y ) = ∂w ∂w
= = −8 xy
∂x ∂y 2x − 2 y
Thus, the joint PDF f UW (u, w) is given by

f UW (u , w) =
f XY ( u+w
2 , u−w
2 )+ f (XY
u+w
2 ,− u−w
2 ) + f (−
XY
u +w
2 , u −w
2 ) + f (−
XY
u+w
2 ,− u−w
2 )
4 u 2 − w2 4 u 2 − w2 4 u 2 − w2 4 u 2 − w2
26
6.29 X and Y are independent normal random variables, where X ~ N ( µ X , σ X2 ) and
Y ~ N ( µ Y , σ Y2 ). If we define U = X + Y and W = X − Y , find the joint PDF of U and W.
(Note: Give the explicit expression for f UW (u, w).)

Solution: The solution to the equations is

U +W
X=
2
U −W
Y=
2
The Jacobian of the transformation is given by
∂u ∂u
∂x ∂y 1 1
J ( x, y ) = ∂w ∂w = = −2
∂x ∂y 1 −1
Thus, the joint PDF f UW (u, w) is given by
f XY ( u +2w , u −2w ) 1
f UW (u, w) = = f XY ( u +2w , u −2w )
| −2 | 2
Since X and Y are independent, their marginal and joint PDFs are given by
1
f X (x ) = e −( x− µ X ) / 2σ X
2 2

σ X 2π
1
fY ( y ) = e − ( y − µY ) / 2σ Y2
2

σ Y 2π

e −{( x− µ X ) }
1
f XY ( x, y ) = f X ( x ) f Y ( y ) = / 2σ X2 + ( y − µY ) 2 / 2σ Y2
2

2πσ X σ Y
Thus, f UW (u, w) is given by
  (u + w − 2 µ X ) 2 (u − w − 2 µ Y ) 2 
f XY ( u +2w , )=
1 1
f UW (u , w) = u−w
exp−  + 
2 2
4πσ X σ Y   8σ 2
X 8σ Y2 

Section 6.10: The Central Limit Theorem

6.30 If 30 fair dice are rolled, what is the probability that the sum obtained is between 95 and
125?

Solution: Let X 1 , X 2 , Κ , X 30 be random variables that denote the outcomes of the rolls of the
dice. Then the mean and variance of X k , k = 1, 2, Κ , 30, are given by
1
E[ X k ] = {1 + 2 + 3 + 4 + 5 + 6} = 21 = 3.5
6 6
1
{
σ X2 k = (1 − 3.5)2 + (2 − 3.5)2 + (3 − 3.5)2 + (4 − 3.5)2 + (5 − 3.5)2 + (6 − 3.5)2 =
6
35
12
}
Let the random variable S denote the sum of these outcomes; that is,

27
 S = X 1 + X 2 + Λ + X 30
Since these random variables are independent and identically distributed, we have that
(30)(21)
E[ S ] = E[ X 1 + X 2 + Λ + X 30 ] = 30 E[ X 1 ] = = 105
6
(30)(35)
σ S2 = 30σ X2 1 = = 87.5
12
σ S = 87.5 = 9.354

Finally, because the number of observations is “large,” we can apply the central limit theorem as
follows:

 125 − 105   95 − 105 

P[95 < S < 125] = FS (125) − FS (95) = Φ  − Φ  = Φ(2.14 ) − Φ(− 1.07 )
 9.354   9.354 
= Φ(2.14 ) − {1 − Φ(1.07 )} = Φ(2.14 ) + Φ (1.07 ) − 1 = 0.9838 + 0.8577 − 1
= 0.8415

6.31 X 1 , X 2 , Κ , X 35 are independent random variables each of which is uniformly distributed

between 0 and 1. Let S = X 1 + X 2 + Λ + X 35 . What is the probability that S > 22 ?

Solution: The mean and variance of X k , k = 1, 2, Κ , 30, are given by

1+ 0
E[ X k ] = = 0.5
2

σ Xk =
2 (1 − 0)
2
=
1
12 12
Because the random variables are independent, the mean, variance and standard deviation of S
are given by
E[ S ] = E[ X 1 + X 2 + Λ + X 35 ] = 35E[ X 1 ] = (35)(0.5) = 17.5
(35)(1) 35
σ S2 = 35σ X2 = =
1
12 12
35
σS = = 1.708
12
Since the number of observations is 35, we can apply the central limit theorem as follows:

 22 − 17.5 
P[ S > 22] = 1 − P[ S ≤ 22] = 1 − FS (22) = 1 − Φ  = 1 − Φ(2.63) = 1 − 0.9957 = 0.0043
 1.708 

6.32 The random variable X is uniformly distributed between 1 and 2. If S is the sum of 40
independent experimental values of X, evaluate P[55 < S ≤ 65].

28
Solution: Let X 1 , X 2 , Κ , X 40 be random variables that denote the experimental values of X.
Thus, the mean and variance of X k , k = 1, 2, Κ , 40, are given by
1+ 2
E[ X k ] = = 1. 5
2

σ X2 k =
(2 − 1)2 = 1
12 12
Given that S = X 1 + X 2 + Λ + X 40 , the mean, variance and standard deviation of S are given by
E[ S ] = E[ X 1 + X 2 + Λ + X 40 ] = 40 E[ X 1 ] = (40)(1.5) = 60
(40)(1) 40 10
σ S2 = 40σ X2 = = =
1
12 12 3
10
σS = = 1.826
3
From the central limit theorem, we have that
 65 − 60   55 − 60 
P[55 < S < 65] = FS (65) − FS (55) = Φ  − Φ  = Φ (2.74) − Φ (− 2.74)
 1.826   1.826 
= Φ(2.74 ) − {1 − Φ(2.74)} = 2Φ(2.74) − 1 = 2(0.9969) − 1
= 0.9938

6.33 Consider the number of times K that the number 4 appears in 600 tosses of a fair die.
Determine the probability that the number appears 100 times using the following methods:

(a) Stirling’s formula, which states that n!≈ 2πn ( ne ) = n n e − n 2πn

n

(b) The Poisson approximation to the binomial distribution

(c) The central limit theorem by replacing K = 100 with 99.5 < K < 100.5

Solution: The probability p that the number 4 appears in any toss is p = 16 . Thus, the number of
times K that the number 4 appears in 600 tosses of the die is a binomially distributed random
variable whose PMF is given by
k 600− k
 600  1   5 
p K (k ) =      k = 0, 1, 2, Κ , 600
 k  6   6 
The probability that the number appears 100 times is given by
100 500 100 500
 600  1   5  600!  1   5 
P[ K = 100] = p K (100) =      =    
100  6   6  100! 500!  6   6 
(a) Using the Stirling’s formula, n!≈ 2πn ( ne ) = 2πn n n e − n , we have that
n

29
 p K (100) =
{ 1200π (600 )(e )(5 )}
600 −600 500
=
1200π
(6 ){ 200π (100 )(e )}{ 1000π (500 )(e )} (
600 100 −100 500 −500
200π )( 1000π )
3
= = 0.0437
500π
(b) Using the Poisson approximation to the binomial distribution, we have that

600
λ = np = = 100
6
λ100 e −λ 100100 e −100 100100 e −100 1
p K (100) ≈ = = = = 0.0399
100! 100! { }{ }
200π 100100 e −100 200π

(c) Using the central limit theorem, we have that

E[ K ] = np = 100
500
σ K2 = np (1 − p ) =
6
500
σK = = 9.129
6
Thus,

P[ K = 100] = p K (100) ≈ P[99.5 < K < 100.5] = FK (100.5) − FK (99.5)

 100.5 − 100   99.5 − 100 
= Φ  − Φ  = Φ (0.05) − Φ(− 0.05) = Φ (0.05) − {1 − Φ(− 0.05)}
 9.129   9.129 
= 2Φ(0.05) − 1 = 2(0.5199) − 1 = 0.0398

Section 6.11: Order Statistics

6.34 A machine has 7 identical components that operate independently with respective lifetimes
X 1 , X 2 , Κ , X 7 hours. Their common PDF and CDF are f X (x) and FX (x), respectively.
Find the probability that the machine lasts at most 5 hours if
(a) It keeps going until all its components fail.
(b) It fails as soon as one of its components fails.
(c) It fails when it has only one component that has not failed.
Solution:
(a) Let Y be a random variable that denotes the time until all components have failed. Then
Y = max {X 1 , X 2 , Κ , X 7 }
Since the X k are independent, then if the machine keeps going until all its components
fail, the probability that it lasts at most 5 hours is given by

30
 P[Y ≤ 5] = P[max {X 1 , X 2 , Κ , X 7 } ≤ 5] = P[ X 1 ≤ 5, X 2 ≤ 5, Κ , X 7 ≤ 5]
= P[ X 1 ≤ 5, ]P[ X 2 ≤ 5] Κ P[ X 7 ≤ 5] = FX1 (5) FX 2 (5)Λ FX 7 (5)
= [FX (5)]
7

(b) Let U be a random variable that denotes the time the first component fails. Now, if the
machine fails as soon as one of its components fails and it lasts at most 5 hours, then the
first component to fail lasted no more than 5 hours. Since the components behave
independently, the time to machine failure can be represented by a binomial distribution
with a “success” probability p = FX (5). Thus,

7 7
P[U ≤ 5] =   p 1 (1 − p ) 7 −1 =   FX (5 ) [1 − FX (5)] = 7 FX (5) [1 − FX (5)]
6 6

1  1 

(c) Let V dente the time until the 6th component failure occurs. Since the machine “fails”
when the 6th component fails, the probability that it lasts at most 5 hours is the probability
that 6 of the 7 components lasted at most 5 hours and is given by
7 7
P[V ≤ 5] =   p 6 (1 − p ) 7 −6 =   [FX (5 )] [1 − FX (5)] = 7 [FX (5 )] [1 − FX (5)]
6 6

6 6

6.35 A machine needs 4 out of its 6 identical independent components to operate. Let
X 1 , X 2 , Κ , X 6 denote the respective lifetimes of the components, and assume that each
component’s lifetime is exponentially distributed with a mean of 1 / λ hours. Find
(a) The CDF of the machine’s lifetime.
(b) The PDF of the machine’s lifetime.
Solution: The PDF and CDF of the life of each component are given by
f X ( x ) = λe − λx x≥0
− λx
FX ( x) = 1 − e
(a) Let Y denote the lifetime of the machine. Since the machine needs at least 4 of the 6
components to operate, Y is the time until the 3rd failure occurs, and its CDF is given by
 6
FY ( y ) = P[Y ≤ y ] =   [FX ( y )] [1 − FX ( y )] = 20[FX ( y )] [1 − FX ( y )]
3 3 3 3

3
 
[
= 20 1 − e −λy ] [e ]
3 − λy 3
[
= 20 e −λy − e −2 λy ]
3

(b) The PDF of Y is given by

fY ( y) =
d
dy
[
FY ( y ) = 60 e −λy − e −2λy ] {− λe
2 − λy
+ 2λe − 2 λ y }
{ }[
= 60 2λe −2 λy − λe −λy e −λy − e −2 λy ]
2
y≥0

31
6.36 Assume that the random variables X 1 , X 2 , Κ , X 6 are independent and identically
distributed with the common PDF f X (x) and common CDF FX (x). Find the PDF and
CDF of the following:
(a) The 2nd largest random variable.
(b) The maximum random variable.
(c) The minimum random variable.

Solution: Let Yk denote the kth largest of the random variables X 1 , X 2 , Κ , X 6 . From the
textbook we know that for a set of n random variables the CDF and PDF of Yk are given by

k −1
 n 
FYk ( y ) = ∑   [FX ( y )] [1 − FX ( y )]
n−m m

m =0  n − m 

f X ( y )[1 − FX ( y )] [FX ( y )]
n! k −1 n−k
f Yk ( y ) = y≥0
(k − 1)!(n − k )!

(a) The CDF and PDF of the 2nd largest random variable are obtained by substituting n = 6
and k = 2 :
1
 6 
FY2 ( y ) = ∑   [FX ( y )]6−m [1 − FX ( y )]m = [FX ( y )]6 + 6[FX ( y )]5 [1 − FX ( y )]
m =0  6 − m 

f Y2 ( y ) = 30 f X ( y )[1 − FX ( y )][FX ( y )] y≥0

4

(b) The CDF and PDF of the maximum random variable are obtained by substituting n = 6
and k = 1 :
0
 6 
FY1 ( y ) = ∑   [FX ( y )]6−m [1 − FX ( y)]m = [FX ( y )]6
m =0  6 − m 

f Y1 ( y ) = 6 f X ( y )[FX ( y )] y≥0
5

(c) The CDF and PDF of the minimum random variable are obtained by substituting n = 6
and k = 6 :
5
 6 
FY6 ( y ) = ∑   [FX ( y )] [1 − FX ( y )]
6−m m

m =0  6 − m 

= [FX ( y )] + 6[FX ( y )] [1 − FX ( y )] + 15[FX ( y )] [1 − FX ( y)]

6 5 4 2

+ 20[FX ( y )] [1 − FX ( y )] + 15[FX ( y )] [1 − FX ( y)] + 6 FX ( y )[1 − FX ( y )]

3 3 2 4 5

f Y6 ( y ) = 6 f X ( y )[1 − FX ( y )] y≥0
5

32
Chapter 7: Transform Methods

Chapter Summary: This chapter discusses transform methods that are useful in computing
moments of random variables. In particular, it discusses the characteristic function, the z-
transform of the probability mass functions of discrete random variables and the s-transform of
the probability distribution functions of continuous random variables.

Section 7.2: Characteristic Functions

7.1 Find the characteristic function of the random variable X with the following PDF:
1
 6 ≤ x ≤ 10
f X ( x) =  4
0 otherwise

Solution: The characteristic function is given by

10
∞ e jwx 10  e jwx  e10 jw − e 6 jw
Φ X ( w) = ∫ e jwx
f X ( x)dx = ∫ dx =   =
x = −∞ x =6 4
 4 jw  6 4 jw

7.2 Find the characteristic function of the random variable Y with the following PDF:
3e −3 y y≥0
f Y ( y) = 
0 otherwise

Solution: The characteristic function is given by

∞
∞ ∞
−3 y
∞
− ( 3− jw ) y  e −( 3− jw) y  3
Φ Y ( w) = ∫ e jwy
f Y ( y )dy = ∫ e jwy
3e dy = 3∫ e dy = 3−  =
y = −∞ y =0 y =0
 3 − jw  0 3 − jw

7.3 Find the characteristic function of the random variable X with the following PDF:
0 x < −3
x +3
 −3≤ x < 0
 9
f X ( x) = 
3 − x 0≤ x<3
 9
0 x≥3

Solution: The characteristic function is given by

1
 ∞ ( x + 3)e jwx 0 3 (3 − x )e
jwx
Φ X ( w) = ∫ e jwx f X ( x)dx = ∫ dx + ∫ dx
x = −∞ x = −3 9 x =0 9

{ 1  0 jwx
}  e jwx  
3
1 0 jwx 3 3 3
= ∫ xe dx − ∫ xe dx + 3∫ e dx = ∫ xe dx − ∫ xe dx + 3
jwx jwx jwx
 
9 x = −3 x =0 x = −3 9  x =−3 x =0
 jw  −3 

1  0 jwx 3
= ∫ xe dx − ∫ xe dx +
{
3 e j 3 w − e − j 3w  1  0 jwx } 3 6 e j 3w − e − j 3w  { }
 = ∫ x =xe dx − ∫ xe dx +
jwx jwx

9  x = −3 x =0 jw  9  −3 x =0 2 jw 
1 0 3 6 
= ∫ xe jwx dx − ∫ xe jwx dx +   sin(3w)
= − =
9 x 3 x 0
 w 

Let u = x ⇒ du = dx, and let dv = e jwx dx ⇒ v = e jwx / jw. Thus,

0 3
0 3  xe jwx  1 0 jwx  xe jwx  1 0 jwx
∫ dx − ∫ xe dx =   − ∫ − + ∫ e dx
jwx jwx
xe e dx  
x = −3 x =0
 jw  −3 jw x =−3  jw  0 jw x =0
3e −3 jw
=
jw
1
{
+ 2 1 − e −3 jw − }
3e 3 jw
jw
1
− 2 e 3 jw − 1 { }
w w

= −
{
2 3 e 3 jw − e −3 jw } − {e 3 jw
+ e −3 jw }
w2 jw w2
{ }
2  6  e 3 jw − e −3 jw  2  e 3 jw + e −3 jw
= 2 −  − 2 
{ }
w  w 2j w  2
2 6  2 
= 2 −   sin(3w) −  2  cos(3w)
w  w w 

This implies that

1  0 jwx 3 6 
Φ X ( w) = ∫ x =xe dx − ∫ xe jwx dx +   sin(3w)
9  −3 x=0
 w 
1 2 6  2  6 
=  2 −   sin(3w) −  2  cos(3w) +   sin(3w)
9 w  w w   w 
=
2
{1 − cos(3w)}
9w 2

7.4 The characteristic function of the random variable X is given by Φ X (w). If we define the
random variable Y = aX + b, what is the characteristic function of Y?

Solution: By definition, Φ X ( w) = E[e jwX ], which means that

Φ Y ( w) = E[e jwY ] = E[e jw( aX +b ) ] = E[e jwaX e jwb ] = e jwb E[e j ( wa ) X ] = e jwb Φ X (aw)
2
Section 7.3: s-Transforms
7.5 Explain why each of the following functions is or is not a valid s-transform of a PDF:
1 − e −5 s
(a) A( s ) =
s
7
(b) B ( s ) =
4 + 3s
5
(c) C ( s ) =
5 + 3s
Solution: The condition under which a function H (s ) can be the s-transform of a PDF is that
H (0) = 1.
1− e −5 s
(a) Given that A( s ) = s , we have that A(0) = 00 . Therefore, using L’Hopital’s rule we
obtain

d
(
1 − e −5 s ) 5e − 5 s
A(0) = ds = = 5 ≠1
d
(s ) 1 s =0
ds s =0

Thus, A(s ) is not a valid s-transform of a PDF.

(b) Given that B ( s ) = 4 +73 s , we have that B (0) = 74 ≠ 1, which means that B(s ) cannot be
the s-transform of a PDF.
(c) Given that C ( s ) = 5+53 s , we have that C (0) = 55 = 1, which means that C (s ) can be the
s-transform of a PDF.

7.6 Assume that the s-transform of the PDF of the random variable Y is given by
K
M Y (s) =
s+2

Determine the following:

(a) the value of K that makes the function a valid s-transform of a PDF
2
(b) E[Y ]

Solution:
(a) The value of K that makes the function a valid s-transform of a PDF can be obtained as
follows:
3
 K
M Y ( 0) = =1⇒ K = 2
2

2
(b) To obtain E[Y ] we proceed as follows:
 d 2 M Y (s)   2K  1
E[Y ] = ( −1) 
2 2
 =  3
=
 s = 0  ( s + 2)  s = 0 2
2
 ds
7.7 X and Y are independent random variables with the PDFs
λe − λx x≥0
f X ( x) = 
0 otherwise
µe − µy y≥0
f Y ( y) = 
0 otherwise

If the random variable R is defined by R = X + Y , determine the following:

(a) M R (s )
(b) E[R ]
(c) σ R2

Solution: First, we determine the s-transforms of X and Y as follows:

λ
M X (s) =
s+λ
µ
M Y ( s) =
s+µ
 λ  µ 
(a) Because X and Y are independent, M R ( s ) = M X ( s ) M Y ( s ) =   
 s + λ  s + µ 
1 1
(b) E[ R] = E[ X ] + E[Y ] = +
λ µ
1 1
(c) σ R2 = σ X2 + σ Y2 = +
λ 2
µ2
7.8 The random variable X has the following PDF:
2 x 0 ≤ x ≤1
f X ( x) = 
0 otherwise

Determine the numerical values of

4
 d 3
(a)  [M X ( s )] 
 ds  s =0
 d3 
(b)  3 M X ( s )
 ds  s =0

Solution: The moments of X that we will need are given by

1
∞ 1  2x3  2
E[ X ] = ∫ xf X ( x)dx = ∫ 2 x dx =  2
 =
−∞ 0
 3 0 3
1
∞  2x5 
1 2
E[ X ] = ∫ x f X ( x)dx = ∫ 2 x dx = 
3 3
 =
4
−∞ 0
 5 0 5
Since we are required to determine the numerical values of the derivatives of an s-transform, we
do not have to explicitly find M X (s ). Instead we proceed as follows:
d 3
(a) To obtain  [M X ( s )]  , we note that this is the negative of the expected value of the
 ds  s =0
sum of three independent and identically distributed random variables X 1 , X 2 , X 3 that
have the same distribution as X; that is,

d 3
 ds [M X ( s )]  = (−1){E[ X 1 ] + E[ X 2 ] + E[ X 3 ]} = −3E[ X ] = −2
  s =0

 d3 
(b) To obtain  3 M X ( s ) , we note that it is related to the third moment of X as follows:
 ds  s =0

 d3  2
 3 M X ( s ) = (−1) E[ X ] = −
3 3

 ds  s =0 5

7.9 The s-transform of the PDF of the random variable X is given by

λ6
M X (s) =
(s + λ ) 6
Determine the following:
(a) E[ X ]
(b) σ X2

Solution: Note that the s-transform can be expressed as follows:

λ6  λ 
6

M X (s) = = 
(s + λ ) s+λ
6

5
Let Y be the random variable whose PDF f Y ( y ) has the s-transform
λ
M Y ( s) =
s+λ
Then E[Y ] = λ , σ =
1 2
Y
1
λ2
and X is the sum of 6 independent and identically distributed random
variables Y1 , Y2 , Κ , Y6 whose common PDF is f Y ( y ); that is, X = Y1 + Y2 + Κ + Y6 . Thus
6
(a) E[ X ] = 6 E[Y ] =
λ
6
(b) σ X2 = 6σ Y2 =
λ2

7.10 The s-transform of the PDF of the random variable X is given as M X (s). If we define the
random variable Y = aX + b, what is the s-transform of the PDF of Y?

Solution: By definition, M X ( s ) = E[e − sX ], which means that

M Y ( s ) = E[e − sY ] = E[e − s ( aX +b ) ] = E[e − saX e − sb ] = e − sb E[e − saX ] = e − sb M X (as)

7.11 The continuous random variables X and Y have the following PDFs:
1 0 < x ≤1
f X ( x) = 
0 otherwise
0.5 2< y≤4
f Y ( y) = 
0 otherwise
Assume that the function L(s ) is defined as follows:
L( s ) = [M X ( s )] [M Y ( s )]
3 2

Determine the value of the following quantity:

2
 d2   d  
 2 L( s )  −  L( s )  
 ds  s =0  ds  s =0 

Solution: The mean and variance of X and Y are given by

1+ 0
E[ X ] = = 0.5
2
(1 − 0) 2 1
σ X2 = =
12 12
4+2
E[Y ] = =3
2
( 4 − 2) 2 1
σ X2 = =
12 3
6
Now, L is the sum of 5 independent random variables, 3 of which are identically distributed as X
and 2 are identically distributed as Y. That is,
L = X 1 + X 2 + X 3 + Y1 + Y2
Since the quantity
2
 d2   d  
 2 L( s ) −  L( s )  = E[ L ] − (E[ L]) = σ L
2 2 2

 ds  s =0  ds  s =0 

we obtain
2
 d2   d  
 2 L( s ) −  L( s )  = E[ L ] − (E[ L]) = σ L = 3σ X + 2σ Y =
3 2 11
+ =
2 2 2 2 2

 ds  s =0  ds  s =0  12 3 12

Section 7.4: z-Transforms

7.12 The z-transform of the PMF of the random variable X is given by
1+ z2 + z4
G X ( z) =
3

Determine
(a) E[ X ]
(b) p X (E[ X ]); that is, P[ X = E[ X ]]

Solution:
dG X ( z )  2z + 4z 3  6
(a) E[ X ] = =  = =2
dz z =1  3  z =1 3
(b) G X ( z ) = ∑ x z x p X ( x) = 13 + 13 z 2 + 13 z 4 . Thus
 13 x=0
1
 x=2
p X ( x) =  13
3 x=4
0
 otherwise
p X ( E[ X ]) = p X (2) = 13

7.13 If the z-transform of the PMF of the random variable X is given by

G X ( z ) = A(1 + 3 z ) 3

determine the numerical values of the following:

7
 (a) E[ X 3 ]
(b) p X (2)

Solution: We first determine the value of A as follows:

1
G X (1) = 1 = A(1 + 3) 3 = 64 A ⇒ A =
64
Thus,
dG X ( z )
= 3 A(3)(1 + 3 z ) 2 = 9 A(1 + 3z ) 2
dz
d 2G X ( z)
= 9 A(2)(3)(1 + 3 z ) = 54 A(1 + 3z )
dz 2
d 3G X ( z )
= 54 A(3) = 162 A
dz 3

Now,

d 3G X ( z)  d3 ∞
 ∞ 
dz 3
= 3 ∑ z x
p X ( x )  = ∑ x( x − 1)( x − 2) z p X ( x)
x −3

z =1  dz x =0  z =1  x =0  z =1

{ }
∞ ∞
= ∑ x( x − 1)( x − 2) p X ( x) = ∑ x 3 − 3 x 2 + 2 x p X ( x)
x =0 x =0

= E[ X 3 ] − 3E[ X 2 ] + 2 E[ X ]
d 3G X ( z)
E[ X 3 ] = + 3E[ X 2 ] − 2 E[ X ]
dz 3 z =1
But
 d2  d 
E[ X 2 ] =  2 G X ( z ) +  G X ( z )
 dz  z =1  dz  z =1
d 
E[ X ] =  G X ( z )
 dz  z =1
(a) Thus, E[ X 3 ] is given by

 d3   d 2  d    d 
E[ X 3 ] =  3 G X ( z ) + 3 2 G X ( z ) +  G X ( z )  − 2  G X ( z )
 dz  z =1  dz  z =1  dz  z =1   dz  z =1
 d3   d2  d 
=  3 G X ( z )  +  2 G X ( z ) +  G X ( z ) 
 dz  z =1  dz  z =1  dz  z =1

{
= 162 A + 3{54 A + 162 Az}z =1 + 9 A(1 + 3 z ) 2
954
64
}
= 14.91
z =1 = 954 A =

(b) To obtain p X (2) we observe that it is the coefficient of z 2 in G X (z ), which is given by

8
 G X ( z ) = A(1 + 3z ) 3 = A{1 + 9 z + 27 z 2 + 27 z 3 } ⇒ p X (2) = 27 A =
27
64

7.14 If the z-transform of the PMF of the random variable K is given by

A(14 + 5 z − 3 z 2 )
GK ( z) =
2−z

determine the values of

(a) A
(b) p K (1)

Solution:
A(14 + 5 − 3) 1
(a) G K (1) = 1 = = 16 A ⇒ A =
2 −1 16
(b) We can express the z-transform in the form

A(14 + 5 z − 3 z 2 ) A(14 + 5 z − 3z 2 ) 1 ∞
GK ( z ) = = = (14 + 5 z − 3 z 2
) ∑ ( 2z )k
2−z 2(1 − 2 )
z
32 k =0

=
1
32
{ 2 3
}
(14 + 5 z − 3 z 2 ) 1 + 2z + ( 2z ) + ( 2z ) + Λ =
1
{
32
14 + 12 z + 3 z 2 + 1.5 z 3 + Λ }
Since p K (1) is the coefficient of z in the above polynomial, we have that

12 3
p K (1) = = = 0.375
32 8

7.15 Explain why the function C ( z ) = z 2 + 2 z − 2 is or is not a valid z-transform of the PMF of
a random variable.

Solution: We apply the first of two tests: C (1) = 1 + 2 − 2 = 1. Next, we look at the coefficients of
the powers of z to ensure that they are all nonnegative and less than or equal to 1. We see two
violations of this rule: the coefficient of z is 2 > 1, and the constant term, which is the coefficient
of z 0 , is − 2 < 0. Thus, we conclude that the function cannot be the z-transform of a PMF.

7.16 Consider the function:

1
D( z ) =
2−z

(a) Is it a valid z-transform of the PMF of a random variable?

(b) If it is, what PMF has the z-transform?

9
Solution:
(a) First, we obtain D (1) = 1. Next, we observe that

D( z ) =
1
=
1
=
1 ∞ z k 1
∑
2 − z 2(1 − 2 ) 2 k =0
z
2
{ 2
} {
( 2 ) = 1 + 2z + ( 2z )2 + ( 2z )3 + Λ = 1 1 + 2z + z2
4 + 8z + Λ
3
}
Since D(1) = 1 and the coefficients of the powers of z are nonnegative and no greater than
1, we conclude that D ( z ) is a valid z-transform of the PMF of a random variable.
(b) From the above polynomial expression for D(z ) we observe that the coefficient of z k is
given by ()
1 1 k
2 2
, k = 0, 1, Κ . Thus, the PMF that has the z-transform is
k k +1
11 1
p K (k ) =   =   k = 0, 1, 2, Κ
22 2

7.17 If the z-transform of the PMF of the random variable N is given by

G N ( z ) = 0.5 z 5 + 0.3 z 7 + 0.2 z 10

determine

(c) the PMF of N

(d) E[N ]
(e) σ N2
Solution:
(a) From the coefficients of the powers of z in G N (z ) we conclude that the PMF of N is
0.5 n=5
0.3 n=7

p N ( n) = 
0.2 n = 10
0 otherwise

(b) E[N] = [dzd GN (z)]z=1 = [2.5z4 + 2.1z6 + 2z9 ]z=1 = 2.5 + 2.1+ 2 = 6.6. This can also be
obtained as follows: E[ N ] = 5(0.5) + 7(0.3) + 10(0.2) = 2.5 + 2.1 + 2.0 = 6.6.
(c) To find σ N2 we first find the 2nd moment of N as follows:

 d2  d 
[
E[ N 2 ] =  2 G N ( z ) +  G N ( z ) = 10 z 3 + 12.6 z 5 + 18 z 8 ]
z =1 + 6.6 = 47.2
 dz  z =1  dz  z =1
σ N2 = E[ N 2 ] − (E[ N ])2 = 47.2 − (6.6) 2 = 47.2 − 43.56 = 3.64
Note that the second moment can also be obtained directly as follows:
E[ N 2 ] = 5 2 (0.5) + 7 2 (0.3) + 10 2 (0.2) = 12.5 + 14.7 + 20.0 = 47.2.
10
7.18 The PMF of the random variable X has the z-transform
6
 zp 
G X ( z) =  
1 − z (1 − p ) 
Determine the following:
(a) E[ X ]
(b) σ X2

Solution: Let Y be a random variable whose PMF has the z-transform

zp
GY ( z ) =
1 − z (1 − p)
Then X is the sum of 6 independent and identically distributed random variables whose PMF is
the same as Y; that is, X = Y1 + Y2 + Λ + Y6 . Now, Y is a geometrically distributed random
variable with PMF, mean and variance as follows:
pY ( y ) = p(1 − p ) y −1 y ≥1
1
E[Y ] =
p
1− p
σ Y2 = 2
p
Thus,
6
(a) E[ X ] = 6 E[Y ] =
p
6(1 − p)
(b) σ X2 = 6σ Y2 =
p2

7.19 The z-transform of the PMF of the random variable X is given as G X (z ). If we define the
random variable Y = aX + b, what is the z-transform of the PMF of Y?

Solution: GY ( z ) = E[ z Y ] = E[ z aX +b ] = E[ z aX z b ] = z b E[ z aX ] = z b E[( z a ) X ] = z b G X ( z a )

Section 7.5: Random Sum of Random Variables

7.20 People arrive at a restaurant by families. The number of families X that arrive over the
period of 1 hour is found to be a Poisson random variable with rate λ . If the number of
people in each arriving family is a random variable N whose PMF has the z-transform that
is given by
1 1 1
GN ( z) = z + z2 + z3
2 3 6

11
 determine the following:

(a) G M ( z ), the z-transform of the PMF of M, which is the total number of people arriving
at the restaurant in an arbitrary hour
(b) E[Y ], where Y is the total number of people that arrive at the restaurant over a three-
hour period
Solution:
(a) Let N k denote the number of people in the kth family to arrive at the restaurant. If we
define M x as the number of people when X = x families have arrived, then we have that
M x = N1 + N 2 + Λ + N x
Since the N k are independent and identically distributed, the z-transform of the PMF of
M x is
G M X ( z ) = [G N ( z )]
x

Thus, the z-transform of the PMF of M, the total number of people arriving at the
restaurant in an arbitrary hour, is given by

[ ]
∞ ∞
G M ( z ) = ∑ G M x ( z ) p X ( x ) = ∑ [G N ( z )] p X ( x ) = E [G N ( z )] X = G X (G N ( z ))
x

x =0 x =0

{[
= exp λ 2z + z2
3 + z3
6
]}
−1
(b) Let M i denote the number of people that arrive in the ith hour, i = 1, 2, 3. Since the M i
are identically distributed, we have that the expected value, E[Y ], of the total number of
people that arrive at the restaurant over a 3-hour period is given by
E[Y ] = E[ M 1 ] + E[ M 2 ] + E[ M 3 ] = 3E[ M ] = 3E[ N ]E[ X ]
E[ X ] = λ
 1 2 z 3z 2 
E[ N ] = [ dzd G N ( z )]z =1 =  +
5
+  =
2 3 6  z =1 3
Thus, E[Y ] = 3( 53 )λ = 5λ .

7.21 The number of customers, K, that shop at the neighborhood store in a day has the PMF
λk e − λ
p K (k ) = k = 0, 1, 2, Κ
k!

Independently of K, the number of items N that each customer purchases has the PMF

12
  14 n=0
1
4 n =1

p N (n) =  13 n=2
1 n=3
6
0 otherwise

What is the z-transform of the PMF of Y, the total number of items that the store sells on an
arbitrary day?

Solution: The z-transforms of the PMFs of K and N are given by

G K ( z ) = e λ ( z −1)
1 1 1 1
GN ( z) = + z + z 2 + z3
4 4 3 6

Let N i denote the number of items bought by the ith customer, i = 1, 2, Κ , and Yk the total
number of items given that k customers arrived at the store that day. Then,

Yk = N 1 + N 2 + Λ + N k

Since the N i are independent and identically distributed, the z-transform of the PMF of Yk is
given by
GYk ( z ) = [G N ( z )]
k

Thus, the z-transform of the PMF of Y is given by

[ ]
∞ ∞
GY ( z ) = ∑ GYk ( z ) p K (k ) = ∑ [G N ( z )] p K (k ) = E [G N ( z )] K = G K (G N ( z ))
k

k =0 k =0

{[
= exp λ 14 + 4z + z2
3 + z3
6
]}
− 1 = exp λ 4z +{[ z2
3 + z3
6 − 34 ]}
7.22 Books are packed into cartons. The weight W of a book in pounds is a continuous random
variable with PDF

1 1≤ w ≤ 5
fW ( w) =  4
0 otherwise

The number K of books in any carton is a random variable with the PMF

13
  14 k =8
1
4 k =9

p K (k ) =  13 k = 10
1 k = 12
6
0 otherwise

If we randomly select a carton and its weight is X, determine

(a) the s-transform of the PDF of X.
(b) E[X].
(c) the variance of X.

Solution: The s-transform of the PDF of W and the z-transform of the PMF of K are given by

e − s − e −5 s
M X (s) =
4s
1 8 1 9 1 10 1 12
GK ( z ) = z + z + z + z
4 4 3 6

(a) Let X k denote the weight of a carton that contains k books, and let Wi denote the weight
off the ith book in the carton. Then

X k = W1 + W2 + Λ + Wk

Since the Wi are independent and identically distributed, the z-transform of the PMF of
X k is given by

M X k ( s ) = [M W ( s )]
k

Thus, the s-transform of the PDF of X is given by

[ ]
∞ ∞
M X ( s ) = ∑ M X k ( s ) p K (k ) = ∑ [M W ( s )] p K (k ) = E [ M W ( s )] K = G K ( M W ( s ))
k

k =0 k =0

1 1 1 1 
=  z 8 + z 9 + z 10 + z 12  − s −5 s
4 4 3 6  z = e −e
4s

(b) The expected value of X is given by E[ X ] = E[ K ]E[W ], where E[K ] and E[W ] are
given by

14
  
E[ K ] = [ dzd G K ( z )]z =1 = 2 z 7 + z 8 + z 9 + 2 z 11  =
9 10 115
 4 3  z =1 12
5 +1
E[W ] = =3
2

 115  115
Thus, E[ X ] = 3 = = 28.75.
 12  4

Note that E[K ] can also be obtained by the more direct approach as follows:

1 1 1  1  8 9 10 12 24 + 27 + 40 + 24 115

E[ K ] = 8  + 9  + 10  + 12  = + + + = =
4 4  3 6 4 4 3 6 12 12

(c) The variance of X is given by

σ X2 = E[ K ]σ W2 + (E[W ])2 σ K2

Now,

(5 − 1) 2 16 4
σ W2 = = =
12 12 3
1 1 1  1  64 81 100 144 1123
E[ K 2 ] = 8 2   + 9 2   + 10 2   + 12 2   = + + + =
4 4  3 6 4 4 3 6 12
2
1123  115 
σ = E[ K ] − (E[ K ])
251
= −  = = 1.743
2 2 2
K
12  12  144

Thus,

 115  4   251  115 251

σ X2 =    + 9 = + = 28.4653
 12  3   144  9 16

15
Chapter 8: Introduction to Descriptive Statistics

Chapter Summary: This chapter presents an introduction to descriptive statistics. It discusses

such topics as measures of central tendency, which include the mean, mode and median;
measures of spread, which include the range, interquartile range, variance and standard
deviation; and graphical displays, which include dot plot, histogram, bar chart, pie chart, and box
and whiskers. Finally, the chapter discusses shapes of frequency distributions, including the
skewness and the peakedness.

Section 8.3: Measures of Central Tendency

8.1 Consider the following set of data: 15, 20, 21, 20, 36, 15, 25, 15.
(a) What is the sample mean?
(b) What is the median?
(c) What is the mode?
Solution:
15 + 20 + 21 + 20 + 36 + 15 + 25 + 15 167
(a) The mean is µ = = = 20.875
8 8
(b) To obtain the median we order the set as follows:
15, 15, 15, 20, 20, 21, 25, 36
Thus, the median is 20.
(c) The mode is 15, which occurs 3 times.

8.2 The following data represents the ages of 21 students in a class: 22, 24, 19, 17, 20, 27, 24,
23, 26, 17, 19, 22, 25, 21, 21, 22, 22, 21, 21, 20, 22.
(a) What is the mean of the data set?
(b) What is the median?
(c) What is the mode?
Solution:
(a) The mean of the set, µ , is obtained as follows. Let M denote the sum of the data. Then
M = 22 + 24 + 19 + 17 + 20 + 27 + 24 + 23 + 26 + 17 + 19 + 22 + 25 + 21 + 21 + 22 + 22
+ 21 + 21 + 20 + 22
M 455
µ= = = 21.667
21 21

(b) To obtain the median we order the set as follows:

17, 17, 19, 19, 20, 20, 21, 21, 21, 21, 22, 22, 22, 22, 22, 23, 24, 24, 25, 26, 27
Thus, the median is 22.

1
 (c) The mode is 22, which occurs 5 times.

8.3 The scores in a class quiz are as follows: 58, 62, 62, 63, 65, 65, 65, 68, 69, 72, 72, 75, 76,
78, 79, 81, 84, 84, 85, 92, 94, 95, 98.
(a) What is the mean score in the quiz?
(b) What is the median score in the quiz?
(c) What is the mode of the quiz?
Solution:
1658
a) The mean is µ = = 72.087
23
b) Since the set is in increasing order, the median is 75.
c) The mode is 65, which occurs 3 times

Section 8.4: Measures of Dispersion

8.4 Consider the following set of data: 15, 20, 21, 20, 36, 15, 25, 15.

(a) What is the variance?

(b) What is the interquartile range?
Solution:
(a) The variance can be obtained as follows:

15 + 20 + 21 + 20 + 36 + 15 + 25 + 15 167
The mean is µ = = = 20.875
8 8
The variance is given by
1 3(15 − 20.875) + 2(20 − 20.875) + (21 − 20.875) 
2 2 2

σ = ∑ (xk − µ ) = 
21 2

8 k 8 + (25 − 20.875)2 + (36 − 20.875)2 

=
1
8
{
3(− 5.875) + 2(− 0.875) + (0.125) + (4.125) + (15.125) =
2 2 2 2 2
}
350.864
8
= 43.858
(b) The interquartile range can be obtained as follows: We re-arrange the set in an ascending
order: 15, 15, 15, 20, 20, 21, 25, 36 . Thus, the median is 20. The lower half of the data
set is 15, 15, 15, 20 whose median is Q1 = 15. The upper half of the data set is
20, 21, 25, 36 whose median is Q3 = (21 + 25) / 2 = 23. Thus, the interquartile range is
IQR = Q3 − Q1 = 8.

8.5 The following data represents the ages of 21 students in a class: 22, 24, 19, 17, 20, 27, 24,
23, 26, 17, 19, 22, 25, 21, 21, 22, 22, 21, 21, 20, 22.
(a) What is the variance?
(b) What is the range?
2
 (c) What is the interquartile range?
Solution:
(a) From Problem 8.2, we have that the mean of the set, µ , is obtained as follows:
M = 22 + 24 + 19 + 17 + 20 + 27 + 24 + 23 + 26 + 17 + 19 + 22 + 25 + 21 + 21 + 22 + 22
+ 21 + 21 + 20 + 22
M 455
µ= = = 21.667
21 21
Thus the variance is
2(17 − 21.667 )2 + 2(19 − 21.667 )2 + 2(20 − 21.667 )2 
 
1 + 4(21 − 21.667 ) + 5(22 − 21.667 ) + (23 − 21.667 ) 
2 2 2

σ2 =
1
∑ ( x k − µ )2 =  
21 k 21 + 2(24 − 21.667 )2 + (25 − 21.667 )2 + (26 − 21.667 )2 
 
+ (27 − 21.667 )
2

1 2(− 4.667 ) + 2(− 2.667 ) + 2(− 1.667 ) + 4(− 0.667 ) + 5(0.333) + (1.333) 
2 2 2 2 2 2

=  
21 + 2(2.333)2 + (3.333)2 + (4.333)2 + +(5.333)2 
136.667
= = 6.5079
21
(b) We first rearrange the data set in an ascending order:
17, 17, 19, 19, 20, 20, 21, 21, 21, 21, 22, 22, 22, 22, 22, 23, 24, 24, 25, 26, 27
Thus, the range is 27 − 17 = 10.
(c) The median is 22, and the lower half of the data set is 17, 17, 19, 19, 20, 20, 21, 21, 21,
21 whose median is Q1 = 20. The upper half of the data set is 22, 22, 22, 22, 23, 24, 24,
25, 26, 27 whose median is Q3 = 23+2 24 = 23.5. Thus, the interquartile rage is
IQR = Q3 − Q1 = 23.5 − 20 = 3.5

8.6 The scores in a class quiz are as follows: 58, 62, 62, 63, 65, 65, 65, 68, 69, 72, 72, 75, 76,
78, 79, 81, 84, 84, 85, 92, 94, 95, 98.
(a) What is the variance?
(b) What is the range?
(c) What is the interquartile range?
Solution:
(a) We first find the mean as follow: Let
M = 58 + 2(62) + 63 + 3(65) + 68 + 69 + 2(72) + 75 + 76 + 78 + 79 + 81 + 2(84) + 85 + 92
+ 94 + 95 + 98 = 1742
1742
µ= = 75.7391
23

3
 Thus, the variance is given by
(58 − 75.7391)2 + 2(62 − 75.7391)2 + (63 − 75.7391)2 
 
+ 3(65 − 75.7391) + (68 − 75.7391) + (69 − 75.7391) 
2 2 2

 2
1 + 2(72 − 75.7391) + (75 − 75.7391) + (76 − 75.7391) 
2 2

σ2 =
1
∑ ( x k − µ )2 =  
23 k 23 + (78 − 75.7391)2 + (79 − 75.7391)2 + (81 − 75.7391)2 
 2
+ 2(84 − 75.7391) + (85 − 75.7391) + (92 − 75.7391) 
2 2

+ (94 − 75.7391)2 + (95 − 75.7391)2 + (98 − 75.7391)2 

 
= 133.236
(b) The range is given by 98 − 58 = 40.
(d) The median is 75. The lower half of the data set is 58, 62, 62, 63, 65, 65, 65, 68, 69, 72,
72 whose median is Q1 = 65. The upper half of the data set is 76, 78, 79, 81, 84, 84, 85,
92, 94, 95, 98 whose median is Q3 = 84. Thus, the interquartile rage is
IQR = Q3 − Q1 = 84 − 65 = 19

Section 8.6: Graphical Displays

8.7 Bob found 15 pens in his new locker: 6 black, 4 blue, 3 red and 2 green.
(a) Draw a dot plot of the data set.
(b) Draw a bar graph of the data set.
Solution:
(a) The dot plot of the data set is as follows:

∗
∗
∗ ∗
∗ ∗ ∗
∗ ∗ ∗ ∗
∗ ∗ ∗ ∗
Black Blue Red Green

(b) The bar graph of the data set is as follows:

4
8.8 The scores in a class quiz are as follows: 58, 62, 62, 63, 65, 65, 65, 68, 69, 72, 72, 75, 76,
78, 79, 81, 84, 84, 85, 92, 94, 95, 98. Draw a box and whiskers diagram of the data set.

Solution: From Problem 8.5 we have that Q1 = 65, M = 75, Q3 = 84. Thus, the interquartile
range is IQR = Q3 − Q1 = 84 − 65 = 19, and the box and whiskers diagram is as follows:
Q1 M Q3

65 75 84

50 58 60 70 80 90 98 100

8.9 Consider a survey concerning the ages of 800 students in a college. The results can be
summarized as follows: 320 students were between 18 and 20 years old, 240 students
were between 20 and 22 years old, 80 were between 22 and 24 years old, and 160
students were between 24 and 26 years old.
(a) Give the frequency distribution of the ages
(b) Plot the histogram of the ages
(c) Plot the frequency polygon of the ages.
Solution:
(a) The frequency distribution of the ages is as follows:

Ages Group Frequency

18 – 20 320
20 – 22 240
22 – 24 80
24 – 26 160

(b) The histogram plot is as follows:

5
 (c) The frequency polygon of the ages is as follows:

8.10 Consider the following set of data: 2, 3, 5, 5, 6, 7, 7, 7, 8, 9, 9, 10, 10, 11, 12, 14, 14, 16,
18, 18, 22, 24, 24, 26, 28, 28, 32, 45, 50, and 55.
(a) Plot the box and whiskers diagram of the set.
(b) Determine if there are any outliers in the data set.
Solution:
(a) The median is 13. The lower data set is 2, 3, 5, 5, 6, 7, 7, 7, 8, 9, 9, 10, 10, 11, 12 whose
median is Q1 = 7. Similarly, the upper half of the data set is 14, 14, 16, 18, 18, 22, 24, 24,
26, 28, 28, 32, 45, 50, 55 whose median is Q3 = 24. Thus, the box and whiskers plot is as
follows:

6
 Q1 M Q3

(b) The interquartile range is IQR = Q3 − Q1 = 17. Now, 1.5 × IRQ = 25.5. Thus,
Q1 − 1.5 × IRQ = −18.5, and Q3 + 1.5 × IRQ = 49.5. Therefore, there are no lower
outliers, but 50 and 55 are outliers at the upper end, as shown below.
Q1 M Q3

• •

8.11 The following data were collected in a study in which 30 students were asked to rate their
professors: very good, good, good, fair, excellent, good, good, very good, fair, good,
good, excellent, very good, good, good, good, fair, very good, good, very good, excellent,
very good, good, fair, fair, very good, very good, good, good, and excellent. Draw a pie
chart to represent the data.
Solution: The tally of the ratings is as follows:

Thus, the proportions of the ratings are as follows:

7
 Rating Proportion
Excellent 13.33%
Very Good 26.67%
Good 43.33%
Fair 16.67%

The pie chart is as shown below:

26.67
25% Very Good 25%
13.33
Excellent
Good

43.33 16.67
25% 25%
Fair

Section 8.7: Shape of Frequency Distribution

8.12 The following data represents the ages of 21 students in a class: 22, 24, 19, 17, 20, 27, 24,
23, 26, 17, 19, 22, 25, 21, 21, 22, 22, 21, 21, 20, 22.
(a) Determine the skewness of the data
(b) Determine the kurtosis of the data
(c) Determine the excess kurtosis.
Solution: From Problem 8.5 we have that µ = 21.667, σ = 6.5079 = 2.5510. Thus, we have
that
(a) The skewness is given by

8
  X − µ 
3
 1 3
 ∑ (xk − µ ) 
1
γ 1 = E  X
 =
σ  (σ X )
3
 X   21 k 
2(17 − 21.667 )3 + 2(19 − 21.667 )3 + 2(20 − 21.667 )3 + 4(21 − 21.667 )3 
 1   1  
3
3
=   + 5(22 − 21.667 ) + (23 − 21.667 ) + 2(24 − 21.667 ) + (25 − 21.667 ) 
3 3 3

 2.551   21  
+ (26 − 21.667 ) + (27 − 21.667 )
3 3


 1   1 2(− 4.667 ) + 2(− 2.667 ) + 2(− 1.667 ) + 4(− 0.667 ) + 5(0.333) 

3 3 3 3 3 3

=    
 2.551   21 + (1.333)3 + 2(2.333)3 + (3.333)3 + (4.333)3 + (5.333)3 
= 0.1328
(b) The kurtosis is given by
 X − µ  4  1 1 4 1 4
 ∑ (xk − µ )  =  ∑ (xk − µ ) 
1
β 2 = E  X
  =
 σ X   (σ X )  21 k
4
 σ X2 ( )
2
 21 k 
2(17 − 21.667 )4 + 2(19 − 21.667 )4 + 2(20 − 21.667 )4 + 4(21 − 21.667 )4 
 1   1  
2
4 
=   + 5(22 − 21.667 ) + (23 − 21.667 ) + 2(24 − 21.667 ) + (25 − 21.667 ) 
4 4 4

 6.5079   21  
+ (26 − 21.667 ) + (27 − 21.667 )
4 4


 1   1 2(− 4.667 ) + 2(− 2.667 ) + 2(− 1.667 ) + 4(− 0.667 ) + 5(0.333) 

2 4 4 4 4 4

=    
 6.5079   21 + (1.333)4 + 2(2.333)4 + (3.333)4 + (4.333)4 + (5.333)4 
= 1.5147

(c) The excess kurtosis is γ 2 = β 2 − 3 = 1.5147 − 3 = −1.4853. Thus, the distribution is

platykurtic.

8.13 The scores in a class quiz are as follows: 58, 62, 62, 63, 65, 65, 65, 68, 69, 72, 72, 75, 76,
78, 79, 81, 84, 84, 85, 92, 94, 95, 98.
(a) What is the skewness of the data?
(b) What is the kurtosis of the data?
(c) What is the excess kurtosis?
Solution: From Problem 8.6 we have that µ = 75.7391, σ = 133.236 = 11.5428. Thus, we have
that
(a) The skewness is given by

9
  X − µ 
3
  1 3  1   1 
3
3
 ∑ (xk − µ )  =     ∑ ( x k − µ ) 
1
γ 1 = E  X
 =
σ  (σ X )
3
 X   23 k   11.5428   23  k 
(58 − 75.7391)3 + 2(62 − 75.7391)3 + (63 − 75.7391)3 
 
+ 3(65 − 75.7391) + (68 − 75.7391) + (69 − 75.7391) 
3 3 3

 3
 1   1 + 2(72 − 75.7391) + (75 − 75.7391) + (76 − 75.7391) 
3 3 3

=    
 11.5428   23 + (78 − 75.7391)3 + (79 − 75.7391)3 + (81 − 75.7391)3 
 3
+ 2(84 − 75.7391) + (85 − 75.7391) + (92 − 75.7391) 
3 3

+ (94 − 75.7391)3 + (95 − 75.7391)3 + (98 − 75.7391)3 

 
(− 17.7391)3 + 2(− 13.7391)3 + (− 12.7391)3 + 3(− 10.7391)3 
 
 1   1 + (− 7.7391) + (− 6.7391) + 2(− 3.7391) + (− 0.7391)
3 3 3 3 3

=    
 11.5428   23 + (0.2609) + (2.2609) + (3.2609) + (5.2609) + 2(8.2609)
3 3 3 3 3

 3
+ (9.2609) + (16.2609) + (18.2609) + (19.2609) + (23.2609) 
3 3 3 3

= 0.4186
(b) The kurtosis is given by
 X − µ  4   1 4
3
 1   1  3
 ∑ (xk − µ )  =     ∑ ( xk − µ ) 
1
β 2 = E  X
  =
 σ X   σ X2 ( ) 2
 23 k   11.5428   23  k 
(58 − 75.7391)4 + 2(62 − 75.7391)4 + (63 − 75.7391)4 
 
+ 3(65 − 75.7391) + (68 − 75.7391) + (69 − 75.7391) 
4 4 4

 4
  1 + 2(72 − 75.7391) + (75 − 75.7391) + (76 − 75.7391) 
2 4 4
 1
=    
 133.236   23 + (78 − 75.7391)4 + (79 − 75.7391)4 + (81 − 75.7391)4 
 4 
+ 2(84 − 75.7391) + (85 − 75.7391) + (92 − 75.7391) 
4 4

+ (94 − 75.7391)4 + (95 − 75.7391)4 + (98 − 75.7391)4 

 
(− 17.7391)4 + 2(− 13.7391)4 + (− 12.7391)4 + 3(− 10.7391)4 
 
  1 + (− 7.7391) + (− 6.7391) + 2(− 3.7391) + (− 0.7391)
2 4 4 4 4
 1 
=    
 133.236   23 + (0.2609) + (2.2609) + (3.2609) + (5.2609) + 2(8.2609)
4 4 4 4 4

 4
+ (9.2609) + (16.2609) + (18.2609) + (19.2609) + (23.2609) 
4 4 4 4

= 2.1348

(c) The excess kurtosis is given by γ 2 = β 2 − 3 = 2.1348 − 3 = −0.8652. Thus, the

distribution is platykurtic. However, because it is very close to zero, we can also say
that it can be mesokurtic.

10
8.14 Consider the following set of data: 15, 20, 21, 20, 36, 15, 25, 15.
(a) What is the skewness of the data?
(b) What is the kurtosis of the data?
(c) What is the excess kurtosis?

Solution: From Problem 8.4 we have that µ = 21.667, σ = 43.858 = 6.6225.

(a) The skewness is given by
 X − µ 
3
 1 1 3
3
 1   1  3
γ 1 = E  X
 =  ∑ ( x − µ )  =     ∑ ( xk − µ ) 
 σ X  (σ X )  8 k
3 k
   6.6225   8  k 

 1   1 3(15 − 20.875) + 2(20 − 20.875) + (21 − 20.875) 

3 3 3 3

=    
 6.6225   8 + (25 − 20.875)3 + (36 − 20.875)3 

{ }
3
 1  1
=    3(− 5.875) + 2(− 0.875) + (0.125) + (4.125) + (15.125)
3 3 3 3 3

 6. 6225   
8
= 1.2570

(b) The kurtosis is given by

 X − µ  4  1 4
2
 1   1  4
 ∑ (xk − µ )  =     ∑ ( xk − µ ) 
1
β 2 = E  X
  =
 σ X   σ X2 ( ) 2
8 k   43.858   8  k 

 1   1 3(15 − 20.875) + 2(20 − 20.875) + (21 − 20.875) 

2 4 4 4

=    
 43.858   8 + (25 − 20.875)4 + (36 − 20.875)4 

{ }
2
 1  1
=    3(− 5.875) + 2(− 0.875) + (0.125) + (4.125) + (15.125)
4 4 4 4 4

 43.858   8 
= 3.6520

(c) The excess kurtosis is given by γ 2 = β 2 − 3 = 3.6520 − 3 = 0.6520. Thus, the

distribution is leptokurtic. As in the previous example, because the excess kurtosis is
close to zero, we can say that it is tending toward being mesokurtic.

11
Chapter 9: Introduction to Inferential Statistics

Chapter Summary: This chapter presents an introduction to inferential statistics. It discusses

such topics as sampling theory; estimation, which includes point estimates and interval
estimates; hypothesis testing; and linear regression analysis.
Section 9.2: Sampling Theory
9.1 A sample size of 5 results in the sample values 9, 7, 1, 4, and 6.
(a) What is the sample mean?
(b) What is the sample variance?
(c) What is the unbiased estimate of the sample variance?
Solution:
9 + 7 + 1 + 4 + 6 27
(a) The sample mean is X = = = 5.4
5 5
(b) The sample variance is given by

( X k − 5.4)2 = (9 − 5.4) + (7 − 5.4) + (1 − 5.4) + (4 − 5.4) + (6 − 5.4)

2 2 2 2 2
1 5
S2 = ∑
5 k =1 5

=
(3.6 ) + (1.6 ) + (− 4.4 ) + (− 1.4 ) + (0.6 )
2 2 2 2 2
=
12.96 + 2.56 + 19.36 + 1.96 + 0.36
5 5
37.20
= = 7.44
5
5
S 2 =  (7.44 ) = 9.3
n
(c) The unbiased estimate of the sample variance is Sˆ 2 =
n −1 4

9.2 The true mean of a quiz conducted in a class of 50 students is 70 points, and the true
standard deviation is 12 points. It is desired to estimate the mean by sampling a subset of
the scores, without replacement.
(a) What is the standard deviation of the sample mean if only 10 scores are used?
(b) How large should the sample size be for the standard deviation of the sample mean to
be 1% of the true mean?
Solution: We are given that the true variance is σ X2 = 144. Since sampling is done without
replacement we proceed as follows:
(a) The standard deviation of the sample mean when only 10 scores are used can be obtained
as follows:
σ X2  50 − 10  144  40 
576 576 24
σ =
2
 =  = ⇒σX = = = 3.428
X
10  50 − 1  10  49  49 49 7

1
 (b) Let n be the sample size required for the standard deviation of the sample mean to be 1%
of the true mean. Then we have that
144  50 − n   1 
σX =   = 70  = 0.7.
n  49   100 
Thus, we have that
144  50 − n 
  = 0.49 ⇒ 144(50 − n) = 0.49(49n) = 24.01n
n  49 
Finally, we have that
144 × 50 7200
n= = = 42.85 ≈ 43
144 + 24.01 168.01

9.3 A random sample of size 81 is taken from a population that has a mean of 24 and
variance 324. Use the central limit theorem to determine the probability that the sample
mean lies between 23.9 and 24.2.
Solution: The population mean and variance are given respectively by µ X = 24, σ X2 = 324.
Thus, for a sample size of 81 the sample standard deviation and sample mean are given by
σ X2 324
σX = = = 4=2
n 81
X = µ X = 24
From the central limit theorem, the probability that the sample mean lies between 23.9 and 24.2
is given by
 24.2 − X   23.9 − X 
P[23.9 < X < 24.2] = FX (24.2) − FX (23.9) = Φ  − Φ
  σ


 σ X   X 
 24.2 − 24   23.9 − 24 
= Φ  − Φ  = Φ(0.1) − Φ (− 0.05)
 2   2 
= Φ(0.1) − {1 − Φ (0.05)} = Φ(0.1) + Φ(0.05) − 1
= 0.5398 + 0.5199 − 1 = 0.0597

9.4 A random number generator produces three-digit random numbers that are uniformly
distributed between 0.000 and 0.999.

(a) If the generator produces the sequence of numbers 0.276, 0.123, 0.072, 0.324, 0.815,
0.312, 0.432, 0.283, 0.717, what is the sample mean?
(b) What is the variance of the sample mean of numbers produced by the random number
generator?
(c) How large should the sample size be in order to obtain a sample mean whose standard
deviation is no greater than 0.01?
Solution: The true mean and true variance are given by

2
 0.000 + 0.999
µX = = 0.4995
2

σ X2 =
(0.999 − 0.000)2 = 0.999 2
12 12
(a) If the generator produces the sequence of numbers 0.276, 0.123, 0.072, 0.324, 0.815,
0.312, 0.432, 0.283, 0.717, the sample mean is
0.276 + 0.123 + 0.072 + 0.324 + 0.815 + 0.312 + 0.432 + 0.283 + 0.717
X = = 0.3727
9
σ X2  1  0.999
 0.998001 2
(b) The variance of the sample mean is σ X2 =  = =  
= 0.00924.
 9108 9  12
(c) Let n be the sample size that is required to obtain a sample mean whose standard
deviation is no greater than 0.01. Then we have that
σ X2 0.999 2 0.999 2 0.999 2
σX = = ≤ 0.01 ⇒ ≤ 0.0001 ⇒ n ≥ = 831.67 = 832
n 12n 12n 12(0.0001)

9.5 Calculate the value of the Student’s t PDF for t = 2 with (a) 6 degrees of freedom and (b)
12 degrees of freedom.
Solution: The PDF of the Student’s t distribution is given by
 v +1
Γ  − ( v +1) / 2
 2   t2 
f T (t ) = 1 + 
 v   v
vπ Γ 
2
where v = n − 1 is the number of degrees of freedom and n is the sample size. When t = 2 we
obtain

 v + 1
Γ  − ( v +1) / 2
 2   4
f T (2) = 1 + 
v v
vπ Γ 
2

(a) At 6 degrees of freedom we obtain

 6 +1
Γ 
Γ(3.5)  5 
− ( 6 +1) / 2 − 3.5 − 3.5
2   4 2.5 × 1.5 × 0.5 π  5 
f T (2) v =6 =   1 +  =   =  
6 6π Γ(3)  3  2! 6 π
6π Γ  
6  3
2
− 3.5
2.5 × 1.5 × 0.5  5 
=   = 0.0640
2 6 3

3
 (b) At 12 degrees of freedom we obtain

 12 + 1 
Γ 
Γ(6.5)  4 
− (12+1) / 2 − 6.5
 2   4
f T (2) v =12 = 1 +  =  
 12   12  12π Γ(6)  3 
12π Γ 
2
− 6.5 − 6.5
5.5 × 4.5 × 3.5 × 2.5 × 1.5 × 0.5 π  4  5.5 × 4.5 × 3.5 × 2.5 × 1.5 × 0.5  4 
=   =  
5! 12 π 3 120 12 3
= 0.0602

Section 9.3: Estimation Theory

9.6 A large number of light bulbs were turned on continuously to determine the average
number of days a bulb can last. The study revealed that the average lifetime of a bulb is
120 days with a standard deviation of 10 days. If the lifetimes are assumed to be
independent normal random variables, find the confidence limits for a confidence level of
90% on the sample mean that is computed from a sample size of (a) 100 and (b) 25.
Solution: We are given that X = 120, σ = 10. When the sample size is n, the confidence limits
for the 90% confidence level are
k 90σ 1.64(10) 16.4
X± = 120 ± = 120 ±
n n n
where k 80 = 1.64 was obtained from Table 9.1 in the textbook.
(a) When n = 100, we obtain the limits as
k 90σ
= 120 ± 1.64 = [118.36, 121.64]
16.4 16.4
X± = 120 ± = 120 ±
n 100 10
(b) When n = 25, we obtain the limits as
k 90σ
= 120 ± 3.28 = [116.72, 123.28]
16.4 16.4
X± = 120 ± = 120 ±
n 25 5

9.7 A random sample of 50 of the 200 electrical engineering students’ grades in applied
probability showed a mean of 75% and a standard deviation of 10%.
(a) What are the 95% confidence limits for the estimate of the mean of the 200 grades?
(b) What are the 99% confidence limits for the estimate of the mean of the 200 grades?
(c) With what confidence can we say that the mean of all the 200 grades is 75 ± 1?
Solution: We are given that the population size is N = 200, and the sample size is n = 50. Since
the population size is not very large compared to the sample size, the confidence limits are given
by

4
 kσ X N −n 10k 200 − 50 10k 150
X ± kσ X = X ± = 75 ± = 75 ±
n N −1 50 200 − 1 50 199
3
= 75 ± 10k = 75 ± 1.23k
199
From Table 9.1 we obtain the following results:
(a) At the 95% confidence level, k = 1.96. Thus, the confidence limits are
75 ± 1.23(1.96) = 75 ± 2.41 = [72.59, 77.41]
(b) At the 99% confidence level, k = 2.58. Thus, the confidence limits are
75 ± 1.23(2.58) = 75 ± 3.17 = [71.83, 78.17]
(c) To obtain the value of k that gives the confidence limit of 75 ± 1, we solve the equation
75 ± 1.23k = 75 ± 1 ⇒ k = 1 / 1.23 = 0.81
The area under the standard normal curve from 0 to 0.81 is 0.7910 − 0.5 = 0.2910 . Thus,
the required degree of confidence is the area 2(0.2910) = 0.5820, which means that the
confidence level is 58%.

9.8 The mean of the grades of 36 freshmen is used to estimate the true average grade for the
freshman class. If µ is the true mean, what is the probability that the estimate differs
from the true mean by 3.6 marks if the standard deviation is known to be 24? (Note: The
problem is asking for the probability that the true mean lies between µ − 3.6 and
µ + 3.6.)
Solution: We are given that n = 36. The probability that the estimate differs from the true mean
by 3.6 marks is given by

[ ]
P X − µ ≤ 3.6 = P[− 3.6 ≤ X − µ ≤ 3.6] = P[µ − 3.6 ≤ X ≤ µ + 3.6]
 µ + 3.6 − µ   
= FX (µ + 3.6) − FX (µ − 3.6 ) = Φ  − Φ µ − 3.6 − µ 
 σ   σ 
 36   36 
 3.6   − 3.6   3.6   − 3.6 
= Φ 24  − Φ 24  = Φ  − Φ  = Φ(0.9 ) − Φ(− 0.9 )
     4   4 
 36   36 
= Φ(0.9 ) − {1 − Φ(0.9 )} = 2Φ(0.9 ) − 1 = 2(0.8159) − 1 = 0.6318

9.9 What is the increase in sample size that is required to increase the confidence level of a
given confidence interval for a normal random variable from 90% to 99.9%?
Solution: The values of k corresponding to the 90% and 99% confidence levels are k = 1.64 and
k = 3.29, respectively. If we denote the samples sizes for the 90% and 99% confidence levels by
m and n, respectively, then

5
  1.64σ X 1.64σ X 
P µ X − ≤ X ≤ µX +  = 0.90
 m m 
 3.29σ X 3.29σ X 
P µ X − ≤ X ≤ µX +  = 0.99
 n n 

If the confidence limits are to be the same for both cases, then we have that
1.64σ X 3.29σ X n 3.29 n
= ⇒ = = 2.0061 ⇒ = 2.00612 = 4.0244
m n m 1.64 m
Thus, we require a four-fold increase in the sample size.

9.10 A box contains red and white balls in an unknown proportion. A random sample of 60
balls selected with replacement from the box showed that 70% were red. Find the 95%
confidence limits for the actual proportion of red balls in the box.
Solution: If we consider selecting a red ball as a success, then the probability of success on each
trial is p = 0.7. Since each trial is a Bernoulli trial, the variance is given by

σ 2 = p(1 − p ) = (0.7)(0.3) = 0.21

If n = 60, the 95% confidence limits for the actual proportion of red balls in the box are given by
k 95σ
= X ± 0.116 = [X − 0.116, X + 0.116]
0.21
X± = X ± 1.96
n 60
9.11 A box contains a mix of red and blue balls whose exact composition of red and blue balls
is not known. If we draw 20 balls from the box with replacement and obtain 12 red balls,
what is the maximum-likelihood estimate of p, the probability of drawing a red ball?
Solution: Let K denote the number of red balls among the 20 balls drawn. If p denotes the
probability of drawing a red ball, then the PMF of K is
 20 
p K (k ) =   p k (1 − p ) 20− k k = 0, 1, Κ , 20
k 
Thus, the likelihood function is given by
 20 
L( p; k ) =   p k (1 − p ) 20− k
k 
The value of p that maximizes the likelihood function can be obtained as follows:
 20 
log L( p; k ) = log  + k log p + (20 − k ) log(1 − p )
k 
∂ k 20 − k
log L( p; k ) = − = 0 ⇒ k (1 − p) = p (20 − k ) ⇒ k = 20 p
∂p p 1− p
k
pˆ =
20
If k = 12, we obtain

6
 k 12
pˆ = = = 0.6
20 20
9.12 A box contains red and green balls whose exact composition is not known. An
experimenter draws balls one by one with replacement until a green ball appears. Let X
denote the number of balls drawn until a green ball appears. This operation is repeated n
times to obtain the sample X 1 , X 2 , Κ , X n . Let p be the fraction of green balls in the
box. What is the maximum-likelihood estimate of p on the basis of this sample?
Solution: The PMF of X is given by
p X ( x) = p(1 − p ) x −1 x = 1, 2, Κ
If the operation is repeated n times to obtain the sample X 1 , X 2 , Κ , X n , then the likelihood
function of the sample is given by

[ ][ ] [
L( p; x1 , x 2 , Λ , x n ) = p (1 − p ) x1 −1 p(1 − p ) x2 −1 Λ p (1 − p) xn −1 ]
x1 + x 2 +Λ + xn − n y−n
= p (1 − p )n
= p (1 − p ) n

where y = x1 + x 2 + Λ + x n . The value of p that maximizes the function can be obtained as

follows:

log L( p; k ) = n log p + ( y − n) log(1 − p )

∂ n y−n
log L( p; k ) = − = 0 ⇒ n(1 − p ) = p ( y − n) ⇒ n = yp
∂p p 1− p
n n
pˆ = =
y x1 + x 2 + Λ + xn

Section 9.4: Hypothesis Testing

9.13 A college provost claimed that 60% of the freshmen at his school receive their degrees
within four years. A curious analyst followed the progress of a particular freshman class
with 36 students and found that only 15 of the students received their degrees at the end
of their fourth year. Determine whether this particular class performed worse than
previous classes at a level of significance of (a) 0.05 and (b) 0.01.

Solution: The population proportion of success (or population mean) is p = 0.6. Since the
experiment is essentially a Bernoulli trial, the population variance is σ 2 = p (1 − p ) = 0.24. The
sample proportion of success (or sample mean) is
15
p= = 0.42
36
Because the sample proportion is less than the population proportion, the null and alternate
hypotheses can be set up as follows:
H 0 : p = 0.60
H 1 : p < 0.60

7
Thus, we have a left-tail test whose z-score is

p− p 0.42 − 0.60 6(0.18)

z= = =− = −2.204
σ/ n = 0.24 / 36 0.24

(a) At the 0.05 level of significance, the critical z-score for the left-tail test is z c = −1.645.
That means that we reject any null hypothesis that lies in the region z ≤ −1.645. Since
the score z = −2.204 lies in this region, we reject H 0 and accept H 1 .
(b) At the 0.01 level of significance, the critical z-score for the left-tail test is z c = −2.33.
Since the score z = −2.204 lies in the acceptance region, which is the region where
z > −2.33, we accept H 0 and reject H 1 .

9.14 An equipment manufacturing company claimed that at least 95% of the equipment it
supplied to a factory conformed to specifications. An examination of a sample of 200
pieces of equipment revealed that 18 of them did not meet the specifications. Determine
whether the company’s claim is legitimate at a level of significance of (a) 0.01, and (b)
0.05.
Solution: The population mean is p = 0.95, and the corresponding variance is
σ 2 = p (1 − p ) = (0.95)(0.05) = 0.0475
The sample mean is
200 − 18
p= = 0.91
200
Because the sample proportion is less than the population proportion, the null and alternate
hypotheses can be set up as follows:
H 0 : p = 0.95
H 1 : p < 0.95
Thus, we have a left-tail test whose z-score is
p− p 0.91 − 0.95 200
z= = = −0.04 = −2.595
σ / n = 0.0475 / 200 0.0475

(a) At the 0.05 level of significance, the critical z-score for the left-tail test is z c = −1.645.
That means that we reject any null hypothesis that lies in the region z ≤ −1.645. Since
the score z = −2.595 lies in this region, we reject H 0 and accept H 1 .
(b) At the 0.01 level of significance, the critical z-score for the left-tail test is z c = −2.33.
Since the score z = −2.595 lies in the rejection region z ≤ −2.33, we also reject H 0 and
accept H 1 .

9.15 A company claims that the boxes of detergent that it sells contain more than the current
500 grams of detergent each. From past experience the company knows that the amount
of detergent in the boxes is normally distributed with a standard deviation of 75 grams. A
8
 worker takes a random sample of 100 boxes and finds that the average amount of
detergent in a box is 510 grams. Test the company’s claim at the 0.05 level of
significance.
Solution: The population mean is µ = 500 with a standard deviation σ = 75. The sample mean
is X = 510 with n = 100 observations. Since the sample mean is greater than the population
mean, we can set up the null and alternate hypotheses as follows:
H 0 : µ = 500
H 1 : µ > 500
Thus, we have a right-tail test whose z-score is
X −µ 510 − 500 100
z= = = = 1.33
σ/ n = 75 100 75
For a right-tail test the critical z-score at the 95% level of confidence is z c = 1.645. Thus, we
reject a null hypothesis that lies in the region z c ≥ 1.645. Since z = 1.33 lies in the acceptance
region, we accept H 0 and reject H 1 . This means that there is no statistical difference between
the sample mean and the population mean at the 95% level of confidence.
9.16 A government agency received many consumers’ complaints that boxes of cereal sold by
a company contain less than the advertised weight of 20 oz of cereal with a standard
deviation of 5 oz. To check the consumers’ complaints, the agency bought 36 boxes of
the cereal and found that the average weight of cereal was 18 oz. If the amount of cereal
in the boxes is normally distributed, test the consumers’ complaint at the 0.05 level of
significance.
Solution: The population mean is µ = 20 with a standard deviation σ = 5. The sample mean is
X = 18 with n = 36 observations. Since the sample mean is less than the population mean, we
can set up the null and alternate hypotheses as follows:
H 0 : µ = 20
H 1 : µ < 20
Thus, we have a left-tail test whose z-score is
X −µ 18 − 20 2(6)
z= = =− = −2.4
σ/ n = 5 / 36 5
For a left-tail test the critical z-score at the 95% level of confidence is z c = −1.645. Thus, we
reject a null hypothesis that lies in the region z c ≤ −1.645. Since z = −2.4 lies in this region, we
reject H 0 and accept H 1 .

Section 9.5: Regression Analysis

9.17 Data were collected for a random variable Y as a function of another random variable X.
The recorded ( x, y ) pairs are as follows:
(3, 2), (5, 3), (6, 4), (8, 6), (9, 5), (11, 8)

9
 (a) Plot the scatter diagram for these data.
(b) Find the linear regression line of y on x that best fits these data.
(c) Estimate the value of y when x = 15.
Solution:
(a) The plot of the scatter diagram is as follows:

∗
∗
∗
∗
∗

(b) To find the regression line y = a + bx of y on x that best fits these data we proceed as
follows:

x y x2 xy
3 2 9 6
5 3 25 15
6 4 36 24
8 6 64 48
9 5 81 45
11 8 121 88
∑ x = 42 ∑ y = 28 ∑ x 2 = 336 ∑ xy = 226
The values of a and b that make the line the best fit for the above data are given by

10
 n∑i =1 xi yi − ∑i =1 xi ∑i =1 y i
n n n
∗ 6(226) − 42(28) 1356 − 1176
b = = = = 0.714
n∑i =1 xi2 −
n
(∑ x ) n
i =1 i
2
6(336) − (42) 2 2016 − 1764

∑ y i − b ∗ ∑i =1 xi
n n
∗ 28 − (0.714)(42) 28 − 29.938
a = i =1
= = = −0.33
n 6 6

Thus, the best line is y = −0.33 + 0.714 x.

(c) When x = 15, we obtain the estimate y = −0.33 + 0.714(15) = 10.38.

9.18 Data were collected for a random variable Y as a function of another random variable X.
The recorded ( x, y ) pairs are as follows:
(1, 11), (3, 12), (4, 14), (6, 15), (8, 17), (9, 18), (11, 19)
(a) Plot the scatter diagram for these data.
(b) Find the linear regression line of y on x that best fits these data.
(c) Estimate the value of y when x = 20.
Solution:
(a) The plot of the scatter diagram is as follows:

∗
∗
∗
∗
∗
∗
∗

11
 (b) To find the regression line y = a + bx of y on x that best fits these data we proceed as
follows:

x y x2 xy
1 11 1 11
3 12 9 36
4 14 16 56
6 15 36 90
8 17 64 136
9 18 81 162
11 19 121 209
∑ x = 42 ∑ y = 106 ∑ x 2 = 328 ∑ xy = 700
The values of a and b that make the line the best fit for the above data are given by

n∑i =1 xi yi − ∑i =1 xi ∑i =1 y i
n n n
∗ 7(700) − 42(106) 4900 − 4452
b = = = = 0.842
n∑i =1 xi2 −
n
(∑ x ) n
i =1 i
2
7(328) − (42) 2 2296 − 1764

∑ y i − b ∗ ∑i =1 xi
n n
∗ 106 − (0.842)(42) 106 − 35.364 70.636
a = i =1
= = = = 10.09
n 7 7 7

Thus, the best line is y = 10.09 + 0.842 x.

(c) When x = 20, we obtain the estimate y = 10.09 + 0.842(20) = 26.93.

9.19 The ages x and systolic blood pressures y of 12 people are shown in the following table:
Age (x) 56 42 72 36 63 47 55 49 38 42 68 60

Blood Pressure (y) 147 125 160 118 149 128 150 145 115 140 152 155

(a) Find the least-squares regression line of y on x.

(b) Estimate the blood pressure of a person whose age is 45 years.
Solution:
(a) To find the regression line y = a + bx of y on x that best fits these data we proceed as
follows:

12
 x y x2 xy
56 147 3136 8323
42 125 1764 5250
72 160 5184 11520
36 118 1296 4248
63 149 3969 9387
47 128 2209 6016
55 150 3025 8250
49 145 2401 7105
38 115 1444 4370
42 140 1764 5880
68 152 4624 10336
60 155 3600 9300
∑ x = 628 ∑ y = 1684 ∑ x = 34416
2
∑ xy = 89985
The values of a and b that make the line the best fit for the above data are given by

n∑i =1 xi yi − ∑i =1 xi ∑i =1 y i
n n n
∗ 12(89985) − 628(1684) 1079820 − 1057552
b = = =
n∑i =1 xi2 −
n
(∑ x ) n
i =1 i
2
12(34416) − (628) 2 412992 − 394384

22268
= = 1.2
18608
∑ y i − b ∗ ∑i =1 xi
n n
∗ 1684 − (1.2)(628) 1684 − 753.6 930.4
a = i =1
= = = = 77.53
n 12 12 12
Thus, the best line is y = 77.53 + 1.2 x.

(b) The estimate of the blood pressure of a person whose age is 45 years is given by
y = 77.53 + 1.2(45) = 131.53.

9.20 The following table shows a random sample of 12 couples who stated the number x of
children they planned to have at the time of their marriage and the number y of actual
children they have.
Couple 1 2 3 4 5 6 7 8 9 10 11 12

Planned Number 3 3 0 2 2 3 0 3 2 1 3 2
of Children (x)

Actual Number 4 3 0 4 4 3 0 4 3 1 3 1
of Children (y)

13
 (a) Find the least-squares regression line of y on x.
(b) Estimate the number of children that a couple who had planned to have 5 children
actually had.
Solution:
(a) To find the regression line y = a + bx of y on x that best fits these data we proceed as
follows:

x y x2 xy
3 4 9 12
3 3 9 9
0 0 0 0
2 4 4 8
2 4 4 8
3 3 9 9
0 0 0 0
3 4 9 12
2 3 4 6
1 1 1 1
3 3 9 9
2 1 4 2
∑ x = 24 ∑ y = 30 ∑ x = 62
2
∑ xy = 76
The values of a and b that make the line the best fit for the above data are given by

n∑i =1 xi yi − ∑i =1 xi ∑i =1 y i
n n n
∗ 12(76) − 24(30) 912 − 720
b = = =
n∑i =1 xi2 −
n
(∑ x ) n
i =1 i
2
12(62) − (24) 2 744 − 576

192
= = 1.143
168
∑ y i − b ∗ ∑i =1 xi
n n
∗ 30 − (1.143)(24) 30 − 27.432 2.568
a = i =1
= = = = 0.214
n 12 12 12
Thus, the best line is y = 0.214 + 1.143x.

(b) The estimate for the number of children that a couple who planned to have 5 children
actually had is given by y = 0.214 + 1.143(5) = 5.929 = 6.

14
Chapter 10: Introduction to Random Processes

Chapter Summary: This chapter presents an introduction to random processes. It discusses

classification of random processes; characterization of random processes including the
autocorrelation function of a random process, autocovariance function, crosscorrelation function
and crosscovariance function; stationary random processes; ergodic random processes; and
power spectral density.
Section 10.3: Mean, Autocorrelation Function and Autocovariance Function
10.1 Calculate the autocorrelation function of the rectangular pulse shown in Figure 10.6; that
is,
X (t ) = A 0≤t ≤T
where A and T are constants.
X (t )

t
0 T

Figure 10.6: Figure for Problem 10.1

Solution: Since the function X (t ) is an aperiodic function, its autocorrelation function is given
by
∞
R XX (t , t + τ ) = ∫ X (t ) X (t + τ )dt
−∞

This is essentially a convolution integral that can be evaluated as follows:

(a) When − T ≤ τ < 0, we obtain the following arrangement:
X(t +τ ) X (t )

τ T +τ
∞
Thus, R XX (t , t + τ ) = ∫ X (t ) X (t + τ )dt = A 2 (T + τ ).
−∞

(b) When 0 ≤ τ < T , we obtain the following arrangement:

X (t )
X(t +τ )

τ T +τ

1
 ∞
Thus, R XX (t , t + τ ) = ∫ X (t ) X (t + τ )dt = A 2 (T − τ ).
−∞

From these two results we have that

 A 2 (T − τ ) τ <T
R XX (t , t + τ ) = 
0 otherwise

10.2 Calculate the autocorrelation function of the periodic function X (t ) = A sin( wt + φ ), where
the period T = 2π / w, and A, φ and w are constants.
Solution: Since the function X (t ) = A sin( wt + φ ) is periodic with period T = 2π / w, its
autocorrelation function is given by
1 T 1 T
R XX (t , t + τ ) = ∫ X (t ) X (t + τ )dt = A sin( wt + φ ) A sin( wt + wτ + φ )dt
2T −T 2T ∫−T
A2 T A2 T
φ τ φ {cos(wτ ) − cos(2wt + wτ + 2φ )}dt
2T ∫−T 4T ∫−T
= sin( wt + ) sin( wt + w + ) dt =

sin (2wt + wτ + 2φ )  w
2π

wA 2 
= t cos( w τ ) −  2π
8π 2w t =− w

wA 2  4π  sin (wτ + 2φ ) sin (wτ + 2φ ) 

=  cos(wτ ) −  − 
8π  w  2w 2w 
2
A
= cos( wτ )
2
10.3 The random process X(t) is given by

X (t ) = Y cos(2πt ) t≥0

where Y is a random variable that is uniformly distributed between 0 and 2. Find the
expected value and autocorrelation function of X(t).
Solution: We are given that
 12 0≤ y≤2
f Y ( y) = 
0 otherwise
2+0
E[Y ] = =1
2

σ Y2 =
(2 − 0)2 = 1
12 3
(a) E[ X (t )] = E[Y cos(2πt )] = E[Y ] cos(2πt ) = cos(2πt )
(b) The autocorrelation function of X (t ) is given by

2
 R XX (t , t + τ ) = E[ X (t ) X (t + τ )] = E[Y cos(2πt )Y cos(2πt + 2πτ )]
{ }
= E[Y 2 ] cos(2πt ) cos(2πt + 2πτ ) = σ Y2 + (E[Y ]) cos(2πt ) cos(2πt + 2πτ )
2

4 2
= cos(2πt ) cos(2πt + 2πτ ) = {cos(2πτ ) + cos(4πt + 2πτ )}
3 3
10.4 The sample function X(t) of a stationary random process Y(t) is given by
X (t ) = Y (t ) sin( wt + Θ)
where w is a constant, Y(t) and Θ are statistically independent, and Θ is uniformly dis-
tributed between 0 and 2π . Find the autocorrelation function of X(t) in terms of RYY (τ ).
Solution: We are given that
1 0 ≤ θ ≤ 2π
f Θ (θ ) =  2π
0 otherwise
2π + 0
E[Θ] = =π
2

σ Θ2 =
(2π − 0)2 = π 2
12 3
The autocorrelation function of X (t ) is given by

R XX (t , t + τ ) = E[ X (t ) X (t + τ )] = E[Y (t ) sin( wt + Θ)Y (t + τ ) sin( wt + wτ + Θ)]

= E[Y (t )Y (t + τ )]E[sin( wt + Θ) sin( wt + wτ + Θ)]
 cos( wτ ) − cos( wt + wτ + 2Θ) 
= RYY (τ ) E[sin( wt + Θ) sin( wt + wτ + Θ)] = RYY (τ ) E  
 2 
1
= RYY (τ ){cos( wτ ) − E[cos( wt + wτ + 2Θ)]}
2
But
∞ 1 2π
E[cos( wt + wτ + 2Θ)] = ∫ cos( wt + wτ + 2θ ) f Θ (θ )dθ = ∫ cos(wt + wτ + 2θ )dθ
−∞ 2π 0

=
1
[sin (wt + wτ + 2θ )]θ2π=0 = 0
4π
Thus,
1
R XX (t , t + τ ) = RYY (τ ) cos( wτ )
2
10.5 The sample function X(t) of a stationary random process Y(t) is given by
X (t ) = Y (t ) sin( wt + Θ)
where w is a constant, Y(t) and Θ are statistically independent, and Θ is uniformly dis-
tributed between 0 and 2π . Find the autocovariance function of X(t).
Solution: We are given that

3
 1 0 ≤ θ ≤ 2π
f Θ (θ ) =  2π
0 otherwise
E[ X (t )] = µ X (t ) = E[Y (t ) sin( wt + Θ)] = E[Y (t )]E[sin( wt + Θ)]
∞ 1 2π
E[sin( wt + Θ)] = ∫ sin( wt + θ ) f Θ (θ )dθ = ∫ sin( wt + θ )dθ
−∞ 2π 0

1
= [− cos wt + θ )]02π = 0
2π
Thus, µ X (t ) = 0 and the autocovariance function of X (t ) is given by

C XX (t , t + τ ) = R XX (t , t + τ ) − µ X (t ) µ X (t + τ ) = R XX (t , t + τ )

1
From Problem 10.4, we know that R XX (t , t + τ ) = RYY (τ ) cos(wτ ). Therefore,
2
1
C XX (t , t + τ ) = R XX (t , t + τ ) = RYY (τ ) cos(wτ )
2
10.6 The random process X(t) is given by
X (t ) = A cos( wt ) + B sin( wt )
where w is a constant, and A and B are independent standard normal random variables (i.e.,
zero mean and variance of 1). Find the autocovariance function of X(t).
Solution: We have that µ A = µ B = 0, and σ A2 = σ B2 = 1 ⇒ E[ A 2 ] = E[ B 2 ] = 1. Thus, we have
that
C XX (t , t + τ ) = R XX (t , t + τ ) − µ X (t ) µ X (t + τ ) = R XX (t , t + τ )
µ X (t ) = E[ X (t )] = E[ A cos(wt ) + B sin( wt )] = E[ A cos( wt )] + E[ B sin( wt )]
= E[ A] cos(wt ) + E[ B] sin( wt ) = 0
R XX (t , t + τ ) = E[ X (t ) X (t + τ )] = E [{A cos( wt ) + B sin( wt )}{A cos(wt + wτ ) + B sin( wt + wτ )}]
= E[ A 2 ] cos( wt ) cos(wt + wτ ) + E[ A]E[ B ] cos( wt ) sin( wt + wτ )
+ E[ A]E[ B ] sin( wt ) cos( wt + wτ ) + E[ B 2 ] sin( wt ) sin( wt + wτ )
= cos( wt ) cos(wt + wτ ) + sin( wt ) sin( wt + wτ )
= cos(− wτ ) = cos( wτ )
Thus, the autocovariance function is
C XX (t , t + τ ) = R XX (t , t + τ ) = R XX (t , t + τ ) = cos(wτ )

10.7 Assume that Y is a random variable that is uniformly distributed between 0 and 2. If we
define the random process X (t ) = Y cos(2πt ), t ≥ 0, find the autocovariance function of
X(t).
Solution: We are given that

4
 1 0≤ y≤2
f Y ( y) =  2
0 otherwise
2+0
E[Y ] = =1
2

σ Y2 =
(2 − 0)2 = 1
12 3
Thus, we have that

C XX (t , t + τ ) = R XX (t , t + τ ) − µ X (t ) µ X (t + τ )
µ X (t ) = E[ X (t )] = E[Y cos(2πt )] = E[Y ] cos(2πt ) = cos(2πt )
R XX (t , t + τ ) = E[ X (t ) X (t + τ )] = E [{Y cos(2πt )}{Y cos(2πt + 2πτ )}]
{ }
= E[Y 2 ] cos(2πt ) cos(2πt + 2πτ ) = σ Y2 + (E[Y ]) cos(2πt ) cos(2πt + 2πτ )
2

4
= cos(2πt ) cos(2πt + 2πτ )
3
Thus, the autocovariance function is
C XX (t , t + τ ) = R XX (t , t + τ ) − µ X (t ) µ X (t + τ )
4
= cos(2πt ) cos(2πt + 2πτ ) − cos(2πt ) cos(2πt + 2πτ )
3
1 1
= cos(2πt ) cos(2πt + 2πτ ) = {cos(2πτ ) + cos(24πt + 2πτ )}
3 6

10.8 A random process X(t) is given by

X (t ) = A cos(t ) + ( B + 1) sin(t ) −∞ <t < ∞
where A and B are independent random variables with E[ A] = E[ B] = 0 and
E[ A 2 ] = E[ B 2 ] = 1. Find the autocovariance function of X(t).

Solution: The autocovariance function can be obtained as follows:

C XX (t , t + τ ) = R XX (t , t + τ ) − µ X (t ) µ X (t + τ )
µ X (t ) = E[ X (t )] = E[ A cos(t ) + ( B + 1) sin(t )]
= E[ A] cos(t ) + E[ B] sin(t ) + sin(t ) = sin(t )
R XX (t , t + τ ) = E[ X (t ) X (t + τ )] = E [{A cos(t ) + ( B + 1) sin(t )}{A cos(t + τ ) + ( B + 1) sin(t + τ )}]
= E[ A 2 ] cos(t ) cos(t + τ ) + E[ A]{E[ B] + 1} cos(t ) sin(t + τ )
{ }
+ E[ B 2 ] + 2 E[ B] + 1 sin(t ) sin(t + τ )
= cos(t ) cos(t + τ ) + 2 sin(t ) sin(t + τ ) = cos(τ ) + sin(t ) sin(t + τ )
Thus, the autocovariance function is

5
 C XX (t , t + τ ) = R XX (t , t + τ ) − µ X (t ) µ X (t + τ )
= cos(τ ) + sin(t ) sin(t + τ ) − sin(t ) sin(t + τ )
= cos(τ )
10.9 Determine the missing elements of the following autocovariance matrix of a zero-mean
wide-sense stationary random process X(t), where the missing elements are denoted by xx.
1.0 xx 0.4 xx 
0.8 xx 0.6 0.4
C XX =  
 xx 0.6 1.0 0.6
 
0.2 xx xx 1.0 
Solution: For any random process X (t ), the autocovariance function is given by
C XX (t , t + τ ) = R XX (t , t + τ ) − µ X (t ) µ X (t + τ )

If X (t ) is a zero-mean wide-sense stationary process, then we have that µ X (t ) = 0 and the

autocovariance function becomes C XX (t , t + τ ) = R XX (τ ). This means that if R XX is the
autocorrelation matrix, then C XX = R XX . Since R XX is a symmetric matrix, we have that

1.0 0.8 0.4 0.2

0.8 1.0 0.6 0.4
C XX =
0.4 0.6 1.0 0.6
 
0.2 0.4 0.6 1.0 

10.10 The random process X(t) is defined as follows:

X (t ) = A + e − B|t|
where A and B are independent random variables. A is uniformly distributed over the range
− 1 ≤ A ≤ 1, and B is uniformly distributed over the range 0 ≤ B ≤ 2. Find the following:
(a) the mean of X(t)
(b) the autocorrelation function of X(t)
Solution: We are given that

6
 1 −1 ≤ a ≤ 1
f A (a) =  2
0 otherwise
1 0≤b≤2
f B (b) =  2
0 otherwise
−1+1
E[ A] = =0
2
2+0
E[ B] = =1
2

σA =
2 (1 − (−1) )
2
1
= = E[ A 2 ]
12 3

σ B2 =
(2 − 0) = 1 ⇒ E[ B 2 ] = 4
2

12 3 3
(a) The mean of X (t ) is given by

[ ] [ ] [
E[ X (t )] = E A + e − B|t| = E[ A] + E e − B|t| = E e − B|t| = ∫ e −b|t| f B (b)db = ] ∞

−∞
1 2 −b|t|
2 ∫0
e db
2
1  e −b|t| 
= −  =
2  t  0 2 t
1
1 − e −2|t| { } t≠0

(b) The autocorrelation of X (t ) is given by

[{ }{ }] [
R XX (t , t + τ ) = E A + e − B|t| A + e − B|t +τ | = E A 2 + Ae − B|t| + Ae − B|t +τ | + e − B|t| e − B|t +τ | ]
= E [A ] + E[ A]E [e ] + E[ A]E [e
2 − B|t | − B|t +τ |
]+ E[e − B|t | − B|t +τ |
e ]
= E [A ] + E [e {
2 }
] − B |t |+|t +τ |

=
1
+
1
3 2{t + t + τ }
1− e {
− 2{ t + t +τ }
}
10.11 The random process X (t ) has the autocorrelation function R XX (τ ) = e −2|τ | . The random
process Y(t) is defined as follows:
t
Y (t ) = ∫ X 2 (u )du
0

Find E[Y (t )].

Solution: The expected value of Y (t ) is given by
E [(Y (t )] = E  ∫ X 2 (u )du 
t

 0 
Interchanging expectation and integration we obtain

0
[ ]
E[Y (t )] = ∫ E X 2 (u ) du = ∫ R XX (0)du = ∫ du = t
t t

0 0
t

7
Section 10.4: Crosscorrelation and Crosscovariance Functions
10.12 Two random processes X(t) and Y(t) are both zero-mean and wide-sense stationary
processes. If we define the random process Z (t ) = X (t ) + Y (t ), determine the
autocorrelation function of Z(t) under the following conditions:
(a) X(t) and Y(t) are jointly wide-sense stationary
(b) X(t) and Y(t) are orthogonal.
Solution: The autocorrelation function of Z (t ) is given by

RZZ (t , t + τ ) = E[ Z (t ) Z (t + τ )] = E [{X (t ) + Y (t )}{X (t + τ ) + Y (t + τ )}]

= E[ X (t ) X (t + τ )] + E[ X (t )Y (t + τ )] + E[Y (t ) X (t + τ )] + E[Y (t )Y (t + τ )]
= R XX (τ ) + R XY (t , t + τ ) + RYX (t , t + τ ) + RYY (τ )
(a) If X (t ) and Y (t ) are jointly wide-sense stationary, then R XY (t , t + τ ) = R XY (τ ) and
RYX (t , t + τ ) = RYX (τ ). Thus, the autocorrelation function of Z (t ) becomes

RZZ (t , t + τ ) = R XX (τ ) + R XY (τ ) + RYX (τ ) + RYY (τ )

(b) If X (t ) and Y (t ) are orthogonal, then R XY (t , t + τ ) = RYX (t , t + τ ) = 0 and we obtain

RZZ (t , t + τ ) = R XX (τ ) + RYY (τ )

10.13 Two random processes X(t) and Y(t) are defined as follows:
X (t ) = A cos(wt ) + B sin( wt )
Y (t ) = B cos(wt ) − A sin( wt )

where w is a constant, and A and B zero-mean and uncorrelated random variables with
variances σ A2 = σ B2 = σ 2 . Find the crosscorrelation function R XY (t , t + τ ).
Solution: The crosscorrelation function R XY (t , t + τ ) is given by
R XY (t , t + τ ) = E[ X (t )Y (t + τ )] = E [{A cos( wt ) + B sin( wt )}{B cos( wt + wτ ) − A sin( wt + wτ )}]
= E[ AB] cos( wt ) cos( wt + wτ ) − E[ A 2 ] cos( wt ) sin( wt + wτ )
+ E[ B 2 ] sin( wt ) cos( wt + wτ ) − E[ AB] sin( wt ) sin( wt + wτ )
Since A and B are uncorrelated, we have that Cov ( A, B ) = E[ AB ] − E[ A]E[ B] = 0. Because
E[ A] = E[ B] = 0, we obtain E[ AB] = 0. Thus, we have that

R XY (t , t + τ ) = E[ B 2 ] sin( wt ) cos( wt + wτ ) − E[ A 2 ] cos(wt ) sin( wt + wτ )

= σ 2 {sin( wt ) cos( wt + wτ ) − cos( wt ) sin( wt + wτ )} = σ 2 sin{wt − ( wt + wτ )}
= σ 2 sin{− wτ } = −σ 2 sin{wτ }
10.14 Two random processes X(t) and Y(t) are defined as follows:
X (t ) = A cos( wt + Θ)
Y (t ) = B sin( wt + Θ)

8
 where w, A and B are constants, and Θ is a random variable that is uniformly distributed
between 0 and 2π .
(a) Find the autocorrelation function R XX (t , t + τ ), and show that X(t) is a wide-sense sta-
tionary process
(b) Find the autocorrelation function RYY (t , t + τ ), and show that Y(t) is a wide-sense sta-
tionary process
(c) Find the crosscorrelation function R XY (t , t + τ ), and show that X(t) and Y(t) are jointly
wide-sense stationary.
Solution: We are given that
1 0 ≤ θ ≤ 2π
f Θ (θ ) =  2π
0 otherwise

(a) The autocorrelation function of X (t ) is given by

R XX (t , t + τ ) = E[ X (t ) X (t + τ )] = E[ A cos( wt + Θ) A cos( wt + wτ + Θ)]

 cos(− wτ ) + cos(2 wt + wτ + 2Θ) 
= A 2 E[cos( wt + Θ) cos( wt + wτ + Θ)] = A 2 E  
 2

=
A2
2
{ 2π
cos( wτ ) + ∫ cos(2 wt + wτ + 2θ ) f Θ (θ )dθ
0
}
A2  1 2π 
= cos( wτ ) + ∫ cos(2 wt + wτ + 2θ )dθ 
2  2π 0

2π
A2  1  sin( 2wt + wτ + 2θ )  
=  cos( w τ ) +  
2  2π  2 0 

A2
= cos( wτ )
2
Since R XX (t , t + τ ) is independent of t and is a function of τ only, we conclude that
X (t ) is a wide-sense stationary process.

(b) The autocorrelation function of Y (t ) is given by

9
 RYY (t , t + τ ) = E[Y (t )Y (t + τ )] = E[ B sin( wt + Θ) B sin( wt + wτ + Θ)]
 cos(− wτ ) − cos(2wt + wτ + 2Θ) 
= B 2 E[sin( wt + Θ) sin( wt + wτ + Θ)] = B 2 E  
 2

=
B2
2
{ 2π
cos( wτ ) − ∫ cos(2 wt + wτ + 2θ ) f Θ (θ )dθ
0
}
2
B  1 2π 
= cos(wτ ) − ∫ cos(2wt + wτ + 2θ )dθ 
2  2π 0

2π
B2  1  sin(2 wt + wτ + 2θ )  
= cos( wτ ) −  
2  2π  2 0 

B2
= cos( wτ )
2

Since RYY (t , t + τ ) is independent of t and is a function of τ only, we conclude that Y (t )

is a wide-sense stationary process.

(c) The crosscorrelation function of X (t ) and Y (t ) is given by

R XY (t , t + τ ) = E[ X (t )Y (t + τ )] = E[ A cos(wt + Θ) B sin( wt + wτ + Θ)]

 sin(2wt + wτ + 2Θ) − sin(− wτ ) 
= ABE[cos( wt + Θ) sin( wt + wτ + Θ)] = ABE  
 2

=
AB
2
{ 2π
sin( wτ ) + ∫ sin(2wt + wτ + 2θ ) f Θ (θ )dθ
0
}
2
B  1 2π 
= sin( wτ ) + ∫ sin(2 wt + wτ + 2θ )dθ 
2  2π 0

2π
AB  1  sin(2wt + wτ + 2θ )  
= sin( wτ ) −  
2  2π  2 0 

AB
= sin( wτ )
2
Since R XY (t , t + τ ) is independent of t and is a function of τ only, we conclude that
X (t ) and Y (t ) are jointly wide-sense stationary.

Section 10.5: Wide-Sense Stationary Processes

10.15 Two random processes X(t) and Y(t) are defined as follows:
X (t ) = A cos(w1t + Θ)
Y (t ) = B sin( w2 t + Φ )

10
 where w1 , w2 , A and B are constants, and Θ and Φ are statistically independent random
variables, each of which is uniformly distributed between 0 and 2π .
(a) Find the crosscorrelation function R XY (t , t + τ ), and show that X(t) and Y(t) are jointly
wide-sense stationary.
(b) If Θ = Φ, show that X(t) and Y(t) are not jointly wide-sense stationary.
(c) If Θ = Φ, under what condition are X(t) and Y(t) jointly wide-sense stationary?
Solution: We are given that
 21π 0 ≤ θ ≤ 2π
f Θ (θ ) = f Φ (θ ) = 
0 otherwise

(a) The crosscorrelation function of X (t ) and Y (t ) is given by

R XY (t , t + τ ) = E[ X (t )Y (t + τ )] = E[ A cos(w1t + Θ) B sin( w2 t + w2τ + Φ)]

= ABE[cos(w1t + Θ) sin( w2 t + w2τ + Θ)]
 sin( w1t + w2 t + w2τ + Θ + Φ ) − sin( w1t − w2 t − w2τ + Θ − Φ) 
= ABE  
 2 
=
AB
{E[sin( w1t + w2t + w2τ + Θ + Φ)] − E[sin(w1t − w2t − w2τ + Θ − Φ ]}
2
Since Θ and Φ are statistically independent random variables, their joint PDF is the
product of their marginal PDFs. Thus,
E [sin( w1t + w2 t + w2τ + Θ + Φ )] = ∫
2π 2π
∫ sin({w1 + w2 }t + w2τ + θ + φ ) f Θ (θ ) f Φ (φ )dθdφ
0 0

1 2π 2π
= ∫ ∫ sin({w1 + w2 }t + w2τ + θ + φ )dθdφ
4π 2 0 0

=0
E [sin( w1t − w2 t − w2τ + Θ − Φ)] = ∫
2π 2π
∫ sin({w1 − w2 }t − w2τ + θ − φ ) f Θ (θ ) f Φ (φ )dθdφ
0 0

1 2π 2π
= ∫ ∫ sin({w1 − w2 }t − w2τ + θ − φ )dθdφ
4π 2 0 0

=0
This implies that R XY (t , t + τ ) = 0, which shows that X (t ) and Y (t ) are jointly wide-
sense stationary.
(b) If Θ = Φ, then we have that
 sin( w1t + w2 t + w2τ + 2Θ) − sin( w1t − w2 t − w2τ ) 
R XY (t , t + τ ) = ABE  
 2 
AB
= sin( w1t − w2 t − w2τ )
2

11
 Since R XY (t , t + τ ) is not a function of τ only, we conclude that X (t ) and Y (t ) are not
jointly wide-sense stationary.

(c) From the result in part (b) above, we can see that when Θ = Φ, the condition under
which X (t ) and Y (t ) are jointly wide-sense stationary is that w1 = w2 .

10.16 Explain why the following matrices can or cannot be valid autocorrelation matrices of a
zero-mean wide-sense stationary random process X(t).
(a)
1.0 1.2 0.4 1.0 
1.2 1.0 0.6 0.9
G= 
0.4 0.6 1.0 1.3 
 
1.0 0.9 1.3 1.0 
(b)
2.0 1.2 0.4 1.0 
1.2 2.0 0.6 0.9
H=
0.4 0.6 2.0 1.3 
 
1.0 0.9 1.3 2.0

(c)
1.0 0.7 0.4 0.8
0.5 1.0 0.6 0.9
K=
0.4 0.6 1.0 0.3
 
1.0 0.9 0.3 1.0 
Solution:
(a) Since the diagonal elements are supposed to be R XX (0), their value puts an upper bound
on the other entries because we know that for a wide-sense stationary process
X (t ), R XX (τ ) ≤ R XX (0) for all τ ≠ 0. Thus, although G is a symmetric matrix, it contains
off-diagonal elements whose values are larger than the value of the diagonal elements.
Therefore, G cannot be the autocorrelation matrix of a wide-sense stationary process.
(b) H is a symmetric matrix and the diagonal elements are identical and have the largest
value in the matrix. Therefore, H can be the autocorrelation matrix of a wide-sense
stationary process.
(c) The fact that K is not a symmetric matrix means that it cannot be the autocorrelation
matrix of a wide-sense stationary process.

10.17 Two jointly stationary random processes X(t) and Y(t) are defined as follows:
X (t ) = 2 cos(5t + Φ )
Y (t ) = 10 sin(5t + Φ)
12
 where Φ is a random variable that is uniformly distributed between 0 and 2π . Find the
crosscorrelation functions R XY (τ ) and RYX (τ ).
Solution: We are given that
1 0 ≤ φ ≤ 2π
f Φ (φ ) =  2π
0 otherwise

(a) The crosscorrelation function of X (t ) and Y (t ), R XY (τ ), is given by

R XY (t , t + τ ) = E[ X (t )Y (t + τ )] = E[2 cos(5t + Φ )10 sin(5t + 5τ + Φ)]

= 20 E[cos(5t + Φ ) sin(5t + 5τ + Φ )] = 20 E[sin(5t + 5τ + Φ ) cos(5t + Φ)]
 sin(10t + 5τ + 2Φ ) + sin(5τ ) 
= 20 E 
 2  = 10 E[sin(10t + 5τ + 2Φ) + sin(5τ )]
2π
= 10 sin(5τ ) + 10 ∫ sin(10t + 5τ + 2φ ) f Φ (φ )dφ
0

5 2π
= 10 sin(5τ ) + sin(10t + 5τ + 2φ )dφ
π∫ 0

= 10 sin(5τ )
(b) The crosscorrelation function of Y (t ) and X (t ), RYX (τ ), is given by

RYX (t , t + τ ) = E[Y (t ) X (t + τ )] = E[10 sin(5t + Φ )2 cos(5t + 5τ + Φ )]

= 20 E[sin(5t + Φ) cos(5t + 5τ + Φ )]
 sin(10t + 5τ + 2Φ ) + sin(−5τ ) 
= 20 E 
 2  = 10 E[sin(10t + 5τ + 2Φ ) − sin(5τ )]
2π
= −10 sin(5τ ) + 10∫ sin(10t + 5τ + 2φ ) f Φ (φ )dφ
0

5 2π
= −10 sin(5τ ) + sin(10t + 5τ + 2φ )dφ
π∫ 0

= −10 sin(5τ )

10.18 State why each of the functions, F (τ ), G (τ ) and H (τ ), shown in Figure 10.7 can or
cannot be a valid autocorrelation function of a wide-sense stationary process.
F(τ ) G(τ ) H (τ )
1 1
1

0.5

τ τ τ
−2 −1 0 1 2 3 −2 −1 0 1 2 −2 −1 0 1 2
(a) (b) (c)
Figure 10.7: Figure for Problem 10.18

13
Solution:
(a) Consider the function F (τ ). Because F (τ ) is not an even function, it cannot be the
autocorrelation function of a wide-sense stationary process.
(b) Consider the function G (τ ). Although it is an even function, G (0) is not the largest value
of the function. In particular, G (0) < G (−1) = G (1). Since the autocorrelation function
R XX (τ ) of a wide-sense stationary process X (t ) has the property that R XX (τ ) ≤ R XX (0)
for all τ ≠ 0, we conclude that G (τ ) cannot be the autocorrelation function of a wide-
sense stationary process.
(c) The function H (τ ) is an even function and has the property that H (τ ) ≤ H (0) for all
τ ≠ 0. Therefore, we conclude that it can be the autocorrelation function of a wide-sense
stationary process.

10.19 A random process Y(t) is given by

Y (t ) = A sin(Wt + Φ )
where A, W and Φ are independent random variables. Assume that A has a mean of 3 and a
variance of 9, Φ is uniformly distributed between − π and π , and W is uniformly distrib-
uted between − 6 and 6. Determine if the process is stationary in the wide sense.
Solution: We are given that
1 −π ≤ φ ≤ π
f Φ (φ ) =  2π
0 otherwise
1 −6 ≤ w≤ 6
f W ( w) =  12
0 otherwise
E[ A] = 3, σ A2 = 9 ⇒ E[ A 2 ] = σ A2 + (E[ A]) = 18
2

The autocorrelation function of Y (t ) is given by

RYY (t , t + τ ) = E[Y (t )Y (t + τ )] = E[ A sin(Wt + Φ) A sin(Wt + Wτ + Φ)]

 cos(−Wτ ) − cos(2Wt + Wτ + 2Φ ) 
= E[ A 2 sin(Wt + Φ ) sin(Wt + Wτ + Φ)] = E  A 2 
 2 
= E[ A 2 ]E [cos(Wτ ) − cos(2Wt + Wτ + 2Φ )]
1
2
= 9 E [cos(Wτ ) − cos(2Wt + Wτ + 2Φ )]
Now,
1  sin( wτ ) 
6

E [cos(Wτ )] = ∫ cos( wτ ) f W ( w)dw = ∫ cos( wτ )dw = 

6 1 6
−6 12 − 6 12  τ 
−6

sin(6τ ) − sin(−6τ ) sin(6τ ) + sin(6τ ) sin(6τ )

= = =
12τ 12τ 6τ

14
Also, because W and Φ are statistically independent random variables, their joint PDF is the
product of their marginal PDFs. Thus,
π
E [cos(2Wt + Wτ + 2Φ)] = ∫
6
∫φ cos(2 wt + wτ + 2φ ) f Φ ,W (φ , w)dφdw
w= −6 = −π
6 π
=∫ ∫φ cos(2 wt + wτ + 2φ ) f Φ (φ ) f W ( w)dφdw
w= −6 = −π

1 6 π
= ∫ ∫ cos(2wt + wτ + 2φ )dφdw
24π w= −6 φ = −π

π
1 6  cos(2 wt + wτ + 2φ ) 
=
24π ∫w=−6  2  dw
−π

=0
Thus, we obtain
 sin(6τ )  3 sin(6τ )
RYY (t , t + τ ) = 9 E [cos(Wτ ) − cos(2Wt + Wτ + 2Φ )] = 9 =
 6τ  2τ
Since RYY (t , t + τ ) is independent of t and is a function of τ only, we conclude that Y (t ) is a
wide-sense stationary process.

10.20 A random process X(t) is given by

X (t ) = A cos(t ) + ( B + 1) sin(t ) −∞ < t < ∞
where A and B are independent random variables with E[ A] = E[ B] = 0 and
E[ A 2 ] = E[ B 2 ] = 1. Is X(t) wide-sense stationary?
Solution: The autocorrelation function of X (t ) is given by

R XX (t , t + τ ) = E[ X (t ) X (t + τ )] = E[{A cos(t )( B + 1) sin(t )}{A cos(t + τ )( B + 1) sin(t + τ )}]

= E[ A 2 ] cos(t ) cos(t + τ ) + E[ A]{E[ B] + 1}[cos(t ) sin(t + τ ) + sin(t ) cos(t + τ )]
{ }
+ E[ B 2 ] + 2 E[ B ] + 1 sin(t ) sin(t + τ )
= E[ A 2 ] cos(t ) cos(t + τ ) + E[ B 2 ] sin(t ) sin(t + τ ) + sin(t ) sin(t + τ )
= cos(t ) cos(t + τ ) + 2 sin(t ) sin(t + τ )
Since R XX (t , t + τ ) is not independent of t, we conclude that X (t ) is not a wide-sense stationary
process.
10.21 A random process has the autocorrelation function
16τ 2 + 28
R XX (τ ) =
τ 2 +1
Find the mean-square value, the mean value and the variance of the process.
Solution: We can rewrite the autocorrelation function as follows:

16τ 2 + 28 16(τ 2 + 1) − 16 + 28 16(τ 2 + 1) + 12 12

R XX (τ ) = = = = 16 + 2
τ +1
2
τ +1
2
τ +1
2
τ +1

15
 (a) E[ X 2 (t )] = R XX (0) = 16 + 12 = 28
(b) E[ X (t )] = ± lim R XX (τ ) = ± 16 + 0 = ±4
|τ |→∞

(c) σ X2 ( t ) = E[ X 2 (t )] − (E[ X (t )]) = 28 − 16 = 12

2

10.22 A wide-sense stationary random process X(t) has a mean-square value (or average power)
E[ X 2 (t )] = 11. Give reasons why the functions given below can or cannot be its
autocorrelation function.
11sin(2τ )
(a) R XX (τ ) =
1+τ 2
11τ
(b) R XX (τ ) =
1 + 3τ 2 + 4τ 4
τ 2 + 44
(c) R XX (τ ) = 2
τ +4
11cos(τ )
(d) R XX (τ ) =
1 + 3τ 2 + 4τ 4
11τ 2
(e) R XX (τ ) =
1 + 3τ 2 + 4τ 4
Solution:
11sin(2τ )
(a) If R XX (τ ) = , then the average power is E[ X 2 (t )] = R XX (0) = 0 ≠ 11. This
1+τ 2

means that this function cannot be the autocorrelation function of X (t ). Note also that the
function is not an even function, which further disqualifies it as a valid autocorrelation
function of a wide-sense stationary process.
11τ
(b) If R XX (τ ) = , then the average power is E[ X 2 (t )] = R XX (0) = 0 ≠ 11. This
1 + 3τ + 4τ
2 4

means that this function cannot be the autocorrelation function of X (t ). Note also that as
in the previous case, the function is not an even function, which further disqualifies it as a
valid autocorrelation function of a wide-sense stationary process
τ 2 + 44
(c) If R XX (τ ) = 2 , then the average power is E[ X 2 (t )] = R XX (0) = 11. Because the
τ +4
function is also an even function and R XX (0) ≥ R XX (τ ), τ ≠ 0, we conclude that this
function can be the autocorrelation function of X (t ).
11cos(τ )
(d) If R XX (τ ) = , then the average power is E[ X 2 (t )] = R XX (0) = 11. Because
1 + 3τ + 4τ
2 4

the function is also an even function and R XX (0) ≥ R XX (τ ), τ ≠ 0, we conclude that this
function can be the autocorrelation function of X (t ).

16
 11τ 2
(e) If R XX (τ ) = , then the average power is E[ X 2 (t )] = R XX (0) = 0 ≠ 11. This
1 + 3τ 2 + 4τ 4
means that even though this function is an even function, it cannot be the autocorrelation
function of X (t ).

10.23 An ergodic random process X(t) has the autocorrelation function

36τ 2 + 40
R XX (τ ) =
τ 2 +1
Determine the mean value, mean-square value, and variance of X(t).
Solution: We can rewrite the autocorrelation function as follows:

36τ 2 + 40 36(τ 2 + 1) − 36 + 40 36(τ 2 + 1) + 4 4

R XX (τ ) = = = = 36 + 2
τ +1
2
τ +1
2
τ +1
2
τ +1

(a) E[ X (t )] = ± lim R XX (τ ) = ± 36 + 0 = ±6
|τ |→∞

(b) E[ X 2 (t )] = R XX (0) = 36 + 4 = 40
(c) σ X2 (t ) = E[ X 2 (t )] − (E[ X (t )]) = 40 − 36 = 4
2

10.24 Assume that X(t) is the sum of a deterministic quantity Q and a wide-sense stationary
noise process N(t). Determine the following:
(a) the mean of X(t)
(b) the autocorrelation function of X(t)
(c) the autocovariance function of X(t)
Solution: We are given that X (t ) = Q + N (t ), where E[ N (t )] = 0 and RNN (t , t + τ ) = RNN (τ ).
(a) E[ X (t )] = E[Q + N (t )] = E[Q ] + E[ N (t )] = Q + 0 = Q
(b) The autocorrelation function of X (t ) is given by

R XX (t , t + τ ) = E[ X (t ) X (t + τ )] = E [{Q + N (t )}{Q + N (t + τ )}]

= E[Q 2 + QN (t + τ ) + N (t )Q + N (t ) N (t + τ )]
= E[Q 2 ] + Q{E[ N (t + τ )] + E[ N (t )]} + E[ N (t ) N (t + τ )]
= Q 2 + R NN (τ )

(c) The autocovariance function of X (t ) is given by

C XX (t , t + τ ) = R XX (t , t + τ ) − µ X (t ) µ X (t + τ ) = Q 2 + R NN (τ ) − Q 2 = R NN (τ )

10.25 Two statistically independent and zero-mean random processes X(t) and Y(t) have the
following autocorrelation functions, respectively:
R XX (τ ) = e −|τ |
RYY (τ ) = cos(2πτ )
17
 Determine the following:
(a) the autocorrelation function of the process U (t ) = X (t ) + Y (t )
(b) the autocorrelation function of the process V (t ) = X (t ) − Y (t )
(c) the crosscorrelation function of U(t) and V(t)
Solution: We are given also that µ X (t ) = µY (t ) = 0.
(a) The autocorrelation function of the process U (t ) = X (t ) + Y (t ) is given by

RUU (t , t + τ ) = E[U (t )U (t + τ )] = E[{X (t ) + Y (t )}{X (t + τ ) + Y (t + τ )}]

= E[ X (t ) X (t + τ )] + E[ X (t )Y (t + τ )] + E[Y (t ) X (t + τ )] + E[Y (t )Y (t + τ )]
= E[ X (t ) X (t + τ )] + E[ X (t )]E[Y (t + τ )] + E[Y (t )]E[ X (t + τ )] + E[Y (t )Y (t + τ )]
= E[ X (t ) X (t + τ )] + E[Y (t )Y (t + τ )]
= R XX (τ ) + RYY (τ ) = e −|τ | + cos(2πτ )
(b) The autocorrelation function of the process V (t ) = X (t ) − Y (t ) is given by

RVV (t , t + τ ) = E[V (t )V (t + τ )] = E[{X (t ) − Y (t )}{X (t + τ ) − Y (t + τ )}]

= E[ X (t ) X (t + τ )] − E[ X (t )Y (t + τ )] − E[Y (t ) X (t + τ )] + E[Y (t )Y (t + τ )]
= E[ X (t ) X (t + τ )] − E[ X (t )]E[Y (t + τ )] − E[Y (t )]E[ X (t + τ )] + E[Y (t )Y (t + τ )]
= E[ X (t ) X (t + τ )] + E[Y (t )Y (t + τ )]
= R XX (τ ) + RYY (τ ) = e −|τ | + cos(2πτ )

(c) The crosscorrelation function of U(t) and V(t) is given by

RUV (t , t + τ ) = E[U (t )V (t + τ )] = E [{X (t ) + Y (t )}{X (t + τ ) − Y (t + τ )}]

= E[ X (t ) X (t + τ )] − E[ X (t )Y (t + τ )] + E[Y (t ) X (t + τ )] − E[Y (t )Y (t + τ )]
= E[ X (t ) X (t + τ )] − E[ X (t )]E[Y (t + τ )] + E[Y (t )]E[ X (t + τ )] − E[Y (t )Y (t + τ )]
= E[ X (t ) X (t + τ )] + E[Y (t )Y (t + τ )]
= R XX (τ ) − RYY (τ ) = e −|τ | − cos(2πτ )

Section 10.6: Ergodic Random Processes

10.26 A random process Y(t) is given by
Y (t ) = A cos(wt + Φ )
where w is a constant, and A and Φ are independent random variables. The random vari-
able A has a mean of 3 and a variance of 9, and Φ is uniformly distributed between − π
and π . Determine if the process is a mean-ergodic process.
Solution: We are given that
1 −π ≤ φ ≤ π
f Φ (φ ) =  2π
0 otherwise
E[ A] = 3, σ A2 = 9

18
The ensemble average of Y (t ) is given by
π
E[Y (t )] = E[ A cos(wt + Φ )] = E[ A]E[cos( wt + Φ)] = 3∫ cos( wt + φ ) f Φ (φ )dφ
−π

3 3
[sin(wt + φ )]π−π = 3 {sin( wt + π ) − sin(wt − π )}
π
= ∫ cos(wt + φ )dφ =
2π −π 2π 2π
3
= {− sin(wt ) + sin( wt )} = 0
2π
Similarly, the time average of Y (t ) is given by
A  sin( wt + Φ ) 
T
1 T 1 T
Y (t ) = lim
T →∞ 2T ∫−T Y (t )dt = Tlim∫−T A cos(wt + Φ)dt = Tlim
→∞ 2T →∞ 2T 
 w 
−T

= lim
A
T →∞ 2 wT
{sin( wT + Φ) − sin(− wT + Φ)} = Tlim A
→∞ wT
sin( wT ) cos(Φ )

 sin( wT ) 
= A cos(Φ ) lim  = A cos(Φ ) lim [sinc( wT )] = 0
T →∞
 wT  T →∞

where sinc( x) = sin( x) / x. Thus, since the ensemble average of Y (t ) is equal to its time average,
we conclude that Y (t ) is a mean-ergodic process.

10.27 A random process X(t) is given by

X (t ) = A
where A is a random variable with a finite mean of µ A and finite variance σ A2 . Determine if
X(t) is a mean-ergodic process.
Solution: The ensemble average of X (t ) is given by

E[ X (t )] = E[ A] = µ A
The time average of X (t ) is given by
1 T 1 T 2 AT
X (t ) = lim ∫ X (t )dt = lim ∫ Adt = lim = A ≠ µA
T →∞ 2T −T T →∞ 2T −T T →∞ 2T

Thus, we conclude that X (t ) is not a mean-ergodic process.

Section 10.7: Power Spectral Density

10.28 Assume that V(t) and W(t) are both zero-mean wide-sense stationary random processes
and let the random process M(t) be defined as follows:
M (t ) = V (t ) + W (t )
(a) If V(t) and W(t) are jointly wide-sense stationary, determine the following in terms of
those of V(t) and W(t):
(i) the autocorrelation function of M(t)
(ii) the power spectral density of M(t)
(b) If V(t) and W(t) are orthogonal, determine the following in terms of those of V(t) and
W(t):
19
 (i) the autocorrelation function of M(t)
(ii) the power spectral density of M(t)
Solution:
(a) Given that V (t ) and W (t ) are jointly wide-sense stationary, then
RMM (t , t + τ ) = E[ M (t ) M (t + τ )] = E [{V (t ) + W (t )}{V (t + τ ) + W (t + τ )}]
= E[V (t )V (t + τ )] + E[V (t )W (t + τ )] + E[W (t )V (t + τ )] + E[W (t )W (t + τ )]
= RVV (τ ) + RVW (τ ) + RWV (τ ) + RWW (τ ) = RMM (τ )
∞
S MM ( w) = ∫ RMM (τ )e − jwτ dτ = SVV ( w) + SVW ( w) + SWV ( w) + SWW ( w)
−∞

(b) Given that V (t ) and W (t ) are orthogonal, then RVW (τ ) = RWV (τ ) = 0, which means that
RMM (t , t + τ ) = RVV (τ ) + RWW (τ ) = RMM (τ )
∞
S MM ( w) = ∫ RMM (τ )e − jwτ dτ = SVV ( w) + SWW ( w)
−∞

10.29 A stationary random process X(t) has an autocorrelation function given by

R XX (τ ) = 2e −|τ | + 4e −4|τ |
Find the power spectral density of the process.
Solution: The power spectral density of the process is given by
∞
S XX ( w) = ∫ R XX (τ )e − jwτ dτ
−∞
0 ∞ 0 ∞
= ∫ 2eτ e − jwτ dτ + ∫ 2e −τ e − jwτ dτ + ∫ 4e 4τ e − jwτ dτ + ∫ 4e −4τ e − jwτ dτ
−∞ 0 −∞ 0
0 ∞ 0 ∞
= 2∫ e (1− jw)τ dτ + 2 ∫ e −(1+ jw )τ dτ + 4∫ e ( 4− jw)τ dτ + 4 ∫ e −(1+ jw )τ dτ
−∞ 0 −∞ 0
0 ∞ 0 ∞
 e (1− jw)τ   e −(1+ jw)τ   e ( 4− jw )τ   e −( 4+ jw)τ 
= 2  − 2   + 4   − 4  
 1 − jw  −∞  1 + jw  0  4 − jw  −∞  4 + jw  0
2 2 4 4
= + + +
1 − jw 1 + jw 4 − jw 4 + jw
4 32
= +
1+ w 2
16 + w 2

10.30 A random process X(t) has a power spectral density given by

 w2
 4 − |w| ≤ 6
S XX ( w) =  9
0 otherwise

Determine (a) the average power and (b) the autocorrelation function of the process.
Solution:
20
(a) The average power of the process is given by
6
1 ∞ 1 6  w2  1  w3 
E[ X (t )] = R XX (0) = ∫−∞ S XX ( w)dw = 2π ∫−6  9  2π 
− = −
2
4 dw 4 w 
2π 27  −6
16
=
π

(b) The autocorrelation function of the process can be obtained as follows:

1 ∞ 1 6  w 2  jwτ
R XX (τ ) = ∫ S XX ( w)e jwτ dw = ∫−6  9 e dw
4 −
2π −∞ 2π
 4e jwτ  6 1 6
1 
2 jwτ
=   − ∫−6w e dw
2π
 jτ  −6 9 
4  e j 6τ − e − j 6τ  1 6 2 jwτ 4 1 6
=  − ∫ w e dw = sin(6τ ) − ∫ w 2 e jwτ dw
πτ  2j  18π − 6 πτ 18π −6

Let u = w 2 ⇒ du = 2wdw, and let dv = e jwτ dw ⇒ v = e jwτ / jτ . Thus,

6
6
jwτ  w 2 e jwτ  6 2 we
jwτ
72  e j 6τ − e − j 6τ  2 6
∫
2
w e dw =   − ∫ − 6 jτ dw = − ∫ we jwτ dw
−6
 jτ  − 6 τ  2j  jτ −6

72 2 6 jwτ
sin(6τ ) −
jτ ∫ − 6
= we dw
τ
Let u = w ⇒ du = dw, and let dv = e jwτ dw ⇒ v = e jwτ / jτ . Thus,
6 6
6
jwτ  we jwτ  6 e
jwτ
12  e j 6τ + e − j 6τ  1  e jwτ 
∫ −6
we dw = 
 jτ  −6
 − ∫−6 jτ dw = 
jτ  2
−  
 jτ  jτ  −6
12 2 j  e j 6τ − e − j 6τ  12 2j
cos(6τ ) + 2  = = cos(6τ ) + 2 sin(6τ )
jτ τ  2j  jτ τ
From these results we obtain
6
2 jwτ 72 2  12 2j 
∫−6w e dw = τ sin(6τ ) − jτ  jτ cos(6τ ) + τ 2 sin(6τ )
72 24 4
= sin(6τ ) + cos(6τ ) − sin(6τ )
τ τ 2
τ3
Therefore,

21
 4 1 6 2 jwτ
R XX (τ ) = sin(6τ ) −
18π ∫−6
w e dw
πτ
4 1  72 24 4 
= sin(6τ ) −  sin(6τ ) + 2 cos(6τ ) − 3 sin(6τ )
πτ 18π  τ τ τ 
4 4 4 2
= sin(6τ ) − sin(6τ ) − cos(6τ ) + sin(6τ )
πτ πτ 3πτ 2
9πτ 3
2
= {sin(6τ ) − 6τ cos(6τ )}
9πτ 3

Note that with repeated application of L-Hopital’s rule, it can be shown that

16
E[ X 2 (t )] = R XX (0) = lim R XX (τ ) =
τ →0 π
as we obtained earlier.

10.31 A random process X(t) has the power spectral density

9
S XX ( w) = 2
w + 64
Find (a) the average power in the process and (b) the autocorrelation function.
Solution: To find the average power in the process and the autocorrelation function of the
process we proceed as follows: From Table 10.1 we observe that
2a
e − a|τ | ↔ 2
w + a2
Now,
9 9 9  2(8) 
= 2 =  2 
w + 64 w + 8
2 2
16  w + 8 2 

Thus, we have that a = 8 and we obtain

9 −8|τ |
R XX (τ ) = e
16
9
E[ X 2 (t )] = R XX (0) =
16

10.32 A random process Z(t) has the autocorrelation function given by

 τ
1 + τ −τ 0 ≤ τ ≤ 0
 0

 τ
RZZ (τ ) = 1 − 0 ≤τ ≤τ0
 τ0
0 otherwise



22
 where τ 0 is a constant. Calculate the power spectral density of the process.
Solution: The power spectral density of the process is given by
∞ 0  τ  τ0  τ 
S XX ( w) = ∫ R XX (τ )e − jwτ dτ = ∫ 1 + e − jwτ dτ + ∫ 1 − e − jwτ dτ
−∞ −τ 0
 τ0  0
 τ0 
0 1 0 τ0 1 τ0
= ∫ e − jwτ dτ + ∫ τe − jwτ dτ + ∫ e − jwτ dτ − ∫ τe − jwτ dτ
−τ 0 τ0 −τ 0 0 τ0 0

− jwτ τ0
 ee  1 0
− jwτ 1 τ0
= −  + ∫ ττe dτ − ∫ τe − jwτ dτ
 jw  −τ 0 τ 0 τ0
− 0 0

jwτ 0 − jwτ 0
2  e e − ee  1 0
− jwτ 1 τ0
=  + ∫ ττe dτ − ∫ τe − jwτ dτ
w  2j  τ 0 − 0 τ0 0

2 1 0
− jwτ 1 τ0
= sin( wτ 0 ) + ∫ ττe dτ − ∫ τe − jwτ dτ
w τ0 − 0 τ0 0

Let u = τ ⇒ du = dτ , and let dv = e − jwτ dτ ⇒ v = e jwτ / jw. Thus,

0 0 0
0  τe − jwτ  1 0 − jwτ  τe − jwτ  1  e − jwτ 
∫ ττe dτ =  − τ
− jwτ

jw ∫−τ 0
 + e d =  −  −  
− jw  −τ jw  −τ jw  jw  −τ
0
 0
 0 0

1 τ 0 e jwτ 0 e jwτ 0
=−
1
jw
0 + τ 0e {
jwτ 0 1
+ 2 1− e
w
jwτ 0
= 2−
w jw
} − 2
w
{ }
τ0 τ0 τ0
τ0  τe − jwτ  1 τ 0 − jwτ  τe − jwτ  1  e − jwτ 
∫ τe − jwτ
dτ =  − τ
jw ∫0
 + e d =  −  −  
0
 jw  0  jw  0 jw  jw  0
e − jwτ 0 1 τ 0 e − jwτ 0
=−
1
jw
τ 0e {
− jwτ 0 1 − jwτ 0
+ 2 e
w
−1 = }w2
− 2−
w
{
jw
}
Therefore,
2 1 0 1 τ0
S XX ( w) = sin( wτ 0 ) + ∫ ττe
− jwτ
dτ − ∫ τe − jwτ dτ
w τ0 − 0 τ0 0

2  1 τ 0 e jwτ 0 e jwτ 0  1  e − jwτ 0

1 1 τ 0 e − jwτ 0 
= sin( wτ 0 ) +
 2 − −  −  − − 
w w τ0 jw w2  τ 0  w2 w2 jw 
2 2 2  e jwτ 0 + e − jwτ 0  2  e jwτ 0 − e − jwτ 0 
= sin( wτ 0 ) + 2 − 2  −  
w w τ0 w τ0  2  w 2j 
2 2 2 2 2 2
= sin( wτ 0 ) + 2 − 2 cos( wτ 0 ) − sin( wτ 0 ) = 2 − 2 cos(wτ 0 )
w w τ0 w τ0 w w τ0 w τ0
2
= {1 − cos(wτ 0 )}
w τ0
2

23
10.33 Give reasons why the functions given below can or cannot be the power spectral density
of a wide-sense stationary random process.
sin( w)
(a) S XX ( w) =
w
cos(w)
(b) S XX ( w) =
w
8
(c) S XX ( w) = 2
w + 16
5w 2
(d) S XX ( w) =
1 + 3w 2 + 4 w 4
5w
(e) S XX ( w) =
1 + 3w 2 + 4 w 4
Solution:
sin( w)
(c) The function S XX ( w) = is an even function. However, in addition to being an even
w
function, a valid power spectral density of a wide-sense stationary process should satisfy
the condition that S XX ( w) ≥ 0. Since the above function can take both positive and negative
values, we conclude that it cannot be the power spectral density of a wide-sense stationary
process.
cos( w)
(d) The function S XX ( w) = is not an even function. In addition it has negative values
w
when w is negative. Therefore, it cannot be the power spectral density of a wide-sense
stationary process.
8
(e) The function S XX ( w) = 2 is an even function that is also a non-negative function.
w + 16
Therefore, it can be the power spectral density of a wide-sense stationary process.
5w 2
(f) The function S XX ( w) = is an even function that is also a non-negative
1 + 3w 2 + 4 w 4
function. Therefore, it can be the power spectral density of a wide-sense stationary process.
5w
(g) The function S XX ( w) = is not an even function. In addition it has negative
1 + 3w 2 + 4 w 4
values when w is negative. Therefore, it cannot be the power spectral density of a wide-
sense stationary process.
10.34 A bandlimited white noise has the power spectral density defined by
0.01 400π ≤ | w | ≤ 500π
S XX ( w) = 
0 otherwise

Find the mean-square value of the process.

Solution: The power spectral density can be sketched as shown below.

24
 SNN (w)

0.01

− 500π − 400π 400π 500π

The mean-square value of the process is given by

E[ N 2 (t )] = R NN (0) =
1
2π ∫
∞

−∞
S NN ( w)dw = {
1 − 400π
∫
2π −500π
0.01dw + ∫
500π

400π
0.01dw }
=
0.01
2π
{
[w]−−500π + [w]400π
400π 500π
} =
0.01
2π
{200π }
=1

10.35 A wide-sense stationary process X(t) has an autocorrelation function

R XX (τ ) = ae −4|τ |
where a is a constant. Determine the power spectral density.
Solution: The power spectral density can be determined by noting from Table 10.1 that
2β
e − β |τ | ↔ 2 . Thus, since β = 4, we have that
β + w2
 2( 4)  8a
S XX ( w) = a  2 2 
= 2
w + 4  w + 4
2

10.36 Two random processes X(t) and Y(t) are defined as follows:
X (t ) = A cos( w0 t ) + B sin( w0 t )
Y (t ) = B cos( w0 t ) − A sin( w0 t )

where w0 is a constant, and A and B zero-mean and uncorrelated random variables with
variances σ A2 = σ B2 = σ 2 . Find the cross-power spectral density of X(t) and Y(t), S XY (w).
(Note that S XY ( w) is the Fourier transform of the crosscorrelation function R XY (τ ). )
Solution: The cross-power spectral density of X(t) and Y(t) can be obtained as follows:

25
 R XY (τ ) = E[ X (t )Y (t + τ )]
= E [{A cos( w0t ) + B sin( w0 t )}{B cos(w0 t + w0τ ) − A sin( w0t + w0τ )}]
= E[ AB ] cos( w0t ) cos(w0 t + w0τ ) − E[ A 2 ] cos(w0 t ) sin( w0t + w0τ )
− E[ BA] sin( w0 t ) sin( w0 t + w0τ ) + E[ B 2 ]sin( w0t ) cos(w0 t + w0τ )
= E[ B 2 ] sin( w0 t ) cos( w0t + w0τ ) − E[ A 2 ] cos( w0 t ) sin( w0 t + w0τ )
= σ 2 {sin( w0t ) cos( w0 t + w0τ ) − cos( w0 t ) sin( w0 t + w0τ )}
= −σ 2 sin( w0τ )
S XY ( w) = jσ 2π {δ ( w − w0 ) − δ ( w + w0 )}
10.37 Two random processes X(t) and Y(t) are both zero-mean and wide-sense stationary
processes. If we define the random process Z (t ) = X (t ) + Y (t ), determine the power
spectral density of Z(t) under the following conditions:
(a) X(t) and Y(t) are jointly wide-sense stationary
(b) X(t) and Y(t) are orthogonal.
Solution: The power spectral density of Z(t) can be obtained as follows:

RZZ (t , t + τ ) = E[ Z (t ) Z (t + τ )] = E[{X (t ) + Y (t )}{X (t + τ ) + Y (t + τ )}]

= E[ X (t ) X (t + τ )] + E[ X (t )Y (t + τ )] + E[Y (t ) X (t + τ )] + E[Y (t )Y (t + τ )]
= R XX (τ ) + E[ X (t )Y (t + τ )] + E[Y (t ) X (t + τ )] + RYY (τ )
(a) If X(t) and Y(t) are jointly wide-sense stationary, then we obtain

RZZ (t , t + τ ) = R XX (τ ) + E[ X (t )Y (t + τ )] + E[Y (t ) X (t + τ )] + RYY (τ )

= R XX (τ ) + R XY (τ ) + RYX (τ ) + RYY (τ ) = RZZ (τ )
∞
SWW ( w) = ∫ RZZ (τ )e − jwτ dτ = S XX ( w) + S XY ( w) + S YX ( w) + S YY ( w)
−∞

(b) If X(t) and Y(t) are orthogonal, then E[ X (t )Y (t + τ )] = E[Y (t ) X (t + τ )] = 0, and we obtain
RZZ (t , t + τ ) = R XX (τ ) + E[ X (t )Y (t + τ )] + E[Y (t ) X (t + τ )] + RYY (τ )
= R XX (τ ) + RYY (τ ) = RZZ (τ )
∞
SWW ( w) = ∫ RZZ (τ )e − jwτ dτ = S XX ( w) + S YY ( w)
−∞

10.38 Two jointly stationary random processes X(t) and Y(t) have the crosscorrelation function
given by:
R XY (τ ) = 2e −2τ τ ≥0

Determine the following:

(a) the cross-power spectral density S XY (w)
(b) the cross-power spectral density S YX (w)

26
Solution:
(a) The cross-power spectral density S XY ( w)
∞
∞ ∞ ∞  e −( 2+ jw)τ 
S XY ( w) = ∫ R XY (τ )e − jwτ dτ = ∫ 2e −2τ e − jwτ dτ = ∫ 2e −( 2+ jw )τ dτ = 2 − 
−∞ 0 0
 2 + jw  0
2
=
2 + jw
(b) The cross-power spectral density S YX (w) is given by
∞ ∞ ∞
SYX ( w) = ∫ RYX (τ )e − jwτ dτ = ∫ R XY (−τ )e − jwτ dτ = ∫ R XY (u )e jwu du = S XY (− w) = S XY
∗
( w)
−∞ −∞ −∞

2
=
2 − jw

10.39 Two jointly stationary random processes X(t) and Y(t) have the cross-power spectral
density given by
1
S XY ( w) =
− w + j 4w + 4
2

Find the corresponding crosscorrelation function.

Solution: First, we observe that

1 1
S XY ( w) = =
− w + j 4 w + 4 (2 + jw)2
2

From Table 10.1, we find that the corresponding crosscorrelation function is

R XY (τ ) = τe −2τ τ ≥0
10.40 Two zero-mean independent wide-sense stationary random processes X(t) and Y(t) have
the following power spectral densities
4
S XX ( w) = 2
w +4
4
SYY ( w) = 2
w +4

respectively. A new random process W(t) is defined as follows: W (t ) = X (t ) + Y (t ). Deter-

mine the following:
(a) the power spectral density of W(t)
(b) the cross-power spectral density S XW (w)
(c) the cross-power spectral density S YW (w)
Solution:
(a) The power spectral density of W(t) can be obtained as follows:

27
 RWW (t , t + τ ) = E[W (t )W (t + τ )] = E[{X (t ) + Y (t )}{X (t + τ ) + Y (t + τ )}]
= E[ X (t ) X (t + τ )] + E[ X (t )Y (t + τ )] + E[Y (t ) X (t + τ )] + E[Y (t )Y (t + τ )]
= E[ X (t ) X (t + τ )] + E[ X (t )]E[Y (t + τ )] + E[Y (t )]E[ X (t + τ )] + E[Y (t )Y (t + τ )]
= R XX (τ ) + RYY (τ ) = RWW (τ )
4 4 8
SWW ( w) = S XX ( w) + SYY ( w) = + 2 = 2
w +4 w +4 w +4
2

(b) The cross-power spectral density S XW (w) can be obtained as follows:

R XW (t , t + τ ) = E[ X (t )W (t + τ )] = E [X (t ){X (t + τ ) + Y (t + τ )}]
= E[ X (t ) X (t + τ )] + E[ X (t )Y (t + τ )] = E[ X (t ) X (t + τ )] + E[ X (t )]E[Y (t + τ )]
= E[ X (t ) X (t + τ )] = R XX (τ ) = R XW (τ )
4
S XW ( w) = S XX ( w) =
w +4
2

(c) The cross-power spectral density S YW (w) can be obtained as follows:

RYW (t , t + τ ) = E[Y (t )W (t + τ )] = E [Y (t ){X (t + τ ) + Y (t + τ )}]

= E[Y (t ) X (t + τ )] + E[Y (t )Y (t + τ )] = E[Y (t )]E[ X (t + τ )] + E[Y (t )Y (t + τ )]
= E[Y (t )Y (t + τ )] = RYY (τ ) = RYW (τ )
4
SYW ( w) = SYY ( w) =
w +4
2

10.41 Two zero-mean independent wide-sense stationary random processes X(t) and Y(t) have
the following power spectral densities:
4
S XX ( w) = 2
w +4
w2
S YY ( w) = 2
w +4

respectively. Two new random processes V(t) and W(t) are defined as follows:
V (t ) = X (t ) + Y (t )
W (t ) = X (t ) − Y (t )

respectively. Determine the cross-power spectral density SVW (w).

Solution: The cross-power spectral density can be obtained as follows:

28
 RVW (t , t + τ ) = E[V (t )W (t + τ )] = E [{X (t ) + Y (t )}{X (t + τ ) − Y (t + τ )}]
= E[ X (t ) X (t + τ )] − E[ X (t )Y (t + τ )] + E[Y (t ) X (t + τ )] − E[Y (t )Y (t + τ )]
= E[ X (t ) X (t + τ )] − E[ X (t )]E[Y (t + τ )] + E[Y (t )]E[ X (t + τ )] − E[Y (t )Y (t + τ )]
= R XX (τ ) − RYY (τ ) = RVW (τ )
4 w2 4 − w2
SVW ( w) = S XX ( w) − SYY ( w) = − =
w2 + 4 w2 + 4 w2 + 4

10.42 A zero-mean wide-sense stationary random process X(t), − ∞ < t < ∞, has the following
power spectral density:
2
S XX ( w) = −∞ < w< ∞
1 + w2

The random process Y(t) is defined by

2
Y (t ) = ∑ X (t + k )
k =0
(a) Find the mean of Y(t).
(b) Find the variance of Y(t).
Solution: We expand Y (t ) as follows:
2
Y (t ) = ∑ X (t + k ) = X (t ) + X (t + 1) + X (t + 2)
k =0
(a) The mean of Y (t ) is given by
E[Y (t )] = E[ X (t )] + E[ X (t + 1)] + E[ X (t + 2)] = 0

(b) To find the variance of Y (t ), we note that because the mean of Y (t ) is zero, the variance
is equal to the second moment. That is, σ Y2 (t ) = E[Y 2 (t )]. Thus, we need to find the
autocorrelation function that will enable us to find the second moment, as follows:

RYY (t , t + τ ) = E[Y (t )Y (t + τ )] = E [{X (t ) + X (t + 1) + X (t + 2)}{X (t + τ ) + X (t + 1 + τ ) + X (t + 2 + τ )}]

= E[ X (t ) X (t + τ )] + E[ X (t ) X (t + 1 + τ )] + E[ X (t ) X (t + 2 + τ )]
+ E[ X (t + 1) X (t + τ )] + E[ X (t + 1) X (t + 1 + τ )] + E[ X (t + 1) X (t + 2 + τ )]
+ E[ X (t + 2) X (t + τ )] + E[ X (t + 2) X (t + 1 + τ )] + E[ X (t + 2) X (t + 2 + τ )]
= R XX (τ ) + R XX (τ + 1) + R XX (τ + 2) + R XX (τ − 1) + R XX (τ ) + R XX (τ + 1) + R XX (τ − 2)
+ R XX (τ − 1) + R XX (τ )
= 3R XX (τ ) + 2 R XX (τ + 1) + R XX (τ + 2) + 2 R XX (τ − 1) + R XX (τ − 2) = RYY (τ )
E[Y 2 (t )] = RYY (0) = 3R XX (0) + 2 R XX (1) + R XX (2) + 2 R XX (−1) + R XX (−2)
= 3R XX (0) + 4 R XX (1) + 2 R XX (2)

29
 where the last equality follows from the fact that for a wide-sense stationary process, the
autocorrelation function is an even function; that is, R XX (−τ ) = R XX (τ ). Also, we know
that
2a
e −a|τ | ↔ 2
a + w2
Thus, we have that
2
S XX ( w) = ⇒ R XX (τ ) = e −|τ |
1+ w 2

σ Y (t ) = E[Y 2 (t )] = 3R XX (0) + 4 R XX (1) + 2 R XX (2) = 3 + 4e −1| + 2e −2| = 4.7422

2

10.43 Consider two individual wide-sense stationary processes X(t) and Y(t). Consider the
random process Z (t ) = X (t ) + Y (t ).
(a) Show that the autocorrelation function of Z(t) is given by
RZZ (t , t + τ ) = R XX (τ ) + RYY (τ ) + R XY (t , t + τ ) + RYX (t , t + τ )
(b) If X(t) and Y(t) are jointly wide-sense stationary, show that the autocorrelation function
of Z(t) is given by

RZZ (t , t + τ ) = R XX (τ ) + RYY (τ ) + R XY (τ ) + RYX (τ )

(c) If X(t) and Y(t) are jointly wide-sense stationary, find the power spectral density of Z(t).
(d) If X(t) and Y(t) are uncorrelated, find the power spectral density of Z(t).
(e) If X(t) and Y(t) are orthogonal, find the power spectral density of Z(t).
Solution:
(a) The autocorrelation function of Z(t) is given by
RZZ (t , t + τ ) = E[ Z (t ) Z (t + τ )] = E[{X (t ) + Y (t )}{X (t + τ ) + Y (t + τ )}]
= E[ X (t ) X (t + τ )] + E[ X (t )Y (t + τ )] + E[Y (t ) X (t + τ )] + E[Y (t )Y (t + τ )]
= R XX (τ ) + R XY (t , t + τ ) + RYX (t , t + τ ) + RYY (τ )
(b) If X(t) and Y(t) are jointly wide-sense stationary, then we have that
RZZ (t , t + τ ) = R XX (τ ) + R XY (τ ) + RYX (τ ) + RYY (τ ) = RZZ (τ )
(c) If X(t) and Y(t) are jointly wide-sense stationary, the power spectral density of Z(t) is
given by
∞
S ZZ ( w) = ∫ RZZ (τ )e − jwτ dτ = S XX ( w) + S XY ( w) + SYX ( w) + SYY ( w)
−∞

(d) If X(t) and Y(t) are uncorrelated, then E[ X (t )Y (t + τ )] = E[Y (t ) X (t + τ )] = µ X µY , and we

obtain
RZZ (t , t + τ ) = R XX (τ ) + E[ X (t )Y (t + τ )] + E[Y (t ) X (t + τ )] + RYY (τ )
= R XX (τ ) + RYY (τ ) + 2µ X µY = RZZ (τ )
Thus, the power spectral density of Z(t) is given by

30
 ∞
S ZZ ( w) = ∫ RZZ (τ )e − jwτ dτ = S XX ( w) + SYY ( w) + 4 µ X µY δ ( w)
−∞

(e) If X(t) and Y(t) are orthogonal, we have that E[ X (t )Y (t + τ )] = R XY (t , t + τ ) = 0 and

E[Y (t ) X (t + τ )] = RYX (t , t + τ ) = 0, which gives
RZZ (t , t + τ ) = R XX (τ ) + RYY (τ ) = RZZ (τ )
Thus, the power spectral density of Z(t) is given by
∞
S ZZ ( w) = ∫ RZZ (τ )e − jwτ dτ = S XX ( w) + SYY ( w)
−∞

Section 10.8: Discrete-time Random Processes

10.44 Find the power spectral density of a random sequence X [n] whose autocorrelation
function is given by R XX [m] = a m , m = 0, 1, 2, Κ , where | a | < 1.
Solution: The power spectral density is given by

[ ]
∞ ∞ ∞
1
S XX (Ω) = ∑ R XX [m]e − jΩm = ∑ a m e − jΩm = ∑ ae − jΩ =
m

m = −∞ m =0 m =0 1 − ae − jΩ

10.45 A wide-sense stationary continuous-time process X(t) has the autocorrelation function
given by
R X C X C (τ ) = e −2|τ | cos( w0τ )

where w0 is a constant. If X(t) is sampled with a sampling period 10 seconds to produce the
discrete-time process X [n], find the power spectral density of X [n]. [Hint: Use Table 10.1
to find S X C X C (w). ]
Solution: From Table 10.1, the power spectral density of X (t ) is given by
2 2
S X C X C ( w) = +
4 + ( w − w0 ) 2
4 + ( w + w0 ) 2
X (t ) is sampled with a sampling period of 10 seconds to produce X [n]. Thus, the power
spectral density of X [n] is given by
 
 
1 ∞
 Ω − 2πm  1 ∞
 2 2 
S XX (Ω) = ∑ S X C X C   = ∑  + 2
 Ω − 2πm  Ω − 2πm
2
Ts m=−∞  Ts  10 m=−∞   
4+ − w0  4+ + w0  
  10   10  
10.46 Periodic samples of the autocorrelation function of white noise N(t) with period T are
defined by
σ 2 k =0
R NN (kT ) =  N
0 k≠0
Find the power spectral density of the discrete-time random process.
31
Solution: The power spectral density of the process is given by
∞ ∞
S NN (Ω) = ∑ RNN [k ]e − jΩk =
k = −∞
∑R
k = −∞
NN (kT )e − jΩk = σ N2 e − jΩ 0 = σ N2

10.47 The autocorrelation function of a discrete-time process is given by

σ X2 k =0
 2
 4σ
R XX [k ] =  2 X2 k ≠ 0, k odd
k π
0 k ≠ 0, k even

Find the power spectral density S XX (Ω) of the process.
Solution: The power spectral density of the process is given by
∞
4σ X2  − jΩ e −3 jΩ + e 3 jΩ e −5 jΩ + e 5 jΩ 
S XX (Ω) = ∑ R XX [k ]e − jΩ k
= σ X + 2 e + e +
2 jΩ
+ +Λ 
k = −∞ π  9 25 
8σ X2  e − jΩ + e jΩ 1  e −3 jΩ + e 3 jΩ  1  e −5 jΩ + e 5 jΩ  
= σ X2 + 2 
+  +   +Λ 
π  2 9 2  25  2  
8σ 2
1 1 1 
= σ X2 + 2X cos(Ω) + cos(3Ω) + cos(5Ω) + cos(7Ω) + Λ 
π  9 25 49 

32
Chapter 11: Linear Systems with Random Inputs

Chapter Summary: This chapter discusses linear systems with random inputs. It also discusses
the autoregressive moving average process.
Section 11.2: Linear Systems with Deterministic Input
11.1 Find the Fourier transform of the following “sawtooth” function x(t) defined in the
interval [−T , T ] :
1 + Tt −T ≤ t ≤ 0
x(t ) =  t
1 − T 0≤t ≤T
Solution: The Fourier transform of x(t ) is given by
∞ 0  t T t
X ( w) = ∫ x(t )e − jwt dt = ∫ 1 + e − jwt dt + ∫ 1 − e − jwt dt
−∞ −T
 T 0
 T
0 T
 e − jwt  1 0 − jwt  e − jwt  1 T
= −  + ∫ te dt +  −  − ∫ te − jwt dt
 jw  −T T −T  jw  0 T 0

=
jw
{
1 jwT
e −1 +
1
}
jw
{ }
1 0
T −T
1 T
1 − e − jwT + ∫ te − jwt dt − ∫ te − jwt dt
T 0
2  e jwT − e − jwT  1 0 − jwt 1 T − jwt
 =  + ∫−Tte dt − ∫0 te dt
w 2j  T T
2 1 0 1 T
= sin (wT ) + ∫ te − jwt dt − ∫ te − jwt dt
w T −T T 0
− jwt − jwt
Let u = t ⇒ du = dt , and let dv = e dt ⇒ v = −e / jw. Thus, we have that
0 0
0
− jwt  te − jwt  1 0 − jwt Te jwT 1  e − jwt 
∫ −T
te dt = −
 jw  −T jw ∫−T
 + e dt = −
jw
+ −
jw  jw  −T


=−
Te jwT
jw
1
[
+ 2 1 − e jwT
w
]
T 0
T
− jwt  te − jwt  1 T − jwt Te − jwT 1  e − jwt 
∫ 0
te dt = −
 jw  0 jw ∫0
 + e dt = −
jw
+ −
jw  jw  −T


=−
Te − jwT
jw
1
[
+ 2 e − jwT − 1
w
]
Thus,

1
 2 1 0 1 T
X ( w) = sin (wT ) + ∫ te − jwt dt − ∫ te − jwt dt
w T −T T 0
2 1 1
= sin (wT ) +  2 1 − e jwT −
w T w
[ jw
+ ]
Te jwT Te − jwT
jw
1 
− 2 e − jwT − 1  [ ]
w 
2 2 2  e jwT + e − jwT  2  e jwT − e − jwT 
= sin (wT ) + 2 − 2  − 2  
w wT wT 2  w  2j 
2 2 2 2 2 2
= sin (wT ) + 2 − 2 cos(wT ) − 2 sin (wT ) = 2 − 2 cos(wT )
w wT wT w wT wT
2
= 2 {1 − cos(wT )}
wT

11.2 Consider a system that performs a differentiation of the input function. This means that
when the input function is x(t), the output function is
d
y (t ) = x(t )
dt
Find the Fourier transform of y(t) in terms of the Fourier transform of x(t).
Solution: The Fourier transform of y (t ) can be obtained as follows:
1 ∞
2π ∫−∞
x(t ) = X ( w)e jwt dw

d
y (t ) = x(t ) =
dt
1 d ∞
2π dt ∫ − ∞
X ( w)e jwt dw =
1 ∞
2π ∫ − ∞
X ( w)
d jwt
dt
e dw { }
1 ∞ 1 ∞
=
2π ∫ − ∞
jwX ( w)e jwt dw =
2π ∫ − ∞
Y ( w)e jwt dw

Thus, Y ( w) = jwX ( w); that is,

d
y (t ) = x(t ) ↔ jwX ( w)
dt
11.3 Consider a system that performs a complex modulation on an input function. This means
that when the input function is x(t), the output function is
y (t ) = e jw0t x(t )
where w0 is a constant. Find the Fourier transform of y(t) in terms of the Fourier
transform of x(t).
Solution: The Fourier transform of x(t ) is given by
∞ ∞ ∞
Y ( w) = ∫ y (t )e − jwt dt = ∫ e jw0t x(t )e − jwt dt = ∫ x(t )e − j ( w− w0 ) t dt
−∞ −∞ −∞

= X ( w − w0 )

2
11.4 Consider a system that introduces a delay of t 0 to an input function. This means that
when the input function is x(t), the output function is
y (t ) = x(t − t 0 )
where t 0 > 0 is a constant. Find the Fourier transform of y(t) in terms of the Fourier
transform of x(t).
Solution: The Fourier transform of x(t ) is given by
∞ ∞
Y ( w) = ∫ y (t )e − jwt dt = ∫ x(t − t 0 )e − jwt dt
−∞ −∞

Let u = t − t 0 ⇒ du = dt. Thus,

∞ ∞ ∞
Y ( w) = ∫ x(t − t 0 )e − jwt dt = ∫ x(u )e − jw ( u +t0 ) du = e − jwt 0 ∫ x(u )e − jwu du
−∞ −∞ −∞
− jwt 0
=e X ( w)
11.5 Consider a system that performs a scaling operation on an input function. This means that
when the input function is x(t), the output function is
y (t ) = x(at )
where a > 0 is a constant. Find the Fourier transform of y(t) in terms of the Fourier
transform of x(t).
Solution: The Fourier transform of x(t ) is given by
∞ ∞
Y ( w) = ∫ y (t )e − jwt dt = ∫ x(at )e − jwt dt
−∞ −∞

Let u = at ⇒ du = adt. Thus,

1 ∞  w
− j  u

du  a ∫−∞
 x(u )e  a  du a>0
∞
− jwu / a
Y ( w) = ∫ x(u )e =
−∞ a  1 ∞  w
− j  u

− a ∫−∞ x(u )e a<0

a
du

1  w
= X 
a a

Section 11.3: Linear Systems with Continuous Random Input

11.6 A stationary zero-mean random signal X(t) is the input to two filters, as shown in Figure
11.7.
Y1 (t )
h1 (t )

X (t )

Y2 (t )
h2 (t )

Figure 11.7: Figure for Problem 11.6

3
 The power spectral density of X(t) is S XX ( w) = N 0 / 2, and the filter impulse responses
are given by

1 0 ≤ t <1
h1 (t ) = 
0 otherwise
2e −t t≥0
h2 (t ) = 
0 otherwise

Determine the following:

(a) The mean E[Yi (t )] and second moment E[Yi 2 (t )] of the output signal Yi (t ) for
i = 1, 2
(b) The crosscorrelation function RY1Y2 (t , t + τ )
Solution: The system responses of the filters are given by
1
 e − jwt 
∞
H 1 ( w) = ∫ h1 (t )e
−∞
− jwt
dt = ∫ e
1
− jwt
dt = −
jw
 =
1
jw
1 − e − jw { }
0
 0
∞
∞
− jwt
∞
−t − jwt
∞
− (1+ jw ) t  e − jwt  2
H 2 ( w) = ∫ h2 (t )e dt = 2 ∫ e e dt = 2 ∫ e dt = 2 −  =
−∞ 0 0
 1 + jw  0 1 + jw

(a) The mean E[Yi (t )] of the output signal Yi (t ) for i = 1, 2, is given by

E[Yi (t )] = µ X (t ) ∗ hi (t )
Thus, if X (t ) is a zero-mean process, the output process is also a zero-mean process.
The second moment E[Yi 2 (t )] of the output signal Yi (t ) for i = 1, 2 is given by
1 ∞ 1 ∞ 2 N0 ∞ 2
E[Yi 2 (t )] = RYiYi (0) =
2π ∫ −∞
S YiYi ( w)dw =
2π ∫ −∞
H i ( w) S XX ( w)dw =
4π ∫ −∞
H i ( w) dw

Thus, for E[Y12 (t )] we obtain

2
N ∞ N ∞ 2 N0 ∞ 2  sin( w / 2) 
∫−∞ w 2 {1 − cos(w)}dw = 4π
2
E[Y (t )] = 0 ∫ H 1 ( w) dw = 0 ∫−∞ w 2  w / 2  dw
2

4π 4π
1
−∞

2
2N0 ∞  sin( w / 2) 
=
π ∫ 0  w / 2  dw

where the equality follows from the fact that the integrand is an even function. Let
u = w / 2 ⇒ dw = 2du. Thus, we obtain

4
 4N0  π 
2 2
2N0 ∞  sin( w / 2)  4N 0 ∞  sin(u ) 
E[Y (t )] = ∫  w / 2  dw = π ∫  u  du = π  2  = 2 N 0
2

π
1
0 0

Similarly, for E[Y22 (t )] we obtain

N0 ∞ 2 N0 ∞ 4 N ∞ 1
E[Y22 (t )] =
4π ∫ −∞
H 2 ( w) dw =
4π ∫ −∞ 1 + w 2
dw = 0
π ∫ −∞ 1 + w 2
dw

Let w = tan(θ ) ⇒ dw = sec 2 (θ )dθ . Now, when w = −∞, θ = −π / 2. Similarly, when

w = ∞, θ = π / 2. Thus, we obtain

π
π π
N0 ∞ 1 N sec 2 (θ ) N0 2 dθ = N  
θ 2
E[Y (t )] = ∫ dw = 0 ∫− 2 1 + tan 2 (θ )dθ = π ∫− 2  π  π = N 0
2 2
π π
π 1+ w π
2 2 0
− ∞
−
2
(b) Since X (t ) is a noise function, R XX (τ ) = ( N 0 / 2)δ (τ ) and the crosscorrelation function
RY1Y2 (t , t + τ ) is given by
∞ ∞
RY1Y2 (t , t + τ ) = E[Y1 (t )Y2 (t + τ )] = E  ∫ h1 (u ) X (t − u )du ∫ h2 (v) X (t + τ − v)dv 
 −∞ −∞ 
h1 (u )h2 (v)E [ X (t − u ) X (t + τ − v)]dvdu
1 ∞
=∫ ∫
0 0

1 ∞ 1 ∞ N 
=∫ ∫ 2e −v R (τ + u − v)dvdu = ∫ ∫ 2e −v  0 δ (τ + u − v)dvdu
0 0 0 0
 2 
1 1
= N 0 ∫ e −( u +τ ) du = N 0 e −τ ∫ e −u du = N 0 e −τ − e −u
0 0
[ ] 1
0 [
= N 0 e −τ 1 − e −1 ]
= 0.632 N 0 e −τ

11.7 A wide-sense stationary process X(t) is the input to a linear system whose impulse
response is h(t ) = 2e −7t , t ≥ 0. If the autocorrelation function of the process is
R XX (τ ) = e −4|τ | and the output process is Y(t), find the following:
(a) The power spectral density of Y(t)
(b) The cross-spectral power density S XY (w)
(c) The crosscorrelation function R XY (τ )
Solution: The power spectral density of X (t ) and the system response of the system are given
by
8
S XX ( w) = 2
w + 16
2
H ( w) =
7 + jw
(a) The power spectral density of Y(t) is given by
5
 2  2  2  8 
S YY ( w) = H ( w) S XX ( w) = H ( w) H ∗ ( w) S XX ( w) =    2 
 7 + jw  7 − jw  w + 16 
32
= 2
( )(
w + 49 w 2 + 16 )
(b) The cross-spectral power density S XY (w) is given by
 2  8  16
S XY ( w) = H ( w) S XX ( w) =  =
 2
(
 7 + jw  w + 16  (7 + jw) w + 16
2
)
(c) The crosscorrelation function R XY (τ ) can be found as follows:
16 16 a b c
S XY ( w) = = ≡ + +
( )
(7 + jw) w + 16 (7 + jw)(4 + jw)(4 − jw) 7 + jw 4 + jw 4 − jw
2

16 16
a = (7 + jw )S XY ( w) jw=−7 = =−
(−3)(11) 33
16 2
b = (4 + jw)S XY ( w) jw= −4 = =
(3)(8) 3
16 2
c = (4 − jw )S XY ( w) jw=4 = =
(11)(8) 11
Thus, we have that
 2 − 4τ 16 −7τ
 3 e − 33 e τ ≥0
2/3 16 / 33 2 / 11
S XY ( w) = − + ⇒ R XY (τ ) = 
4 + jw 7 + jw 4 − jw  2 e 4τ
11
τ <0

11.8 A linear system has a transfer function given by

w
H ( w) = 2
w + 15w + 50

Determine the power spectral density of the output when the input function is

(a) a stationary random process X(t) with an autocorrelation function R XX (τ ) = 10e −|τ |
(b) white noise that has a mean-square value of 1.2 V 2 / Hz
Solution:
(a) The power spectral density of the output when the input function is a stationary random
process X(t) with an autocorrelation function R XX (τ ) = 10e −|τ | can be obtained as follows:
20
S XX ( w) =
1 + w2
2
2  w   20  20w 2
S YY ( w) = H ( w) S XX ( w) =  2   2 
=
( )(
 w + 15w + 50  1 + w  1 + w 2 w 2 + 15w + 50 )
2

6
 (b) The power spectral density of the output when the input function is white noise that has a
mean-square value of 1.2 V 2 / Hz can be obtained as follows:
R XX (τ ) = 1.2δ (τ ) ⇒ S XX ( w) = 1.2
2
  1.2 w 2
2
S YY ( w) = H ( w) S XX ( w) =  2
w
 {1 .2} =
 w + 15w + 50  (
w 2 + 15w + 50 )
2

11.9 A linear system has the impulse response h(t ) = e − at , where t ≥ 0 and a > 0. Find the
power transfer function of the system.
2
Solution: The power transfer function of the system, H ( w) , can be obtained as follows:
1 2  1  1  1
H ( w) = ⇒ H ( w) = H ( w) H ∗ ( w) =   = 2
a + jw  a + jw  a − jw  a + w
2

11.10 Consider the system with the impulse response h(t ) = e − at , where t ≥ 0 and a > 0.
Assume that the input is white noise with power spectral density N 0 / 2. What is the
power spectral density of the output process?
Solution: The power spectral density of the output process can be obtained as follows:

N0
S XX ( w) =
2
1
H ( w) =
a + jw
2 N0
S YY ( w) = H ( w) S XX ( w) =
(
2 a + w2
2
)
11.11 The power transfer function of a system is given by
2 64
H ( w) =
[16 + w 2 ] 2
Use Table 10.1 to obtain the impulse function h(t) of the system.
Solution: From Table 10.1 we have that
8 2(4)
H ( w) = = 2 ↔ e − 4|t|
16 + w 2
4 +w 2

Therefore, the impulse function h(t) of the system is given by

h(t ) = e −4|t|

11.12 A wide-sense stationary process X(t) has the autocorrelation function given by
R XX (τ ) = cos( w0τ )
The process is input to a system with the power transfer function

7
 2 64
H ( w) =
[16 + w 2 ] 2
(a) Find the power spectral density of the output process.
(b) If Y(t) is the output process, find the cross-power spectral density S XY (w).
Solution: The power spectral density of the input process is
S XX ( w) = π {δ ( w − w0 ) + δ ( w + w0 )}
(a) The power spectral density of the output process is given by
2 64π 64π 128π
S YY ( w) = H ( w) S XX ( w) = + =
(
2 2
16 + w0 ) (
2 2
16 + w0 )
16 + w02 ( )
2

(b) Given that Y(t) is the output process, the cross-power spectral density S XY (w) is given by
8π 8π 16π
S XY ( w) = H ( w) S XX ( w) = + =
16 + w0 16 + w0 16 + w02
2 2

11.13 A causal system is used to generate an output process Y(t) with the power spectral density
2a
S YY ( w) = 2
a + w2
Find the impulse response h(t) of the system.
Solution: Since
2a 2 2 2a
S YY ( w) = 2 = H ( w) S XX ( w) = 1 2 ⇒ H ( w) = 1
a +w 2
a + w2
we conclude that the impulse response h(t ) of the system is h(t ) = δ (t ).

11.14 X(t) is a wide-sense stationary process. It is the input to a linear system with impulse
response h(t), and Y(t) is the output process. Consider another process Z(t) that is
obtained as follows: Z (t ) = X (t ) − Y (t ). The scheme is illustrated in Figure 11.8.

X (t ) h(t ) Y (t ) −+ Z (t )
+

Figure 11.8: Figure for Problem 11.14

Determine the following in terms of the parameters of X(t):

(a) The autocorrelation function RZZ (τ )
(b) The power spectral density S ZZ (w)
8
 (c) The crosscorrelation function R XZ (τ )
(d) The cross-power spectral density S XZ (w)
Solution:
(a) The autocorrelation function RZZ (τ ) is given by
RZZ (t , t + τ ) = E[ Z (t ) Z (t + τ )] = E [{X (t ) − Y (t )}{X (t + τ ) − Y (t + τ )}]
= E[ X (t ) X (t + τ )] − E[ X (t )Y (t + τ )] − E[Y (t ) X (t + τ )] + E[Y (t )Y (t + τ )]
= R XX (τ ) − R XY (τ ) − RYX (τ ) + RYY (τ )
= R XX (τ ) − R XX (τ ) ∗ h(τ ) − R XX (τ ) ∗ h(−τ ) + R XX (τ ) ∗ h(−τ ) ∗ h(τ )

(b) The power spectral density S ZZ (w) is given by

2
S ZZ ( w) = S XX ( w) − H ( w) S XX ( w) − H ∗ ( w) S XX ( w) + H ( w) S XX ( w)
{ 2
= 1 − H ( w) − H ∗ ( w) + H ( w) S XX ( w) }
(c) The crosscorrelation function R XZ (τ ) is given by
R XZ (t , t + τ ) = E[ X (t ) Z (t + τ )] = E [X (t ){X (t + τ ) − Y (t + τ )}]
= E[ X (t ) X (t + τ )] − E[ X (t )Y (t + τ )] = R XX (τ ) − R XY (τ )
= R XX (τ ) − R XX (τ ) ∗ h(τ )

(d) The cross-power spectral density S XZ (w) is given by

S XZ ( w) = S XX ( w) − H ( w) S XX ( w) = {1 − H ( w)}S XX ( w)

11.15 Consider the system shown in Figure 11.9 in which an output process Y(t) is the sum of
an input process X(t) and a delayed version of X(t) that is scaled (or multiplied) by a
factor a.
+
X (t ) + Y (t )
+

Delay, T

Figure 11.9: Figure for Problem 9.15

Determine the following:

(a) The equation that governs the system (i.e., the equation that relates Y(t) to X(t))
(b) The crosscorrelation function R XY (τ )
9
 (c) The cross-power spectral density S XY (w)
(d) The transfer function H(w) of the system
(e) The power spectral density of Y(t)
Solution:
(a) The equation that governs the system is given by Y (t ) = X (t ) + aX (t − T ).
(b) The crosscorrelation function R XY (τ ) is given by
R XY (τ ) = E[ X (t )Y (t + τ )] = E [ X (t ){X (t + τ ) + aX (t + τ − T )}]
= E[ X (t ) X (t + τ )] + aE[ X (t ) X (t + τ − T )]
= R XX (τ ) + aR XX (τ − T )

(c) The cross-power spectral density S XY (w) is given by

S XY ( w) = S XX ( w) + aS XX ( w)e − jwT
{ }
= 1 + ae − jwT S XX ( w)

(d) The transfer function H(w) of the system is given by

S ( w)
H ( w) = XY = 1 + ae − jwT
S XX ( w)

(e) The power spectral density of Y(t) is given by

2
{ }{ }
S YY ( w) = H ( w) S XX ( w) = H ( w) H ∗ ( w) S XX ( w) = 1 + ae − jwT 1 + ae jwT S XX ( w)
  e jwT + e − jwT 
{ }
= 1 + ae jwT + ae − jwT + a 2 S XX ( w) = 1 + 2a 
2

 + a S XX ( w)
2

   
{ }
= 1 + 2a cos( wT ) + a 2 S XX ( w)

11.16 X(t) and Y(t) are two jointly wide-sense stationary processes. If Z (t ) = X (t ) + Y (t ) is the
input to a linear system with impulse response h(t), determine the following:
(a) The autocorrelation function of Z(t)
(b) The power spectral density of Z(t)
(c) The cross-power spectral density S ZV (w) of the input process Z(t) and the output pro-
cess V(t)
(d) The power spectral density of the output process V(t).
Solution:
(a) The autocorrelation function of Z(t) is given by
RZZ (t , t + τ ) = E[ Z (t ) Z (t + τ )] = E[{X (t ) + Y (t )}{X (t + τ ) + Y (t + τ )}]
= E[ X (t ) X (t + τ )] + E[ X (t )Y (t + τ )] + E[Y (t ) X (t + τ )] + E[Y (t )Y (t + τ )]
= R XX (τ ) + R XY (τ ) + RYX (τ ) + RYY (τ ) = RZZ (τ )

(b) The power spectral density of Z(t) is given by

10
 ∗
S ZZ ( w) = S XX ( w) + S XY ( w) + S YX ( w) + S YY ( w) = S XX ( w) + S XY ( w) + S XY ( w) + S YY ( w)
(c) The cross-power spectral density S ZV (w) of the input process Z(t) and the output process
V(t) is given by
S ZV ( w) = H ( w) S ZZ ( w) = H ( w){S XX ( w) + S XY ( w) + S YX ( w) + S YY ( w)}
(d) The power spectral density of the output process V(t) is given by

SVV ( w) = H ( w) S ZZ ( w) = H ( w) {S XX ( w) + S XY ( w) + S YX ( w) + S YY ( w)}
2 2

11.17 X(t) is a wide-sense stationary process. Assume that Z (t ) = X (t − d ), where d is a

constant delay. If Z(t) is the input to a linear system with impulse response h(t),
determine the following:
(a) The autocorrelation function of Z(t)
(b) The power spectral density S ZZ (w)
(c) The crosscorrelation function RZX (τ )
(d) The cross-power spectral density S ZX (w)
(e) The power spectral density S YY (w) of the output process Y(t), which is obtained by
passing Z(t) through a linear system with the system response H(w)
Solution: The following figure shows the configuration of the system.

(a) The autocorrelation function of Z(t) is given by

RZZ (τ ) = E[ Z (t ) Z (t + τ )] = E[ X (t − d ){X (t + τ − d )}] = R XX (τ )

(b) The power spectral density S ZZ (w) is S ZZ ( w) = S XX ( w).

(c) The crosscorrelation function RZX (τ ) is given by
RZX (τ ) = E[ Z (t ) X (t + τ )] = E [ X (t − d ){X (t + τ )}] = R XX (τ + d )

(d) The cross-power spectral density S ZX (w) is S ZX ( w) = S XX ( w)e jwd .

(e) The power spectral density S YY (w) of the output process Y(t) is given by
2 2
S YY ( w) = H ( w) S ZZ ( w) = H ( w) S XX ( w)

11.18 X(t) is a wide-sense stationary process that is the input to a linear system with the transfer
function
1
H ( w) =
a + jw
11
 where a > 0. If X(t) is a zero-mean white noise with power spectral density N 0 / 2, deter-
mine the following:
(a) The impulse response h(t) of the system
(b) The cross-power spectral density S XY (w) of the input process and the output process
Y(t)
(c) The crosscorrelation function R XY (τ ) of X(t) and Y(t)
(d) The crosscorrelation function RYX (τ ) of Y(t) and X(t)
(e) The cross-power spectral density S YX (w) of Y(t) and X(t)
(f) The power spectral density S YY (w) of the output process
Solution: The autocorrelation function of the white noise is
N N
R XX (τ ) = 0 δ (τ ) ⇒ S XX ( w) = 0
2 2
− at
(a) Since from Table 10.1 e ↔ a + jw for a > 0, t ≥ 0, the impulse response of the system
1

is given by
h(t ) = e − at a > 0, t ≥ 0
(b) The cross-power spectral density S XY (w) of the input process and the output process Y(t)
is given by

N0
S XY ( w) = H ( w) S XX ( w) =
2(a + jw)

(c) The crosscorrelation function R XY (τ ) is the inverse Fourier transform of S XY (w) and is
given by
N
R XY (τ ) = 0 e −aτ τ ≥0
2

(d) The crosscorrelation function RYX (τ ) of Y(t) and X(t) is given by

N
RYX (τ ) = R XY (−τ ) = 0 e aτ τ <0
2

(e) The cross-power spectral density S YX (w) of Y(t) and X(t) is given by
N0
S YX ( w) = H ∗ ( w) S XX ( w) =
2(a − jw)

(f) The power spectral density S YY (w) of the output process is given by
2 N0
S YY ( w) = H ( w) S XX ( w) =
2( a + w 2 )
2

12
Section 11.4: Linear Systems with Discrete Random Input
11.19 A linear system has an impulse response given by
e − an n≥0
h[n] = 
0 n<0
where a > 0 is a constant. Find the transfer function of the system.
Solution: The transfer function of the system is given by
∞ ∞ ∞
1
H ( Ω) = ∑ h[n]e − jΩn = ∑ e −an e − jΩn = ∑ e −(a + jΩ) n =
n = −∞ n=0 n =0 1− e − ( a + jΩ )

11.20 A linear system has an impulse response given by

e − an n≥0
h[n] = 
0 n<0
where a > 0 is a constant. Assume that the autocorrelation function of the input sequence
to this system is defined by
R XX [n] = b n 0 < b < 1, n ≥ 0
Find the power spectral density of the output process.
Solution: The power spectral density of the output process can be obtained as follows:
1
H (Ω) = − ( a + jΩ )
1− e
[ ]
∞ ∞
1
S XX (Ω) = ∑ b n e − jΩn = ∑ be − jΩ =
n

n=0 n=0 1 − be − jΩ
2
S YY (Ω) = H (Ω) S XX (Ω) = H (Ω) H ∗ (Ω) S XX (Ω)
 1  1  1 
= − ( a + jΩ )  − ( a − jΩ )   − jΩ 
1 − e 1 − e 1 − be 
1  1 
=  − jΩ 
1 − 2e cos(Ω) + e −2 a
−a
1 − be 
11.21 The autocorrelation function of a discrete-time random sequence X[n] is given by
R XX [m] = e −b|m|
where b > 0 is a constant. Find the power spectral density of the sequence.
Solution: The power spectral density of the sequence can be obtained as follows. First, we have
that
e −bm m≥0
R XX [m] =  bm
e m<0
Thus,

13
 ∞ 1 ∞ ∞ ∞
S XX (Ω) = ∑ R XX [m]e − jΩm =
m = −∞
∑ e bm e − jΩm + ∑ e −bm e − jΩm = ∑ e −bm e jΩm + ∑ e −bm e − jΩm
m = −∞ m =0 m =1 m =0
jΩ m − jΩm
+e
{ } e 
∞ ∞
= 1 + ∑ e −bm e jΩm + e − jΩm = 1 + 2∑ e −bm  
m =1 m =1  2 
∞
= 1 + 2∑ e −bm cos(mΩ)
m =1

11.22 A linear system has an impulse response given by

e − an n≥0
h[n] = 
0 n<0
where a > 0 is a constant. Assume that the autocorrelation function of the input discrete-
time random sequence X[n] is given by
R XX [m] = e −b|m|
where b > 0 is a constant. Find the power spectral density of the output process.
Solution: From earlier results the power spectral density of the output process can be obtained as
follows:
1
H (Ω) = − ( a + jΩ )
1− e
∞
S XX (Ω) = 1 + 2∑ e −bm cos(mΩ)
m =1
2
SYY (Ω) = H (Ω) S XX (Ω) = H (Ω) H ∗ (Ω) S XX (Ω)
∞
 1  1  
=
1 − e
−( a + jΩ ) 
1 − e
−( a − jΩ ) 

1 + 2 ∑
m =1
e −bm cos(mΩ)

∞
1  
=
1 − 2e cos(Ω) + e −2 a
−a 

1 + 2 ∑
m =1
e −bm cos(mΩ)


11.23 A wide-sense stationary continuous-time process has the autocorrelation function given
by
R X C X C (τ ) = e −4|τ |
If X C (t ) is sampled with a sampling period 10 seconds to produce the discrete-time pro-
cess X [n], find the power spectral density of X [n].
Solution: The power spectral density of X C (t ) is given by
8
S X C X C ( w) =
16 + w 2
If X C (t ) is sampled with a sampling period 10 seconds to produce the discrete-time process
X [n], the power spectral density of X [n] is given by

14
 1 ∞  Ω − 2πk  1 ∞ 8
S XX (Ω) = ∑ XX
S = ∑
 Ω − 2πk 
2
T k =1 C C  T  10 k =1
16 + 
 10 
11.24 A wide-sense stationary continuous-time process has the autocorrelation function given
by
R X C X C (τ ) = e −4|τ |
X C (t ) is sampled with a sampling period 10 seconds to produce the discrete-time
sequence . The sequence is then input to a system with the impulse response
e − an n≥0
h[n] = 
0 n<0
where a > 0 is a constant. Find the power spectral density of the output process.
Solution: From earlier results we have that
8
S X C X C ( w) =
16 + w 2
1 ∞ 8
S XX (Ω) = ∑
 Ω − 2πk 
2
10 k =1
16 + 
 10 
1
H (Ω ) = − ( a + jΩ )
1− e
2
SYY (Ω) = H (Ω) S XX (Ω) = H (Ω) H ∗ (Ω) S XX (Ω)
1  1 ∞ 8
=  ∑
10 1 − 2e −a cos(Ω) + e −2 a  k =1  Ω − 2πk 
2

16 + 
 10 

11.25 Consider the system shown in Figure 11.10 in which an output sequence Y[n] is the sum
of an input sequence X[n] and a version of X[n] that has been delayed by one unit and
scaled (or multiplied) by a factor a.
+
X [n] + Y [n]
+

Unit Delay

Figure 11.10: Figure for Problem 11.25

15
 Determine the following:
(a) The equation that governs the system
(b) The crosscorrelation function R XY [m]
(c) The cross-power spectral density S XY (Ω)
(d) The transfer function H (Ω) of the system
Solution:
(a) The equation that governs the system is Y [n] = X [n] + aX [n − 1].
(b) The crosscorrelation function R XY [m] is given by

R XY [m] = E[ X [n]Y [n + m]] = E [X [n]{X [n + m] + aX [n + m − 1]}]

= E[ X [n] X [n + m]] + aE[ X [n] X [n + m − 1]]
= RXX [m] + aR XX [m − 1]

(c) From the result in part (b), the cross-power spectral density S XY (Ω) is given by

S XY (Ω) = S XX (Ω) + aS XX (Ω)e − jΩ = [1 + ae − jΩ ]S XX (Ω)

(d) The transfer function H (Ω) of the system is given by

S XY (Ω)
H (Ω) = = 1 + ae − jΩ
S XX (Ω)

Section 11.5: Autoregressive Moving Average Processes

11.26 A discrete-time feedback control system has the property that its output voltage Y[n] at
time n is a linear combination of the output voltage at time n − 1 scaled by a factor a and
a random error W[n] at time n that is independent of past outputs, as shown in Figure
11.11, where | a | < 1. The random process W[n] is a sequence of independent and
identically distributed random variables with zero mean and standard deviation β .
Assume also that the random process Y[n] has zero mean and W [n] = 0 for n < 0.
(a) Determine the equation that governs the system
(b) Is the output process Y[n] a wide-sense stationary?
(c) If Y[n] is a wide-sense stationary process, find the power transfer function
(d) Find the crosscorrelation function RWY [n, n + m]
(e) Find the autocorrelation function RWW [n, n + m]

16
 +
W [n] + Y [n]
+

Unit Delay

Figure 11.11: Figure for Problem 11.26

Solution:
(a) This is an example of a first-order autoregressive process, and the equation that governs
the system is Y [n] = W [n] + aY [n − 1].
(b) The general structure of the output process Y[n] can be obtained as follows:
Y [ 0] = W [ 0]
Y [1] = aY [0] + W [1] = aW [0] + W [1]
Y [2] = aY [1] + W [2] = a{aW [0] + W [1]} + W [2] = a 2W [0] + aW [1] + W [2]
In general,
n
Y [ n ] = ∑ a kW [ n − k ]
k =0

Thus, the autocorrelation function of Y[n] is given by

n n 
RYY [n, n + m] = E[Y [n]Y [n + m]] = E ∑ a kW [n − k ]∑ a jW [n + m − j ]
 k =0 j =0 
n n
= ∑∑ a k a j E[W [n − k ]W [n + m − j ]]
k = 0 j =0

Since the W[n] are independent and identically distributed with E[W [n]] = 0 and
E[W 2 [n]] = β 2 , we have that E[W [n − k ]W [n + m − j ]] = 0 except when
n − k = n + m − j; that is, when j = m + k . Thus, the autocorrelation function becomes

n n
β 2 a m {1 − a 2( n+1) }
RYY [n, n + m] = β 2
∑a a
k =0
k m+ k
=β a2 m
∑a
k =0
2k
=
1 − a2

Since RYY [n, n + m] is not independent of n, Y[n] is not a wide-sense stationary process.

(c) Since we have established that Y[n] is not a wide-sense stationary process, we are not
required to obtain the power transfer function.
(d) The crosscorrelation function RWY [n, n + m] is given by

17
 RWY [n, n + m] = E[W [n]Y [n + m]] = E [W [n]{W [n + m] + aY [n + m − 1]}]
= E[W [n]W [n + m]] + aE[W [n]Y [n + m − 1]]
= RWW [n, n + m] + aRWY [n, n + m − 1]

(e) The autocorrelation function RWW [n, n + m] of the input process is given by

RWW [n, n + m] = β 2δ [m] = RWW [m]

11.27 Find the mean, autocorrelation function and variance of the MA(2) process, assuming
that n > 2 for the output process Y[n].
Solution: If we assume that W[n] is a zero-mean process with variance σ W2 , we have that
Y [n] = β 0W [n] + β1W [n − 1] + β 2W [n − 2]
E[Y [n]] = β 0 E[W [n]] + β1 E[W [n − 1]] + β 2 E[W [n − 2]] = 0
[
σ Y2 = E[Y 2 [n]] = E {β 0W [n] + β1W [n − 1] + β 2W [n − 2]}2 ]
{
= σ W2 β 02 + β12 + β 22 }
RYY [n, n + m] = E[Y [n]Y [n + m]]
= E [{β 0W [n] + β1W [n − 1] + β 2W [n − 2]}{β 0W [n + m] + β1W [n + m − 1] + β 2W [n + m − 2]}]
= β 02 r00 + β 0 β1r01 + β 0 β 2 r02 + β 0 β1r10 + β12 r11 + β1 β 2 r12 + β 0 β 2 r20 + β1 β 2 r21 + β 22 r22
= β 02 r00 + β12 r11 + β 22 r22 + β 0 β1 {r01 + r10 } + β 0 β 2 {r02 + r20 } + β1 β 2 {r12 + r21 }

where
r00 = E[W [n]W [n + m]] = RWW [n, n + m]
r01 = E[W [n]W [n + m − 1]] = RWW [n, n + m − 1]
r02 = E[W [n]W [n + m − 2]] = RWW [n, n + m − 2]
r10 = E[W [n − 1]W [n + m]] = RWW [n − 1, n + m]
r11 = E[W [n − 1]W [n + m − 1]] = RWW [n − 1, n + m − 1]
r12 = E[W [n − 1]W [n + m − 2]] = RWW [n − 1, n + m − 2]
r20 = E[W [n − 2]W [n + m]] = RWW [n − 2, n + m]
r21 = E[W [n − 2]W [n + m − 1]] = RWW [n − 2, n + m − 1]
r22 = E[W [n − 2]W [n + m − 2]] = RWW [n − 2, n + m − 2]

Thus, we obtain
(
 β 02 + β12 + β 22 σ W2 ) m=0

(β β + β1 β 2 )σ W m = ±1
2
RWW [n, n + m] =  0 1
β 0 β 2σ W m = ±2
2

0 otherwise


18
11.28 Find the autocorrelation function of the following MA(2) process:
Y [n] = W [n] + 0.7W [n − 1] − 0.2W [n − 2]
Solution: We use the result of Problem 11.27 with β 0 = 1, β1 = 0.7 and β 2 = −0.2 to obtain the
following result:
1.53σ W2 m=0

0.66σ W m = ±1
2
RWW [n, n + m] = 
− 0.2σ W m = ±2
2

0 otherwise


11.29 Find the autocorrelation function of the following AR(2) process:

Y [n] = 0.7Y [n − 1] − 0.2Y [n − 2] + W [n]
Solution: The autocorrelation function of the process can be obtained as follows:

RYY [ n, n + m] = E[Y [n]Y [ n + m]]

= E [{0.7Y [n − 1] − 0.2Y [ n − 2] + W [n]}{0.7Y [n + m − 1] − 0.2Y [ n + m − 2] + W [n + m]}]
= 0.49 RYY [n − 1, n + m − 1] + 0.04 RYY [n − 2, n + m − 2] + RWW [ m] + A + B + C + D + F + G

where
A = −(0.7)(0.2) E[Y [n − 1][Y [n + m − 2]] = −0.14 RYY [n − 1, n + m − 2]
B = 0.7 E[Y [n − 1][W [n + m]] = 0.7 RYW [n − 1, n + m]
C = −(0.2)(0.7) E[Y [n − 2][Y [n + m − 1]] = −0.14 RYY [n − 2, n + m − 1]
D = −0.2 E[Y [n − 2][W [n + m]] = −0.2 RYW [n − 2, n + m]
F = 0.7 E[W [n][Y [n + m − 1]] = 0.7 RWY [n, n + m − 1]
G = −0.2 E[W [n][Y [n + m − 2]] = −0.2 RWY [n, n + m − 2]

11.30 Consider the following ARMA(1,1) process, where | α | < 1, | β | < 1, and Y [n] = 0 for
n < 0:
Y [n] = αY [n − 1] + W [n] + βW [n − 1]

Assume that W[n] is a zero-mean random process with W [n] = 0 when n < 0, and the
variance is E[W [n]W [k ]] = σ W2 δ [n − k ].
(a) Carefully find a general expression for the Y[n] in terms of only W[n] and its delayed
versions.
(b) Using the above results, find the autocorrelation function of the ARMA(1, 1) process.
Solution:
(a) A general expression for the Y[n] is given by

19
Y [n] = αY [n − 1] + W [n] + βW [n − 1]
= α {αY [n − 2] + W [n − 1] + βW [n − 2]} + W [n] + βW [n − 1]
= α 2 {αY [n − 3] + W [n − 2] + βW [n − 3]} + αβ W [n − 2] + (α + β )W [n − 1] + W [n]
= α 3Y [n − 3] + α 2 βW [n − 3] + α (α + β )W [n − 2] + (α + β )W [n − 1] + W [n]
= α 3 {αY [n − 4] + W [n − 3] + βW [n − 4]} + α 2 βW [n − 3] + α (α + β )W [n − 2] + (α + β )W [n − 1] + W [n]
= W [n] + (α + β )W [n − 1] + α (α + β )W [n − 2] + α 2 (α + β )W [n − 3] + α 3 βW [n − 4] + α 4Y [n − 4]
Μ
n
= W [n] + (α + β )∑ α k −1W [n − k ]
k =1

(b) Using the above results, the autocorrelation function of the ARMA(1, 1) process is given
by
RYY [n, n + m] = E[Y [n]Y [n + m]]
 n
 n+m 
= E W [n] + (α + β )∑ α k −1W [n − k ]W [n + m] + (α + β ) ∑ α j −1W [n + m − j ]
 k =1  j =1 
= S11 + S12 + S 21 + S 22
where
S11 = E[W [n]W [n + m]] = RWW [n, n + m]
n+ m n+m
S12 = (α + β ) ∑ α j −1 E[W [n]W [n + m − j ]] = (α + β ) ∑ α j −1 RWW [n, n + m − j ]
j =1 j −1
n+ m n+ m
S 21 = (α + β ) ∑ α k −1 E[W [n − k ]W [n + m]] = (α + β ) ∑ α k −1 RWW [n − k , n + m]
k =1 k =1
n n+ m
S 22 = (α + β ) 2 ∑∑ α k −1α j −1 E[W [n − k ]W [n + m − j ]]
k =1 j =1
n n+ m
= (α + β ) 2 ∑∑ α k −1α j −1 RWW [n − k , n + m − j ]
k =1 j =1

Since E[W [n]W [k ]] = RWW [n, k ] = σ W2 δ [n − k ], we have that

20
 S11 = RWW [n, n + m] = σ W2 δ [m]
n+ m
S12 = (α + β ) ∑ α j −1 RWW [n, n + m − j ] = σ W2 (α + β )α m−1
j −1
n+ m
S 21 = (α + β ) ∑ α k −1 RWW [n − k , n + m] = σ W2 (α + β )α −m −1
k =1
n n+ m n
S 22 = (α + β ) 2 ∑∑ α k −1α j −1 RWW [n − k , n + m − j ] = σ W2 (α + β ) 2 ∑ α k −1α m + k −1
k =1 j =1 k =1

 σ W (α + β ) α
2 2 m
n
 1
= σ W2 (α + β ) 2 α m −2 ∑ α 2 k = σ W2 (α + β ) 2 α m −2  − 1 =
1 − α 1−α 2
2
k =1 
Thus, the autocorrelation function is given by

α + β  (α + β ) α 

( )
2 m
RYY [n, n + m] = σ W2 δ [m] + α m + α −m   + 
  α  1−α 2 

11.31 Write out the expression for the MA(5) process.

Solution: The expression for the MA(5) process is

Y [n] = β 0W [n] + β1W [n − 1] + β 2W [n − 2] + β 3W [n − 3] + β 4W [n − 4] + β 5W [n − 5]

11.32 Write out the expression for the AR(5) process.

Solution: The expression for the AR(5) process is

Y [n] = α1Y [n − 1] + α 2Y [n − 2] + α 3Y [n − 3] + α 4Y [n − 4] + α 5Y [n − 5] + β 0W [n]

11.33 Write out the expression for the ARMA(4, 3) process.

Solution: The expression for the ARMA(4, 3) process is

Y [n] = α 1Y [n − 1] + α 2Y [n − 2] + α 3Y [n − 3] + α 4Y [n − 4] + β 0W [n] + β1W [n − 1]

+ β 2W [n − 2] + β 3W [n − 3]

21
Chapter 12: Special Random Processes

Chapter Summary: This chapter discusses some special random processes that are used to
model many physical systems. These include the Bernoulli process, Gaussian process, random
walk, Poisson process and Markov process.
Section 12.2: Bernoulli Process
12.1 A random process Y[n] is defined by Y [n] = 3 X [n] + 1, where X[n] is a Bernoulli process
with a success probability p. Find the mean and variance of Y[n].
Solution: The mean and variance of X [n] are given by
E[ X [n]] = p
σ X2 [ n ] = p (1 − p )

Therefore, the mean and variance of Y [n] are given by

E[Y [n]] = E[3 X [n] + 1] = 3E[ X [n]] + 1 = 3 p + 1

σ Y2[ n ] = 3 2 σ X2 [ n ] = 9 p (1 − p)

12.2 A sequence of Bernoulli trials consists of choosing seven components at random from a
batch of components. A selected component is classified as either defective or non-
defective. A non-defective component is considered a success, while a defective
component is considered a failure. If the probability that a selected component is non-
defective is 0.8, what is the probability of three successes?
Solution: Let the random variable K (7) denote the number of nondefective components among
the 7 components. Then the PMF of K (7) has the binomial distribution with p = 0.8, as
follows:

7
p K ( 7 ) (k ) =   (0.8) (0.2) 7 − k k = 0, 1, Κ , 7
k

k 
Thus, the probability of selecting 3 nondefective components is

7 7!
p K ( 7 ) (3) =   (0.8) (0.2) 4 = (0.8)3 (0.2) 4 = 0.0287
3

3
  3! 4!

12.3 The probability that a patient recovers from a rare blood disease is 0.3. If 15 people are
known to have contracted this disease, find the following probabilities:
(a) At least 10 survive.
(b) From 3 to 8 survive.
(c) Exactly 6 survive.
Solution: Let the random variable N (15) denote the number of survivors of the disease. Then
the PMF of N (15) has the binomial distribution with p = 0.3, as follows:

1
 15 
p N (15) (n) =   (0.3) (0.7)15−n n = 0, 1, Κ , 15
n

 
n

(a) The probability that at least 10 survive is given by

15
15  15  15 
P[ N (15) ≥ 10] = ∑p (n) =   (0.3) (0.7) 5 +   (0.3) (0.7) 4 +   (0.3) (0.7) 3
10 11 12
N (15)
n =10 10  11  12 
15  15 
+   (0.3) (0.7) 2 +   (0.3) (0.7) + (0.3)
13 14 15

13  14 
= 0.00365
(b) The probability that the number of survivors is at least 3 and at most 8 is given by
8
15  15  15 
P[3 ≤ N (15) ≤ 8] = ∑ p N (15) (n) =   (0.3) (0.7)12 +   (0.3) (0.7)11 +   (0.3) (0.7)10
3 4 5

n =3 3 4 5

15  15  15 
+   (0.3) (0.7) 9 +   (0.3) (0.7) 8 +   (0.3) (0.7) 7
6 7 8

6 7 8

= 0.8579
(c) The probability that exactly 6 survive is given by
15 
P[ N (15) = 6] =   (0.3) (0.7) 9 = 0.1472
6

6

12.4 A sequence of Bernoulli trials consists of choosing components at random from a batch
of components. A selected component is classified as either defective or non-defective. A
non-defective component is considered a success, while a defective component is
considered a failure. If the probability that a selected component is non-defective is 0.8,
determine the probabilities of the following events:
(a) The first success occurs on the fifth trial.
(b) The third success occurs on the eighth trial.
(c) There are 2 successes by the fourth trial, there are 4 successes by the tenth trial, and
there are 10 successes by the eighteenth trial.
Solution: Let X k be a random variable that denotes the number of trials up to and including the
trial that results in the kth success. Then X k is a kth-order Pascal random variable whose PMF is
given by
 n − 1 k
p X k (n) =   p (1 − p ) n − k k = 1, 2, Κ ; n = k , k + 1, Κ
 k − 1
where p = 0.8.
(a) The probability that the first success occurs on the fifth trial is given by

2
  5 − 1 1 4
P[ X 1 = 5] = p X 1 (5) =   p (1 − p ) 4 =   p (1 − p) 4 = p (1 − p ) 4 = 0.8(0.2) 4
1 − 1  0
= 0.00128

(b) The probability that the third success occurs on the eighth trial is given by
 8 − 1 3 7
P[ X 3 = 8] = p X 3 (8) =   p (1 − p ) 5 =   p 3 (1 − p) 5 = 21 p 3 (1 − p) 5 = 21(0.8) 3 (0.2) 5
 3 − 1  2
= 0.00344
(c) The probability that there are 2 successes by the fourth trial, there are 4 successes by the
tenth trial, and there are 10 successes by the eighteenth trial can be obtained by
partitioning the timeline as follows:
1. There are 2 successes in the first 4 trials
2. There are 2 successes in the next 6 trials
3. There are 6 successes in the next 8 trials
These intervals are illustrated in the following diagram.
{
{
{
Since these intervals are non-overlapping, the events occurring within each interval are
independent of those that occur in other intervals, and the events have a binomial
distribution. Thus, the probability, Q, of the event is given by
 4   6   8    4  6  8 
Q =   p 2 (1 − p) 2   p 2 (1 − p) 4   p 6 (1 − p ) 2  =       p 10 (1 − p ) 8
 2   2   6    2  2  6
= (6 )(15)(28) p 10 (1 − p) 8 = 2520(0.8) (0.2) 8 = 0.00069
10

12.5 A lady invites 12 people for dinner at her house. Unfortunately the dining table can only
seat 6 people. Her plan is that if 6 or fewer guests come, then they will be seated at the
table (i.e., they will have a sit-down dinner); otherwise she will set up a buffet-style meal.
The probability that each invited guest will come to dinner is 0.4, and each guest’s
decision is independent of other guests’ decisions. Determine the following:
(a) The probability that she has a sit-down dinner
(b) The probability that she has a buffet-style dinner
(c) The probability that there are at most three guests
Solution: Let the random variable N denote the number of guests that come for the dinner. Then
N has the binomial distribution with the PMF
3
 12  12 
p N (n) =   p n (1 − p)12−n =   (0.4 ) (0.6)12−n n = 0, 1, Κ , 12
n

 
n  
n
Let A denote the event that she has a sit-down dinner. Thus, A denotes the event that she has a
buffet-style dinner.
(a) The probability that she has a sit-down dinner is given by
6 6
12 
P[ A] = ∑ p N (n) = ∑   (0.4 ) (0.6)12−n = (0.6)12 + 12(0.4)(0.6)11 + 66(0.4) 2 (0.6)10
n

n =0 n=0  n 

+ 220(0.4) (0.6) 9 + 495(0.4) 4 (0.6) 8 + 792(0.4) 5 (0.6) 7 + 924(0.4) 6 (0.6) 6

3

= 0.8418
(b) The probability that she has a buffet-style dinner is given by

P[ A ] = 1 − P[ A] = 0.1582

(c) The probability that there are at most 3 guests is given by

3 3
12 
P[ N ≤ 3] = ∑ p N (n) = ∑   (0.4) (0.6)12− n
n

n =0 n=0  n 

= (0.6)12 + 12(0.4)(0.6)11 + 66(0.4) 2 (0.6)10 + 220(0.4) 3 (0.6) 9

= 0.2253

12.6 A Girl Scout troop sells cookies from house to house. One of the parents of the girls
figured out that the probability that they sell a set of packs of cookies at any house they
visit is 0.4, where it is assumed that they sell exactly one set to each house that buys their
cookies.
(a) What is the probability that the house where they make their first sale is the fifth
house they visit?
(b) Given that they visited 10 houses on a particular day, what is the probability that they
sold exactly 6 sets of cookie packs?
(c) What is the probability that on a particular day the third set of cookie packs is sold at
the seventh house that the girls visit?
Solution: Let X k be a random variable that denotes the number of houses they visited up to and
including the house where they sold their kth set of cookie packs. Then X k is a kth-order Pascal
random variable whose PMF is given by
 n − 1 k  n − 1
p X k (n) =   p (1 − p ) n − k =   (0.4) k (0.6) n − k k = 1, 2, Κ ; n = k , k + 1, Κ
 k − 1   k − 1 
(a) The probability that the house where they make their first sale is the fifth house they visit
is given by

4
  5 − 1
P[ X 1 = 5] = p X 1 (5) =   (0.4)1 (0.6) 4 = (0.4)(0.6) 4 = 0.05184
 1 − 1 
(b) Let the random variable X (10) denote the number of sets of cookie packs they sell, given
that they visited 10 houses on a particular day. Then X (10) is a binomially distributed
random variable with the PMF
10  10 
p X (10) ( x) =   p x (1 − p )10− x =   (0.4) x (0.6)10− x x = 0, 1, Κ , 10
x x
Thus, the probability that that they sold exactly 6 sets of cookie packs is given by

10 
P[ X (10 = 6] = p X (10) (6) =   (0.4) 6 (0.6) 4 = 210 (0.4) 6 (0.6) 4 = 0.1115
6
(c) The probability that on a particular day the third set of cookie packs is sold at the seventh
house that the girls visit is given by
 7 − 1 6
P[ X 3 = 7] = p X 3 (7) =   (0.4) 3 (0.6) 4 =   (0.4) 3 (0.6) 4 = 15(0.4) 3 (0.6) 4 = 0.1244
 3 − 1  2

Section 12.3: Random Walk

12.7 A bag contains 3 red balls, 6 green balls, and 2 blue balls. Jack plays a game in which he
bets $1 to draw a ball from the bag. If he draws a green ball, he wins $1 dollar; otherwise
he loses a dollar. Assume that the balls are drawn with replacement and that Jack starts
the game with $50 with the hope of reaching $100 before going bankrupt. What is the
probability that he will succeed?
Solution: Since there are 11 balls and the balls are drawn with replacement, the probability of
success (i.e., drawing a green ball) in each game is p = 6 / 11. With k = 50 and N = 100, the
probability that Jack will go bankrupt (i.e., ruined) is given by

( ) −( )
1− p k 1− p N
( 56 )50 − ( 56 )100
= = = 0.00011
p p
r50
1− ( ) 1 − ( 56 )
1− p N 100
k =50
p N =100
p = 6 / 11

12.8 Consider a gambler’s ruin game in which p = q = 1 / 2. Let d i , 0 < i < N , denote the
expected duration of a game in which a gambler starts from state i, such as starting with
$i. He gets $1 when he wins a game and loses $1 when he loses a game. The boundary
conditions for the series of games are d 0 = 0 and d N = 0; that is, the gambler is ruined
when he enters state 0, and the series ends when he enters state N.
(a) Show that d i is given by

5
 0 i = 0, N

d i =  d i +1 + d i −1
1 + 2
i = 1, 2, Κ , N − 1

(b) Obtain the general expression for d i from the above equation terms of N.
Solution: Let Di be a random variable that denotes the duration of a game in which a player
starts in state i. Thus, E[ Di ] = d i , where d 0 = d N = 0. Let W denote the event that he wins a
game and let L denote the event that he loses a game.
(a) Let p denote the probability that he wins a game. Then d i is given by

d i = E[ Di | W ]P[W ] + E[ Di | L]P[ L] = p[1 + d i +1 ] + (1 − p)[1 + d i −1 ]

= 1 + pd i +1 + (1 − p )d i −1
Since p = 1 / 2, we have that
0 i = 0, N

d i =  d i +1 + d i −1
1 + 2
i = 1, 2, Κ , N − 1

(b) From the previous result we have that

d i +1 + d i −1
di = 1 + ⇒ d i +1 = 2d i − d i −1 − 2
2
Thus,
d 2 = 2d1 − d 0 − 2 = 2(d 1 − 1)
d 3 = 2d 2 − d1 − 2 = 2{2(d1 − 1)} − d1 − 2 = 3(d1 − 2)
d 4 = 2d 3 − d 2 − 2 = 2{3(d1 − 2)} − 2(d 1 − 1) − 2 = 4(d1 − 3)
d 5 = 2d 4 − d 3 − 2 = 2{4(d1 − 3)} − 3(d1 − 2) − 2 = 5(d1 − 4)
From these results we see that in general we have that d i = i (d1 − i + 1). Since
d N = 0 = N (d1 − N + 1), we have that d1 = N − 1. Thus,

d i = i (d1 − i + 1) = i ( N − 1 − i + 1) = ( N − 1)i i = 1, 2, Κ , N − 1

12.9 Consider a variation of the gambler’s ruin problem with parameters p (i.e., the probability
that player A wins a game) and N modeled in terms of a random walk with reflecting
barrier at zero. Specifically, when state 0 is reached, the process moves to state 1 with
probability p 0 or stays at state 0 with probability 1 − p 0 . Thus, the only trapping state is
N. That is, only player B can be ruined.
(a) Give the state transition diagram of the process.

6
 (b) If ri is the probability of player B being ruined when the process is currently in state
i, obtain the expression for ri to show what happens on the first game when the pro-
cess is in state i.
Solution:
(a) The state transition diagram of the process is as follows:

p0 p p p p
1
0 1 2 Λ N−2 N −1
p
N

1 − p0 1− p 1− p 1− p
1− p 1− p

(b) The probability of player B being ruined when the process is currently is state i, ri , can
be obtained from the following relationship:
 pri +1 + (1 − p)ri −1 i = 1, 2, Κ , N − 1

ri =  p 0 r1 + (1 − p 0 )r0 i=0
0 i=N


12.10 Ben and Jerry play a series of games of checkers. During each game each player bets a
$1, and whoever wins the game gets the $2. Ben is a better player than Jerry and has a
probability 0.6 of winning each game. Initially Ben had $9, while Jerry had $6, and the
game is over when either player is wiped out.
(a) What is the probability that Ben is ruined; that is, that Jerry will wipe him out?
(b) What is the probability that Jerry is ruined?
Solution: The total available amount is N = 9 + 6 = 15.
(a) Since Ben started with $9 and the probability that he wins a game is p = 0.6, the
probability that he is ruined is given by

( ) −( )
1− p 9 1− p 15
( 00..46 )9 − ( 00..46 )15 ( 46 )9 − ( 64 )15
= = = = 0.02378
p p
r9
1− ( ) 1− p 15
1 − ( 00..46 ) 1 − ( 46 )
15 15
p

(b) Since Jerry started with $6 and the probability that he wins a game is q = 0.4, the
probability that he is ruined is given by

( ) −( )
1− q 6 1− q 15
( 00..64 )6 − ( 00..64 )15 ( 64 )6 − ( 64 )15
= = = = 0.97622 = 1 − r9
q q
r6
1− ( ) 1− q 15
1 − ( 00..64 ) 1 − ( 64 )
15 15
q

12.11 Ben and Jerry play a series of games of cards. During each game each player bets a $1,
and whoever wins the game gets the $2. Sometimes a game can end in a tie, in which
case neither player loses his money. Ben is a better player than Jerry and has a probability

7
 0.5 of winning each game, a probability 0.3 of losing each game, and probability 0.2 of
tying with Jerry. Initially Ben had $9, while Jerry had $6, and the game is over when
either player is wiped out.
(a) Give the state transition diagram of the process.
(b) If rk denotes the probability that Ben is ruined, given that the process is currently in
state k, obtain an expression for rk in the first game when the process is in state k.
Solution: Let k denote the state in which Ben has a total of $k left. The total amount is N = 15.
(a) The state transition diagram of the process is given by

(b) If rk denotes the probability that Ben is ruined, given that the process is currently in state
k, the expression for rk in the first game when the process is in state k is given by

rk = 0.5rk +1 + 0.3rk −1 + 0.2rk ⇒ 0.8rk = 0.5rk +1 + 0.3rk −1 ⇒ rk = 1.25{0.5rk +1 + 0.3rk −1 }

Section 12.4: Gaussian Process

12.12 Suppose that X(t) is a wide-sense stationary Gaussian process with the autocorrelation
function
R XX (τ ) = 4 + e −|τ |
Determine the covariance matrix for the random variables X (0), X (1), X (3) , and X (6).
Solution: The expected value of X (t ) is

E[ X (t )] = ± lim R XX (τ ) = ± lim 4 + e −|τ | = ± 4 = ±2

τ →∞ τ →∞

Let X 1 = X (0), X 2 = X (1), X 3 = X (3), X 4 = X (6). Then

Cij = Cov ( X i , X j ) = R XX (i, j ) − µ X (i) µ X ( j ) = R XX (i − j ) − 4 = e −| j −i|

Thus, the covariance matrix for the random variables X (0), X (1), X (3) , and X (6) is given by

1 e −1 e −3 e −6 
 −1 
e 1 e −2 e −5 
C XX =  −3
e e −2 1 e −3 
 −6 −5 
e e e −3 1 

8
12.13 A Gaussian process X(t) has an autocorrelation function
4 sin(πτ )
R XX (τ ) =
πτ
Determine the covariance matrix for the random variables X (t ), X (t + 1), X (t + 2) and
X (t + 3).
Solution: We have that

 sin(πτ ) 
RXX (τ ) = 4  = 4 sinc(πτ )
 πτ 

Thus, the expected value of X (t ) is given by

E[ X (t )] = ± lim RXX (τ ) = 0
τ →∞

Let X 1 = X (t ), X 2 = X (t + 1), X 3 = X (t + 2), X 4 = X (t + 3). Then

C ji = Cov ( X i , X j ) = R XX (i, j ) − µ X (i ) µ X ( j ) = R XX ( j − i )

Thus, the covariance matrix for the random variables X (t ), X (t + 1), X (t + 2), and X (t + 3) is
given by

 4 sin(π ) 4 sin(2π ) 4 sin(3π ) 

 4 
π 2π 3π
 4 sin(−π ) 4 sin(πτ ) 4 sin(2π )  4 0 0 0
 4   4 0 0
C XX = −π πτ 2π  = 0
 4 sin(−2π ) 4 sin(−π )
4
4 sin(π )  0 0 4 0
 − 2π −π π   
 4 sin(−3π ) 4 sin(−2π ) 4 sin(−π )  0 0 0 4
 4 
 − 3π − 2π −π 

12.14 Suppose X(t) is a Gaussian random process with a mean E[ X (t )] = 0 and autocorrelation
function R XX (τ ) = e −|τ | . Assume that the random variable A is defined as follows:
1
A = ∫ X (t )dt
0

Determine the following:

(a) E[ A]
(b) σ A2
Solution:
(a) The expected value of A is E[ A] = E  ∫ X (t )dt  = ∫ E[ X (t )]dt = 0.
1 1

 0  0
(b) Since E[ A] = 0, σ X = E[ A ]. Thus,
2 2

9
 σ X2 = E[ A 2 ] = E  ∫ X (t )dt ∫ X (u )du  = ∫
1 1 1 1

 0 0  ∫ E[ X (t )X (u)]dtdu
0 0

1 1 1 1
−|u −t |
=∫ ∫ R XX (u − t )dtdu = ∫ ∫e dtdu
0 0 0 0

Consider the following figure:

u =t

t >u

u>t

Since

−|u −t |
e − ( u −t ) u≥t
e =  −( t −u )
e u<t

we have that
1 1 1 u 1 t 1 u
σ X2 = ∫ ∫e
−|u −t |
dtdu = ∫ ∫ e −( u −t ) dtdu + ∫ ∫e
− ( t −u )
dudt = 2 ∫ ∫ e −( u −t ) dtdu
0 0 u =0 t = 0 t =0 u =0 u = 0 t =0

= 2∫
1

u =0
[ ] 1
e −u e u − 1 du = 2∫ [1 − e −u ]du = 2 u + e −u
u =0
[ ]
1
0 = 2(1 + e −1 − 1) = 2e −1
= 0.7357
12.15 Suppose X(t) is a Gaussian random process with mean E[ X (t )] = 0 and autocorrelation
function R XX (τ ) = e −|τ | . Assume that the random variable A is defined as follows:
B
A = ∫ X (t )dt
0

where B is a uniformly distributed random variable with values between 1 and 5 and is
independent of the random process X(t). Determine the following:
(a) E[ A]
(b) σ A2
Solution:
10
 (a) The mean of A is given by
E[ A] = ∫ E[ A | B = b] f B (b)db = ∫ E  ∫ X (t )dt  f B (b)db
5 5 b

b =1 b =1 
 t =0 
5 b
=∫ ∫ E[ X (t )] f B (b)dtdb = 0
b =1 t =0

(b) Since the mean of A is zero, the variance of A is σ X2 = E[ A 2 ]; that is,

σ A2 = E[ A 2 ] = ∫ E[ A 2 | B = b] f B (b)db = ∫ E  ∫ ∫ X (t )X (u )dudt  f B (b)db
5 5 b b

b =1 b =1 
 t =0 u =0 
5 b b 5 b b
=∫ ∫ ∫ E[ X (t )X (u )] f B (b)dudtdb = ∫ ∫ ∫ R XX (t , u ) f B (b)dudtdb
b =1 t =0 u =0 b =1 t = 0 u = 0
5 b b 5 b b
=∫ ∫ ∫ e −|u −t| f B (b)dudtdb = 2∫ ∫ ∫ e −(u −t ) f B (b)dudtdb
b =1 t =0 u =0 b =1 t =0 u =0
5

= 2∫
5

b =1
(b + e −b
) 2 5
(
1 b2 
− 1 f B (b)db = ∫ b + e −b − 1 db =  − e −b − b 
4 = 2 2
)
b 1
1

=
1
2
[8 + e −1 − e −5 = 4.1806 ]
Section 12.5: Poisson Process
12.16 University buses arrive at the Students Center to take students to their classes according
to a Poisson process with an average rate of 5 buses per hour. Chris just missed the last
bus. What is the probability that he waits more than 20 minutes before boarding a bus?
Solution: Since buses arrive according to a Poisson process with an average of 5 buses per hour,
the times X between bus arrivals are exponentially distributed with the PDF
f X ( x ) = λe − λ x x≥0
Now, λ = 5 buses/hour or λ = 5 / 60 = 1 / 12 buses per minute. Since Chris just missed the last
bus, the time until the next bus arrives is the random variable X. Therefore, the probability that
he waits more than 20 minutes before boarding a bus is given by
∞
P[ X > 20] = ∫ f X ( x)dx = − e −λx
20
[ ] ∞
20 = e −20λ = e −20 / 12 = e −5 / 3 = 0.18887
12.17 Cars arrive at a gas station according to a Poisson process at an average rate of 12 cars
per hour. The station has only one attendant. If the attendant decides to take a 2-minute
coffee break when there are no cars at the station, what is the probability that one or more
cars will be waiting when he comes back from the break, given that any car that arrives
when he is on coffee break waits for him to get back?
Solution: Let N denote the number of cars that arrive over an interval of length t. Since cars
arrive according to a Poisson process at an average of 12 cars per hour, the PMF of N is given by

p N (n, t ) =
(λt )n e − λt n = 0, 1, 2, Κ ; t ≥ 0
n!

11
where λ = 12 / 60 = 1 / 5 cars per minute. The probability that one or more cars will be waiting
when the attendant comes back from a 2-minute break is the probability that one or more cars
arrive over the 2-minute break period, which is given by

P[ N ≥ 1, t = 2] = 1 − P[ N = 0, t = 2] = 1 − p N (0, 2) = 1 − e −2 / 5 = 0.3297
12.18 Cars arrive at a gas station according to a Poisson process at an average rate of 50 cars
per hour. There is only one pump at the station, and the attendant takes 1 minute to fill up
each car. What is the probability that a waiting line will form at the station? (Note: A
waiting line occurs if two or more cars arrive in any 1-minute period.)
Solution: Let N denote the number of cars that arrive over an interval of length t. Since cars
arrive according to a Poisson process at an average of 50 cars per hour, the PMF of N is given by

pN (n, t ) =
(λ t )n e − λ t
n = 0, 1, 2, Κ ; t ≥ 0
n!
where λ = 50 / 60 = 5 / 6 cars per minute. Let A denote the event that a waiting line occurs. Then,
the probability that event A occurs at the station is given by

P[ A] = P[ N ≥ 2, t = 1] = 1 − P[ N < 2, t = 1] = 1 − p N (0, 1) − p N (1, 1)

11 −5 / 6
= 1 − e −λ − λe −λ = 1 − (1 + λ )e −λ = 1 − e = 0.2032
6
12.19 Studies indicate that the probability that three cars will arrive at a parking lot in a 5-
minute interval is 0.14. If cars arrive according to a Poisson process, determine the
following:
(a) The average arrival rate of cars
(b) The probability that no more than 2 cars arrive in a 10-minute interval
Solution: Let λ denote the average arrival rate of cars per minute and let K denote the number
of cars that arrive over an interval of t minutes. Then, the PMF of K is given by

p K (k , t ) =
(λt )k e − λt k = 0, 1, 2, Κ ; t ≥ 0
k!
(a) Given that the probability that 3 cars will arrive in a 5-minute interval is 0.14, we have
that

pK (3, 5) =
(5λ ) e −5λ
3
=
125λ3 e −5λ
= 0.14 ⇒ λ3 e −5λ =
6(0.14)
= 0.00672
3! 6 125
Solving this equation numerically we obtain λ = 1.
(b) The probability that no more than 2 cars arrive in a 10-minute interval is given by

{
P[ K ≤ 2, t = 10] = p K (0, 10) + p K (1, 10) + p K (2, 10) = e −10λ 1 + 10λ + 50λ2 }
=e −10
{1 + 10 + 50} = 61e −10
= 0.00277

12
12.20 Telephone calls arrive at a switching center according to a Poisson process at an average
rate of 75 calls per minute. What is the probability that more than three calls arrive within
a 5-second period.
Solution: Let N denote the number of calls that arrive over an interval of length t. Since calls
arrive according to a Poisson process at an average of 75 cars per minute, the PMF of N is given
by

pN (n, t ) =
(λ t )n e − λ t
n = 0, 1, 2, Κ ; t ≥ 0
n!
where λ = 75 / 60 = 1.25 calls per second. The probability that more than three calls arrive
within a 5-second period is given by

P[ N > 3, t = 5] = 1 − P[ N ≤ 2, t = 5] = 1 − {p N (0, 5) + p N (1, 5) + p N (2, 5)}

{ }
= 1 − e −5λ 1 + 5λ + 12.5λ2 = 1 − e −6.25 {1 + 6.25 + 19.53125}
= 1 − 26.78125e −6.25 = 0.9483

12.21 An insurance company pays out claims on its life insurance policies in accordance with a
Poisson process with an average rate of 5 claims per week. If the amount of money paid
on each policy is uniformly distributed between $2,000 and $10,000, what is the mean of
the total amount of money that the company pays out in a four-week period?
Solution: Let M denote the number of claims that are paid in an n-week period. Then, the PMF
and expected value of M are given by

pM (m, n) =
(λ n ) m e − λn
m = 0, 1, 2, Κ ; n ≥ 0
m!
E[ M ] = λn
where λ = 5. Let X denote the amount paid on a policy. Since X is uniformly distributed between
$2,000 and $10,000, its mean is

2,000 + 10,000
E[ X ] = = $6,000
2

Thus, the expected total amount of money in dollars, E[T ], that the company pays out in a 4-
week period is given by

E[T ] = E[ M ]E[ X ] = (5)(4)(6,000) = $120,000

12.22 Customers arrive at the neighborhood bookstore according to a Poisson process with an
average rate of 10 customers per hour. Independent of other customers, each arriving
customer buys a book with probability 1/8.
(a) What is the probability that the bookstore sells no book during a particular hour?
(b) What is the PDF of the time until the first book is sold?

13
Solution: This is an example of subdivision of a Poisson process, which is illustrated in the
figure below.

λB
1
λ 8

7
8
λ NB

Let λ denote the arrival rate of customers, let λ B denote the arrival rate of customers who buy
books at the bookstore, and let λ NB denote the arrival rate of customers who do not buy books.
Thus, we have that
λ 10
λ B = = = 1.25
8 8
Let K denote the number of books that the bookstore sells in one hour. Then we know that K is a
Poisson random variable with PMF

p K (k ) =
λkB e − λ B

=
(1.25) e −1.25
k
k = 0, 1, 2, Κ
k! k!
(a) The probability that the bookstore sells no book during a particular hour is given by

P[ K = 0] = p K (0) = e −1.25 = 0.2865

(b) Let X denote the time between book sales. Then X is an exponentially distributed random
variable with the PDF

f X ( x) = λ B e − λB x = 1.25e −1.25 x x≥0

12.23 Joe is a student who is conducting experiments with a series of lightbulbs. He started with
10 identical lightbulbs, each of which has an exponentially distributed lifetime with a
mean of 200 hours. Joe wants to know how long it will take until the last bulb burns out
(or fails). At noontime Joe stepped out to get some lunch with 6 bulbs still on. Assume
that Joe came back and found that none of the 6 bulbs has failed.
(a) After Joe came back, what is the expected time until the next bulb failure?
(b) What is the expected length of time between the fourth bulb failure and the fifth bulb
failure?
Solution: Let Y denote the life of a bulb. Since Y is an exponentially distributed random variable
with rate λ (or mean 1 / λ = 200), the failure (or burnout) rate when k bulbs are still operational
is kλ .
(a) Since the lifetimes are exponentially distributed, when Joe comes back the lifetimes of
the bulbs start from scratch because of the forgetfulness property of the exponentially
distributed random variable. Thus, the 6 bulbs will operate as a “superbulb” whose failure

14
 rate is 6λ . Since the time until a member of the superbulb fails is also exponentially
distributed, the expected time until the next bulb failure occurs is
1 1  1  200
=  = = 33.33 hours
6λ 6  λ  6

(b) If 4 bulbs have failed, then only 2 bulbs are operational and the failure rate is 2λ . Thus,
the expected length of time between the fourth bulb failure and the fifth bulb failure is

1 1  1  200
=  = = 100 hours
2λ 2  λ  2

12.24 Three customers A, B, and C simultaneously arrive at a bank with two tellers on duty.
The two tellers were idle when the three customers arrived, and A goes directly to one
teller, B goes to the other teller, and C waits until either A or B leaves before she can
begin receiving service. If the service times provided by the tellers are exponentially
distributed with a mean of 4 minutes, what is the probability that customer A is still in the
bank after the other two customers leave?
Solution: Let X denote the time to serve a customer. Then the PDF of X is given by

f X ( x) = λe − λx = 0.25e −0.25 x x≥0

The time Y to serve customers B and C is the second-order Erlang random variable whose PDF is
given by
f Y ( y ) = λ2 ye − λx = 0.0625 ye −0.25 y y≥0

The probability that customer A is still in the bank after the other two customers leave is simply
the probability that X is greater than Y, which can be obtained through the figure below as
follows:

X=Y

Y>X

X>Y

∞ x ∞ x
P[ X > Y ] = ∫ ∫ f XY ( x, y )dydx = ∫ ∫ f X ( x) f Y ( y )dydx
x =0 y =0 x =0 y =0

= λ3 ∫
∞

x =0
e − λx {∫ x

y =0
}
ye −λy dy dx

15
Let u = y ⇒ du = dy, and let dv = e − λy dy ⇒ v = −e − λy / λ. Thus,

P[ X > Y ] = λ3
∫
∞

x =0
e − λx
{∫ x

y =0
ye − λy
}
dy dx = λ 3
∫
∞

x =0
e − λx
 ye −λy  x 1 x
 −
 λ  0 λ ∫ y =0
+ e − λy
dy

dx


∞  xe −λx 1  e −λy  x  ∞ e
− λx
e − 2λx xe − 2λx 
− λx
= λ ∫ e −
3
+ − dx = λ3
∫x=0  λ2 λ2 − λ dx
−
x =0
 λ λ  λ  0 
 − λx ∞ ∞

3  e   e − 2 λx  1 ∞ 1 1 1  1 
= λ − 3  +  3  − 2 ∫ 2λxe −2 λx dx  = λ3  3 − 3 − 2  
 λ  0  2λ  0 2λ x =0  λ 2λ 2λ  2λ 
1 1 1
=1− − =
2 4 4

Alternative Solution Method: Another way to solve the problem is to use the forgetfulness
property of the exponential distribution as follows. Let X A denote the time to serve A, X B the
time to serve B and X C the time to serve C. Let 1 / λ A be the mean of X A , let 1 / λ B be the mean
of X B , and let 1 / λC be the mean of X C , where λ A = λ B = λC = 1 / 4. The probability that B
leaves before A is given by
λB
P[ X A > X B ] =
λ A + λB
Because of the forgetfulness property of the exponential distribution, after B leaves, A’s service
starts from scratch. Thus, the probability that C leaves before A is given by
λC
P[ X A > X C ] =
λ A + λC
Therefore, the probability that customer A is still in the bank after the other two customers leave
is given by

P[ X A > X B + X C ] = P[ X A > X B ]P[ X A > X C | X A > X B ] = P[ X A > X B ]P[ X A > X C ]

 λ B  λ C   1  1  1
=   =    =
 λ A + λ B   λ A + λC   2   2  4
12.25 The times between component failures in a certain system are exponentially distributed
with a mean of 4 hours. What is the probability that at least one component failure occurs
within a 30-minute period?
Solution: Since the times between component failures are exponentially distributed, the number
N of failures within an interval of length t is a Poisson random variable with rate λ , where
1 / λ = 4 × 60 = 240 minutes or λ = 1 / 240. Thus, the PMF of N is given by

p N (n, t ) =
(λt )n e −λt n = 0, 1, 2, Κ ; t ≥ 0
n!

16
Therefore, the probability that at least one component failure occurs within a 30-minute period is
given by
P[ N ≥ 1, t = 30] = 1 − P[ N = 0, t = 30] = 1 − p N (0, 30) = 1 − e −30 / 240 = 1 − e −0.125 = 0.1175

12.26 Students arrive at the professor’s office for extra help according to a Poisson process with
an average rate of 4 students per hour. The professor does not start the tutorial until at
least 3 students are available. Students who arrive while the tutorial is going on will have
to wait for the next session.
(a) Given that a tutorial has just ended and there are no students currently waiting for the
professor, what is the mean time until another tutorial can start?
(b) Given that one student was waiting when the tutorial ended, what is the probability
that the next tutorial does not start within the first 2 hours?
Solution: Let T denote the interval between student arrival times at the professor’s office. Then
the PDF of T is given by

f T (t ) = λe − λt t≥0

where λ = 4 students/hour. Let X denote the time that elapses from the instant one session ends
until the time the next session begins.
(a) Given that a tutorial has just ended and there are no students currently waiting for the
professor, the mean time until another tutorial can start is the mean time until the third
student arrives, which is given by

3 3
E[ X ] = 3E[T ] = = hours = 45 minutes
λ 4
(b) Given that one student was waiting when the tutorial ended, the probability that the next
tutorial does not start within the first 2 hours is the probability that the time until the
second of 2 other students arrives is greater than 2 hours measured from the time the last
session ended. This is essentially the probability that a second-order Erlang random
variable, X 2 , is greater than 2; that is,

1 (2λ )k e −2 λ 
P[ X 2 > 2] = 1 − P[ X 2 ≤ 2] = ∑   = e + 2λ e
−2λ − 2λ
= (1 + 2λ )e −2 λ
k =0  k! 
−8
= 9e = 0.0030

12.27 Customers arrive at a bank according to a Poisson process with an average rate of 6
customers per hour. Each arriving customer is either a man with probability p or a woman
with probability 1 − p. It was found that in the first 2 hours the average number of men
who arrived at the bank was 8. What is the average number of women who arrived over
the same period?

17
Solution: This is an example of the subdivision of a Poisson process. If λ M is the arrival rate of
male customers and λW is the arrival rate of female customers, we can represent the process as
shown below.

λM
p
λ
1− p
λW

Let N M denote the number of men who arrive in an interval of length 2 hours, and let N W
denote the number women who arrive in an interval of length 2 hours. Since both N M and N W
are Poisson random variables with rates λ M = pλ and λW = (1 − p )λ , respectively, where
λ = 6, we have that
8 2
E[ N M ] = 2λ M = 2 p (6) = 12 p = 8 ⇒ p = =
12 3
Thus, the average number of women who arrived over the same period is given by

1
E[ N W ] = 2λW = 2(1 − p )(6) = 12(1 − p ) = 12  = 4
 3

12.28 Alan is conducting an experiment to test the mean lifetimes of two sets of electric bulbs
labeled A and B. The manufacturer claims that the mean lifetime of bulbs in set A is 200
hours, while the mean lifetime of the bulbs in set B is 400 hours. The lifetimes for both
sets are exponentially distributed. Alan’s experimental procedure is as follows. He started
with one bulb from each set. As soon as a bulb from a given set fails (or burns out), he
immediately replaces it with a new bulb from the same set and writes down the lifetime
of the burnt-out bulb. Thus, at any point in time he has two bulbs on, one from each set.
If at the end of the week Alan tells you that 8 bulbs have failed, determine the following:
(a) The probability that exactly 5 of those 8 bulbs are from set B
(b) The probability that no bulb will fail in the first 100 hours
(c) The mean time between two consecutive bulb failures
Solution: Let X denote the time until a bulb from set A fails, and let Y denote the time until a
bulb from set B fails. Then the PDFs and expected values of X and Y are given by

f X ( x) = λ A e − λ A x x≥0
f Y ( y ) = λ B e − λB y y≥0
1
E[ X ] = = 200
λA
1
E[Y ] = = 400
λB
18
Let p A denote the probability that a bulb from set A fails before a bulb from set B. Then we have
that
λA 1
2
pA = = 1 200 1 =
λ A + λ B 200 + 400 3

Thus, the probability p B that a bulb from set B fails before a bulb from set A is given by
1
pB = 1 − p A =
3
(a) Let K denote the number of bulbs from set B that fail out of the 8 bulbs that failed. Then
K has a binomial distribution whose PMF is given by
k 8− k
8  8   1   2 
p K (k ) =   p Bk (1 − p B ) 8− k =       k = 0, 1, Κ , 8
k  k  3  3 
Thus, the probability that exactly 5 of the 8 bulbs are from set B is given by
5 3
8  1   2 
P[ K = 5] = p K (5) =       = 0.0683
 5  3   3 
(b) Since the two-bulb arrangement constitutes a competing Poisson process, the composite
failure rate if λ = λ A + λ B = 2001
+ 400
1
= 4003
. The time V until a bulb fails is exponentially
distributed with the PDF and CDF
f V (v) = λe − λv v≥0
FV (v) = 1 − e −λv
Thus, the probability that no bulb will fail in the first 100 hours is given by

P[V > 100] = 1 − P[V ≤ 100] = 1 − FV (100) = e −100( 3 / 400) = e −3 / 4 = 0.4724

(c) The mean time between two consecutive bulb failures is given by

1 1 400
E[V ] = = = = 133.33
λ 1
200 + 1
400 3

12.29 The Merrimack Airlines company runs a commuter air service between Manchester, New
Hampshire, and Cape Cod, Massachusetts. Since the company is a small one, there is no
set schedule for their flights, and no reservation is needed for the flights. However, it has
been determined that their planes arrive at the Manchester airport according to a Poisson
process with an average rate of 2 planes per hour. Vanessa arrived at the Manchester
airport and had to wait to catch the next flight.
(a) What is the mean time between the instant Vanessa arrived at the airport until the
time the next plane arrived?

19
 (b) What is the mean time between the arrival time of the last plane that took off from the
Manchester airport before Vanessa arrived and the arrival time of the plane that she
boarded?
Solution: Let X be a random variable that denotes the times between plane arrivals. Since the
number of planes arriving within any time interval is a Poisson random variable with a mean of
λ = 2 planes per hour, the PDF of X is given by
f X ( x) = λe − λx x≥0
where E[ X ] = 1 / λ = 1 / 2 hours or 30 minutes. We are told that Vanessa arrived at the airport
and had to wait to catch the next flight.
(a) Due to the forgetfulness property of the exponential distribution, the mean time between
the instant that Vanessa arrived at the airport until the time the next plane arrived is the
same E[ X ] = 30 minutes.
(b) The time T between the arrival time of the last plane that took off from Manchester
airport before Vanessa arrived and the arrival time of the plane that she boarded is the
gap that Vanessa entered by random incidence. Thus, E[T ] = 2 E[ X ] = 1 hour.

12.30 Bob has a pet that requires the light in his apartment to always be on. To achieve this,
Bob keeps three lightbulbs on with the hope that at least one bulb will be operational
when he is not at the apartment. The lightbulbs have independent and identically
distributed lifetimes T with PDF f T (t ) = λe − λt , t ≥ 0.
(a) Probabilistically speaking, given that Bob is about to leave the apartment and all three
bulbs are working fine, what does he gain by replacing all three bulbs with new ones
before he leaves?
(b) Suppose X is the random variable that denotes the time until the first bulb fails. What
is the PDF of X?
(c) Given that Bob is going away for an indefinite period of time and all three bulbs are
working fine before he leaves, what is the PDF of Y, the time until the third bulb fail-
ure after he leaves?
(d) What is the expected value of Y?
Solution:
(a) Probabilistically speaking, given that Bob is about to leave the apartment and all three
bulbs are working fine, he gains nothing by replacing the bulbs with new ones before he
leaves because the time until any of the 3 bulbs fails is statistically identical to the time to
failure of a new bulb. This is the result of the forgetfulness property of the exponential
distribution.
(b) The 3 bulbs behave like a single system with a failure rate λ X = 3λ. Thus, the time X
until the first bulb fails is exponentially distributed with the PDF

f X ( x) = λ X e − λ X x = 3λe −3λx x≥0

(c) Given that Bob is going away for an indefinite period of time and all three bulbs are
working fine before he leaves, the random variable Y, which denotes the time until the 3rd
20
 bulb failure after he leaves, can be obtained as follows. Let X 1 denote the time that
elapses from the instant Bob leaves until the first bulb fails, X 2 the time between the first
bulb failure and the second bulb failure, and X 3 the time between the second bulb failure
and the third bulb failure. Then X 1 is exponentially distributed with parameter 3λ , X 2 is
exponentially distributed with parameter 2λ , and X 3 is exponentially distributed with
parameter λ. That is, the PDFs of X 1 , X 2 and X 3 are given, respectively, by
f X1 ( x) = 3λe −3λx x≥0
f X 2 ( x ) = 2λ e − 2 λx
x≥0
f X 3 ( x ) = λ e − λx x≥0
Thus, we have that Y = X 1 + X 2 + X 3 . Because of the forgetfulness property of the
underlying exponential distribution, the random variables X 1 , X 2 and X 3 are
independent. Therefore, the PDF of Y is the convolution of the PDFs of the 3 random
variables. That is,

f Y ( y) = f X1 ( y) ∗ f X 2 ( y) ∗ f X 3 ( y)

(d) The expected value of Y is

1 1 1 11
E[Y ] = E[ X 1 ] + E[ X 2 ] + E[ X 3 ] = + + =
3λ 2λ λ 6λ

12.31 Joe just replaced two lightbulbs one of which is rated 60 watts with an exponentially
distributed lifetime whose mean is 200 hours, and the other is rated 100 watts with an
exponentially distributed lifetime whose mean is 100 hours.
(a) What is the probability that the 60-watt bulb fails before the 100-watt bulb?
(b) What is the mean time until the first of the two bulbs fails?
(c) Given that the 60-watt bulb has not failed after 300 hours, what is the probability that
it will last at least another 100 hours?
Solution: Let X denote the lifetime of the 60-watt bulb and Y the lifetime of the 100-watt bulb.
Then the PDFs and means of these random variables are given by
f X ( x ) = λ e − λx x ≥ 0, λ > 0
− µy
f Y ( y ) = µe y ≥ 0, µ > 0
1 1
E[ X ] = = 60 ⇒ λ =
λ 60
1 1
E[Y ] = = 100 ⇒ µ =
µ 100
(a) The probability that the 60-watt bulb fails before the 100-watt bulb is given by

21
 λ 1
600 5
P[ X < Y ] = = 60
= =
λ+µ 1
60
+ 1
100
60 × 16 8

(b) The mean time until the first of the two bulbs fails is T = min {X , Y }. It is well known
(see Chapter 6) that the PDF of T is given by f T (t ) = γe −γt , t ≥ 0, where γ = λ + µ . Thus,
the mean of T is

1 1 1 600
E[T ] = = = = = 37.5
γ λ+µ 1
60
+ 100
1
16

(c) Due to the forgetfulness property of the exponential distribution, given that the 60-watt
bulb has not failed after 300 hours, the probability that it will last another 100 hours is
given by

P[ X ≥ 100] = e −100λ = e −100 / 60 = e −5 / 3 = 0.18887

12.32 A 5-motor machine can operate properly if at least 3 of the 5 motors are functioning. If
the lifetime X of each motor has the PDF f X ( x) = λe − λx , x ≥ 0, λ > 0, and if the lifetimes
of the motors are independent, what is the mean of the random variable Y, the time until
the machine fails?
Solution: This is an example of a combination of independent Poisson processes. Thus, initially
the 5 motors probabilistically operate as one unit with failure rate λ5 = 5λ . Then, after the first
failure the 4 remaining motors operate as a unit with failure rate λ 4 = 4λ starting from “scratch”
due to the forgetfulness property of the exponential distribution, and so on until only one motor
is left and the failure rate is λ1 = λ . The machine fails when the third motor fails. Thus, if the
random variable Y denotes the time until the machine fails, then we have that

1 1 1 1 1 1 47
E[Y ] = + + = + + =
λ5 λ4 λ3 5λ 4λ 3λ 60λ

12.33 Alice has two identical personal computers, which she never uses at the same time. She
uses one PC at a time and the other is a backup. If the one she is currently using fails, she
turns it off, calls the PC repairman and turns on the backup PC. The time until either PC
fails when it is in use is exponentially distributed with a mean of 50 hours. The time
between the moment a PC fails until the repairman comes and finishes repairing it is also
exponentially distributed with a mean of 3 hours. What is the probability that Alice is idle
because neither PC is operational?
Solution: Let X denote the time until a PC fails. Then the PDF of X is given by

f X ( x ) = λ e − λx x ≥ 0, λ > 0

where E[ X ] = 1 / λ = 50 ⇒ λ = 1 / 50. Similarly, let Y denote the time to repair a PC after it has
failed. Then the PDF of Y is given by

22
 f Y ( y ) = µe − µy y ≥ 0, µ > 0

where E[Y ] = 1 / µ = 3 ⇒ µ = 1 / 3. Since Alice has 2 identical PCs and uses one PC at a time
with the other as a backup that is used when the one in use fails, the probability that she is idle
because neither PC is operational is the probability that the time to repair a failed PC is greater
than the time until an operational PC fails. Thus, if A is the event that Alice is idle, we have that

λ 1
150 3
P[ A] = P[ X < Y ] = = 50
= = = 0.0566
λ+µ 1
50 + 1
3 50 × 53 53

12.34 Cars arrive from the northbound section of an intersection according to a Poisson process
at the rate of λ N cars per minute and from the eastbound section according to a Poisson
process at the rate of λ E cars per minute.
(a) Given that there is currently no car at the intersection, what is the probability that a
northbound car arrives before an eastbound car?
(b) Given that there is currently no car at the intersection, what is the probability that the
fourth northbound car arrives before the second eastbound car?
Solution: Let the random variable X denote the time between arrivals from the northbound
section of the intersection. Then the PDF of X is given by
f X ( x) = λ N e − λN x x≥0
Similarly, let the random variable Y denote the time between arrivals of cars from the eastbound
section of the intersection. Then the PDF of Y is given by
f Y ( y ) = λ E e − λE y y≥0
(a) Given that there is currently no car at the intersection, the probability that a northbound
car arrives before an eastbound car is given by the probability that X is less than Y, which
is
λN
P[ X < Y ] =
λN + λE
(b) Given that there is currently no car at the intersection, the event that the fourth
northbound car arrives before the second eastbound car can occur as follows:
1. The first 4 arrivals are northbound cars. The probability of this subevent is the
probability that there are 4 successes in 4 Bernoulli trials, where the probability of
success is p = λ N /(λ N + λ E ). Thus the subevent is defined by a binomial random
variable with 4 successes in 4 trials; that is, there is no failure.
2. There are 3 successes in the first 4 Bernoulli trials and the 5th trial results in a success.
Thus, this subevent is defined by the 4th-order Pascal random variable in which the 4th
success occurs on the 5th trial.
Since these two subevents are mutually exclusive, the probability q that the 4th
northbound car arrives before the 2nd eastbound car is given by

23
 4  5 − 1 4 4
q =   p 4 (1 − p) 0 +   p (1 − p )1 = p 4 +   p 4 (1 − p) = p 4 + 4 p 4 (1 − p )
4  4 − 1 3 
4
 λN    λE 
= p {1 + 4(1 − p )} = 
4
 1 + 4 
 λN + λE    λN + λE 

12.35 A one-way street has a fork in it, and cars arriving at the fork can either bear right or left.
A car arriving at the fork will bear right with probability 0.6 and will bear left with
probability 0.4. Cars arrive at the fork according to a Poisson process with a rate of 8 cars
per minute.
(a) What is the probability that at least four cars bear right at the fork in three minutes?
(b) Given that three cars bear right at the fork in three minutes, what is the probability
that two cars bear left at the fork in three minutes?
(c) Given that 10 cars arrive at the fork in three minutes, what is the probability that four
of the cars bear right at the fork?
Solution: This is an example of the subdivision of a Poisson process. Let λR denote the arrival
rate of cars that bear right, and let λ L denote the arrival rate of cars that bear left. Now,
λ R = 0.6λ = 0.6 × 8 = 4.8
λ L = 0.4λ = 0.4 × 8 = 3.2
The process is illustrated in the following figure.

λL
λ =8 0.4

0 .6
λR

(a) Let R denote the number of cars that bear right in an interval of length t. Since R is a
Poisson random variable, its PMF is given by

pR (r , t ) =
(λ R t )r e − λ t
R

=
(4.8t ) e −4.8t
r
r = 0, 1, 2, Κ ; t ≥ 0
r! r!
The probability that at least 4 cars bear right at the fork in 3 minutes is given by

P[ R ≥ 4, t = 3] = 1 − P[ R < 4, t = 3] = 1 − {p R (0, 3) + p R (1, 3) + p R (2, 3) + p R (3, 3)}

=1− e −14.4
 (14.4)
2
(14.4 ) 
3

1 + 14.4 + +  = 0.9996
 2 6 

24
 (b) Let L denote the number of cars that bear left in an interval of length t. Its PMF is also
given by

pL (l , t ) =
(λ L t )l e − λ t
L

=
(3.2t ) e −3.2t
l
l = 0, 1, 2, Κ ; t ≥ 0
l! l!
Since R and L are independent Poisson random variables, the probability that 2 cars bear
left at the fork in 3 minutes, given that 3 cars bear right at the fork in 3 minutes, is simply
the probability that 2 cars bear left in 3 minutes, which is given by

P[ L = 2, t = 3 | R = 3, t = 3] = P[ L = 2, t = 3] =
(3λ L )2 e −3λ L

=
(9.6)2 e −9.6 = 0.00312
2! 2
(c) Given that 10 cars arrive at the fork in 3 minutes, the probability that 4 of the cars bear
right at the fork is given by the binomial distribution
4 6
10   λ R   λ L 
4 6
10   4.8   3.2 
P[(R = 4, t = 3), (L = 6, t = 3)] =       =      
 4   λR + λL   λR + λL  4 8   8 
10 
=   (0.6 ) (0.4 ) = 0.1115
4 6

4

Section 12.7: Discrete-Time Markov Chains

12.36 Determine the missing elements denoted by x in the following transition probability
matrix:
 x 1/ 3 1/ 3 1/ 3 
1 / 10 x 1 / 5 2 / 5
P= 
 x x x 1 
 
 3/ 5 2 / 5 x x 

Solution: Since the sum of the entries in each row must sum to 1, the transition probability
matrix is given by
 0 1/ 3 1/ 3 1/ 3 
1 / 10 3 / 10 1 / 5 2 / 5
P= 
 0 0 0 1 
 
3/5 2/5 0 0 
12.37 Draw the state-transition diagram for the Markov chain with the following transition
probability matrix.

25
 1 / 2 0 0 1 / 2
1 / 2 1 / 2 0 0 
P=
1 / 4 0 1 / 2 1 / 4
 
 0 1 / 2 1 / 4 1 / 4
Solution: The state transition diagram is as follows:
1
1 2
1 1
2 2 2
1
4 1
1 2
2

1 1 1
4
4 1 4 3 2
4

12.38 Consider a Markov chain with the state-transition diagram shown in Figure 12.21.
2
1
1

1 1 3

1
3 1
1
3 1
3
4 5 6
1
2 1
2

Figure 12.21: Figure for Problem 12.38

(a) Give the transition probability matrix.

(b) Identify the recurrent states.
(c) Identify the transient states.
Solution:
(a) The transition probability matrix is given by

26
 0 1 0 0 0 0
0 0 1 0 0 0

P = 11 0 0 0 0 0
0 0 13 13 0
3 
0 0 0 0 0 1
0 0 0 0 12 12 
(b) Recurrent states: {1, 2, 3, 5, 6}
(c) The only transient state is state 4

12.39 Consider the Markov chain with the state-transition diagram shown in Figure 12.22.
1
2
6
1
3
2 5 2 3
1
4 3 4
1
2
1 1 7
1 4 4
6
1 1
4 6
1
2
1
3 1
6
1 3

1
3
1 1
1
4 3
3
1
2 2
3

4 8 9
1 1 3
2 4
4

Figure 12.22: Figure for Problem 12.39

(a) List the transient states, the recurrent states, and the periodic states.
(b) Identify the members of each chain of recurrent states.
(c) Give the transition probability matrix of the process.
(d) Given that the process starts in state 1, either determine the numerical value of the
probability that the process is in state 8 after an infinitely large number of transitions
or explain why this quantity does not exist.
Solution:
(a) Transient states: {1, 2, 3, 4}
Recurrent states: {5, 6, 7, 8, 9}

27
 Periodic states: None
(b) There are two chains of recurrent states, which are
Chain 1: {5, 6, 7}
Chain 2: {8, 9}
(c) The transition probability matrix is given by
 13 16 16 13 0 0 0 0 0
1 1 1 0 
 4 14 2 1 01 0 0 01 0
0 4 0 4 6 0 0 3 0 
0 0 1 1 0 0 0 0 0 
 2 2

P = 0 0 0 0 0 12 12 0 0
0 0 0 0 1 0 2 0 0 
 3
1 3
3
0 0
 0 0 0 0 4 4
0
0 0 0 0 0 0 0 13 23 
0 0 0 0 0 0 0 1 3 
 4 4

(d) Given that the process starts in state 1, let A denote the event that it leaves transient states
{1, 2, 3, 4}. Given event A, the probability that the process enters the chain {8, 9} is given
by
1
2
P[1 → 8 | A] = 3
=
1
3 + 1
6 3
After entering the chain {8, 9} the limiting probability that it is in state 8 can be obtained
as follows. Given that the process is in chain {8, 9}, let π k denote the limiting-state
probability that the process is in state k, k = 8, 9. Then
π 8 = 13 π 8 + 14 π 9 ⇒ 23 π 8 = 14 π 9 ⇒ π 9 = 83 π 8
1 = π 8 + π 9 = π 8 {1 + 83 } = 113 π 8 ⇒ π 8 = 113
Thus, given that the process starts in state 1, the probability that it is in state 8 after an
infinitely large number of transitions is the probability that it enters the chain {8, 9}
multiplied by the limiting state probability of being in state 8 once it enters that chain.
That is, this probability exists and is equal to
2 3 2
× =
3 11 11

12.40 Consider the three-state Markov chain shown in Figure 12.23:

28
 1 2 1
3 10

1 1 9
6 10
2
3

1 3 7
12
4

Figure 12.23: Figure for Problem 12.40

(a) Identify the transient states, the recurrent states, the periodic states, and the members
of each chain of recurrent states.
(b) Either determine the limiting-state probabilities or explain why they do not exist.
(c) Given that the process is currently in state 1, determine P[ A], the probability that it
will be in state 3 at least once during the next two transitions.
Solution:
(a) Transient states: None
Recurrent states: {1, 2, 3}
Periodic states: None
Chain of recurrent states: 1 chain: {1, 2, 3}
(b) Since the process is an irreducible and aperiodic Markov chain, the limiting-state
probabilities exist and can be obtained as follows. Let π k denote the limiting-state
probability that the process is in state k, k = 1, 2, 3.
π 1 = 14 π 3 ⇒ π 3 = 4π 1
π 2 = 13 π 1 + 101 π 2 + 16 π 3 ⇒ 109 π 2 = 13 π 1 + 64 π 1 = π 1 ⇒ π 2 = 109 π 1
55π 1
1 = π 1 + π 2 + π 3 = π 1 + 109 π 1 + 4π 1 =
9
From this we obtain
9
π1 =
55
10 10 9 10 2
π 2 = π1 = × = =
9 9 55 55 11
36
π 3 = 4π 1 =
55
(c) Given that the process is currently in state 1, the probability P[ A] that it will be in state 3
at least once during the next two transitions is given by

29
 P[ A] = P[(1 → 3 → 2 ) ∪ (1 → 3 → 3) ∪ (1 → 2 → 3)]
= P[1 → 3 → 2] + P[1 → 3 → 3] + P[1 → 2 → 3]
 2 1 2  7  1  9  1 7 3 72 4
=   +   +   = + + = =
 3   6   3   12   3   10  9 18 10 90 5

12.41 Consider the Markov chain shown in Figure 12.24:

2/3 1/ 5
1/ 3 1 2 4 4/5

1 1
3

Figure 12.24: Figure for Problem 12.41

(a) Which states are transient?
(b) Which states are periodic?
(c) Does state 3 have a limiting-state probability? If so, determine this probability.
(d) Assuming that the process begins in state 4, determine the z-transform of the PMF of
K, where K is the number of trials up to and including the trial in which the process
enters state 2 for the second time.
Solution:
(a) Transient states: {4}
(b) Periodic states: None
(c) State 3 is a recurrent state that belongs to the only chain of recurrent states, which is
{1, 2, 3}. Therefore, it has a limiting-state probability, which can be determined as
follows. Let π k denote the limiting-state probability that the process is in state k,
k = 1, 2, 3, 4.
π 1 = 13 π 1 + π 3 ⇒ π 3 = 23 π 1
π 2 = 23 π 1 + 15 π 4
π3 = π2
π 4 = 45 π 4 ⇒ π 4 = 0 ⇒ π 2 = 23 π 1
7π 1
1 = π 1 + π 2 + π 3 + π 4 = π 1 + 23 π 1 + 23 π 1 + 0 =
3
Thus, we have that
3 2 2 3 2
π 1 = , π 2 = π 3 =  π 1 = × =
7 3 3 7 7

30
 (d) Assuming that the process begins in state 4, let X denote the number of trials (or
transitions) up to and including the trial in which the process enters state 2 for the first
time. Then X is a geometrically distributed random variable with success probability
p = 15 and PMF
x −1
x −1 14
p X ( x) = p (1 − p ) =   x = 1, 2, Κ
55
When the process leaves state 2, it takes 2 trials to reach state 1. Given that it has just
entered state 1, let Y denote the number of trials up to and including the trial in which the
process enters state 2 for the first time. Then Y is a geometrically distributed random
variable with success probability q = 23 and PMF
y −1
y −1 2 1
pY ( y ) = q (1 − q ) =   y = 1, 2, Κ
3  3
Thus, K, the number of trials up to and including the trial in which the process enters
state 2 for the second time, given that it started in state 4, is given by K = X + 2 + Y .
Since X and Y are independent random variables, we have that the z-transform of K is
given by
G K ( z ) = E[ z K ] = E[ z X + 2 +Y ] = E[ z 2 ]E[ z X ]E[ z Y ] = z 2 G X ( z )GY ( z )
 z ( 1 )  z ( 2 )  2z 4
= z 2  54  31  =
1 − 5 z 1 − 3 z  (5 − 4 z )(3 − z )

12.42 Find the limiting-state probabilities associated with the following transition probability
matrix:

0.4 0.3 0.3

P = 0.3 0.4 0.3
0.3 0.3 0.4
Solution: Since P is a doubly stochastic matrix, the limiting state probabilities are equal and are
given by π 1 = π 2 = π 3 = 13 .

12.43 Consider the following transition probability matrix:

0.6 0.2 0.2
P = 0.3 0.4 0.3
0.0 0.3 0.7 
(a) Give the state-transition diagram.
(b) Given that the process is currently in state 1, what is the probability that it will be in
state 2 at the end of the third transition?
(c) Given that the process is currently in state 1, what is the probability that the first time
it enters state 3 is the fourth transition?

31
Solution:
(a) The state-transition diagram is as follows:

(b) Given that the process is currently in state 1, the probability that it will be in state 2 at the
end of the third transition, p12 (3), can be obtained as follows:
p12 (3) = P[1 → 1 → 1 → 2] + P[1 → 1 → 2 → 2] + P[1 → 1 → 3 → 2] + P[1 → 3 → 2 → 2]
+ P[1 → 3 → 3 → 2] + P[1 → 2 → 3 → 2] + P[1 → 2 → 2 → 2] + P[1 → 2 → 1 → 2]
= (0.6) 2 (0.2) + (0.6)(0.2)(0.4) + (0.6)(0.2)(0.3) + (0.2)(0.7)(0.3) + (0.2)(0.3)(0.4)
+ (0.2)(0.3) 2 + (0.2)(0.4) 2 + (0.2) 2 (0.3)
= 0.284
Alternative Solution Method: Another way to obtain p12 (3) is that it is the entry on row 1 and
column 2 of the matrix P 3 , which is given by

3
0.6 0.2 0.2 0.330 0.284 0.386
P = 0.3 0.4 0.3 = 0.273 0.301 0.426
3  

0.0 0.3 0.7  0.153 0.324 0.523

(c) Given that the process is currently in state 1, the probability f 13 (4) that the first time it
enters state 3 is the 4th transition is given by

f 13 (4) = P[1 → 1 → 1 → 1 → 3] + P[1 → 1 → 1 → 2 → 3] + P[1 → 1 → 2 → 2 → 3]

+ P[1 → 1 → 2 → 1 → 3] + P[1 → 2 → 1 → 1 → 3] + P[1 → 2 → 2 → 1 → 3]
+ P[1 → 2 → 1 → 2 → 3] + P[1 → 2 → 2 → 2 → 3]
= (0.6) 3 (0.2) + (0.6) 2 (0.2)(0.3) + (0.6)(0.2)(0.4)(0.3) + (0.6)(0.2)(0.3)(0.2)
+ (0.2)(0.3)(0.6)(0.2) + (0.2)(0.4)(0.3)(0.2) + (0.2)(0.3)(0.2)(0.3) + (0.2)(0.4) 2 (0.3)
= 0.1116

32
12.44 Consider the following social mobility problem. Studies indicate that people in a society
can be classified as belonging to the upper class (state 1), middle class (state 2), and
lower class (state 3). Membership in any class is inherited in the following probabilistic
manner. Given that a person is raised in an upper-class family, he will have an upper-
class family with probability 0.45, a middle-class family with probability 0.48, and a
lower-class family with probability 0.07. Given that a person is raised in a middle-class
family, he will have an upper-class family with probability 0.05, a middle-class family
with probability 0.70, and a lower-class family with probability 0.25. Finally, given that a
person is raised in a lower-class family, he will have an upper-class family with
probability 0.01, a middle-class family with probability 0.50, and a lower-class family
with probability 0.49. Determine the following:
(a) The state-transition diagram of the process
(b) The transition probability matrix of the process
(c) The limiting-state probabilities. Interpret what they mean to the layperson.
Solution:
(a) The state-transition diagram of the process is given by the following:

(b) The transition probability matrix of the process is given by

0.45 0.48 0.07 
P = 0.05 0.70 0.25
 0.01 0.50 0.49
(c) The limiting-state probabilities can be obtained as follows. Let π k denote the limiting-
state probability that the process is in state k, k = 1, 2, 3. Then,
π 1 = 0.45π 1 + 0.05π 2 + 0.01π 3 ⇒ 0.55π 1 = 0.05π 2 + 0.01π 3
π 2 = 0.48π 1 + 0.70π 2 + 0.50π 3 ⇒ 0.48π 1 = 0.3π 2 − 0.50π 3
1 = π1 + π 2 + π 3
The solution to the above system of equations is

33
 π 1 = 0.057
π 2 = 0.555
π 3 = 0.388
This result can be interpreted as follows to the layperson. On the long run, 5.7% of the
population will be in the upper class, 55.5% of the population will be in the middle class,
and 38.8% of the population will be in the lower class.

12.45 A taxi driver conducts his business in three different towns 1, 2, and 3. On any given day,
when he is in town 1, the probability that the next passenger he picks up is going to a
place in town 1 is 0.3, the probability that the next passenger he picks up is going to town
2 is 0.2, and the probability that the next passenger he picks up is going to town 3 is 0.5.
When he is in town 2, the probability that the next passenger he picks up is going to town
1 is 0.1, the probability that the next passenger he picks up is going to town 2 is 0.8, and
the probability that the next passenger he picks up is going to town 3 is 0.1. When he is in
town 3, the probability that the next passenger he picks up is going to town 1 is 0.4, the
probability that the next passenger he picks up is going to town 2 is 0.4, and the
probability that the next passenger he picks up is going to town 3 is 0.2.
(a) Determine the state-transition diagram for the process.
(b) Give the transition probability matrix for the process.
(c) What are the limiting-state probabilities?
(d) Given that the taxi driver is currently in town 2 and is waiting to pick up his first
customer for the day, what is the probability that the first time he picks up a pas-
senger to town 2 is when he picks up his third passenger for the day?
(e) Given that he is currently in town 2, what is the probability that his third passenger
from now will be going to town 1?
Solution:
(a) Let state k represent town k, k = 1, 2, 3. Then the state-transition diagram for the process
is as follows:

34
(b) The transition probability matrix for the process is given by
0.3 0.2 0.5
P =  0.1 0.8 0.1
0.4 0.4 0.2
(c) The limiting-state probabilities of the weather can be obtained as follows. Let π k denote
the limiting-state probability that the process is in state k, k = 1, 2, 3. Then,
π 2 + 4π 3
π 1 = 0.3π 1 + 0.1π 2 + 0.4π 3 ⇒ 0.7π 1 = 0.1π 2 + 0.4π 3 ⇒ π 1 =
7
π 2 = 0.2π 1 + 0.8π 2 + 0.4π 3 ⇒ 0.2π 1 = 0.2π 2 − 0.4π 3 ⇒ π 1 = π 2 − 2π 3
1 = π1 + π 2 + π 3
The solution to the above system of equations is
π 1 = 0.2
π 2 = 0.6
π 3 = 0.2
(d) Given that the taxi driver is currently in town 2 and is waiting to pick up his first
customer for the day, the probability that the first time he picks up a passenger to town 2
is when he picks up his third passenger for the day is f 22 (3), which is given by
f 22 (3) = P[2 → 1 → 1 → 2] + P[2 → 1 → 3 → 2] + P[2 → 3 → 3 → 2] + P[2 → 3 → 1 → 2]
= (0.1)(0.3)(0.2) + (0.1)(0.5)(0.4) + (0.1)(0.2)(0.4) + (0.1)(0.4)(0.2)
= 0.042
(e) Given that he is currently in town 2, the probability that his third passenger from now
will be going to town 1 is p 21 (3), which is given by
p 21 (3) = P[2 → 2 → 2 → 1] + P[2 → 2 → 1 → 1] + P[2 → 2 → 3 → 1] + P[2 → 1 → 3 → 1]
+ P[2 → 1 → 1 → 1] + P[2 → 3 → 2 → 1] + P[2 → 3 → 3 → 1] + P[2 → 3 → 1 → 1]
+ P[2 → 1 → 2 → 1]
= (0.8) 2 (0.1) + (0.8)(0.1)(0.3) + (0.8)(0.1)(0.4) + (0.1)(0.5)(0.4) + (0.1)(0.3) 2
+ (0.1)(0.4)(0.1) + (0.1)(0.2)(0.4) + (0.1)(0.4)(0.3) + (0.1)(0.2)(0.1)
= 0.175
Note that p 21 (3) can also be obtained from the entry in the first column of the second
row of the matrix P 3 as follows:
3
0.3 0.2 0.5 0.243 0.506 0.251
P 3 =  0.1 0.8 0.1 = 0.175 0.650 0.175
 
0.4 0.4 0.2 0.232 0.544 0.224

35
12.46 New England fall weather can be classified as sunny, cloudy, or rainy. A student
conducted a detailed study of the weather pattern and came up with the following
conclusion: Given that it is sunny on any given day, then on the following day it will be
sunny again with probability 0.5, cloudy with probability 0.3 and rainy with probability
0.2. Given that it is cloudy on any given day, then on the following day it will be sunny
with probability 0.4, cloudy again with probability 0.3 and rainy with probability 0.3.
Finally, given that it is rainy on any given day, then on the following day it will be sunny
with probability 0.2, cloudy with probability 0.5 and rainy again with probability 0.3.
(a) Give the state-transition diagram of New England fall weather with the state “sunny”
as state 1, the state “cloudy” as state 2, and the state “rainy” as state 3.
(b) Using the same convention as in part (a), give the transition probability matrix of
New England fall weather.
(c) Given that it is sunny today, what is the probability that it will be sunny four days
from now?
(d) Determine the limiting-state probabilities of the weather.
Solution:
(a) The state-transition diagram of the process is given by the following:

(b) The transition probability matrix of the process is given by

0.5 0.3 0.2
P = 0.4 0.3 0.3
0.2 0.5 0.3
(c) Given that it is sunny today (i.e., in state 1), the probability that it will be sunny 4 days
from today is p11 (4), which is obtained from the entry in the first row and first column of
the matrix P 4 , where
4
0.5 0.3 0.2 0.3873 0.3518 0.2609
P = 0.4 0.3 0.3 = 0.3862 0.3524 0.2614
4  
0.2 0.5 0.3 0.3852 0.3528 0.2620
36
 Thus, the required probability is p11 (4) = 0.3873.
(d) The limiting-state probabilities of the weather can be obtained as follows. Let π k denote
the limiting-state probability that the process is in state k, k = 1, 2, 3. Then,
π 1 = 0.5π 1 + 0.4π 2 + 0.2π 3 ⇒ π 1 = 0.8π 2 + 0.4π 3
7 5
π 2 = 0.3π 1 + 0.3π 2 + 0.5π 3 ⇒ π 1 = π 2 − π 3
3 3
1 = π1 + π 2 + π 3
From the above system of equations we obtain the solution
34
π1 = = 0.3863
88
31
π 2 = = 0.3523
88
23
π 3 = = 0.2614
88

12.47 A student went to a gambling casino with $3. He wins $1 at each round with a probability
p and loses $1 with a probability 1 − p. Being a very cautious player, the student has
decided to stop playing when he doubles his original $3 (i.e., when he has a total of $6)
or when he loses all his money.
(a) Give the state-transition diagram of the process.
(b) What is the probability that he stops after being ruined (i.e., he lost all his money)?
(c) What is the probability that he stops after he has doubled his original amount?
Solution: Let state k denote the event that the student currently has a total of $k, k = 0, 1, Κ , 6.
(a) The state-transition diagram of the process is given by the following:

(b) We know that the ruin probability rk for a player that starts with $k is given by
 (q / p) k − (q / p ) N
 p≠q
1 − (q / p) N
rk = 
N − k p=q
 N

where q = 1 − p. Thus, when N = 6 and k = 3 we obtain

37
  (q / p) 3 − (q / p) 6
 p≠q
 1 − (q / p) 6
r3 = 
1 p = q = 12
 2
(c) The probability that he stops after he has doubled his original amount is the probability
that he is not ruined, which is given by r6 = 1 − r3 .

Section 12.8: Continuous-Time Markov Chains

12.48 A small company has two identical servers that are running at the same time. The time
until either server fails is exponentially distributed with a mean of 1 / λ . When a server
fails, a technician starts repairing it immediately. The two servers fail independently of
each other. The time to repair a failed server is exponentially distributed with a mean of
1 / µ . As soon as the repair is completed the server is brought back on line and is assumed
to be as good as new.
(a) Give the state-transition-rate diagram of the process.
(b) What is the fraction of time that both servers are down?
Solution: Let state k denote the number of operational servers.
(a) Since each server fails independently of the other, the failure rate when both servers are
operational (or up) is 2λ . Thus, the state-transition-rate diagram of the process is given
by
2λ λ
2 1 0

µ µ
(b) Let p k denote the limiting-state probability that the process is in state k, k = 0, 1, 2.
Then the fraction of time that both servers are down, p 0 , can be found by using local
balance equations as follows:
µ 1
2λp 2 = µp1 ⇒ p 2 = p1 = p1
2λ 2ρ
µ 1 1
λp1 = µp 0 ⇒ p1 = p 0 = p 0 ⇒ p 2 = p0
λ ρ 2ρ 2
 1 1 
1 = p 0 + p1 + p 2 = p 0 1 + + 2 
 ρ 2ρ 
This gives
1 2ρ 2
p0 = =
1+ ρ +
1 1
2ρ 2
1 + 2ρ + 2ρ 2
where ρ = λ / µ .
38
12.49 Customers arrive at Mike’s barber shop according to a Poisson process with rate λ
customers per hour. Unfortunately Mike, the barber, has only five chairs in his shop for
customers to wait when there is already a customer receiving a haircut. Customers who
arrive when Mike is busy and all the chairs are occupied leave without waiting for a
haircut. Mike is the only barber in the shop, and the time to complete a haircut is
exponentially distributed with a mean of 1 / µ hours.
(a) Give the state-transition-rate diagram of the process.
(b) What is the probability that there are three customers waiting in the shop?
(c) What is the probability that an arriving customer leaves without receiving a haircut?
(d) What is the probability that an arriving customer does not have to wait?
Solution: Let the state k denote the number of chairs that are occupied by customers, including
the chair that the customer who is currently receiving a haircut is sitting on. Thus, k has the
values k = 0, 1, 2, Κ , 6.
(a) The state-transition-rate diagram of the process is as follows:
λ λ λ λ λ λ

µ µ µ µ µ µ
(b) Let p k denote the limiting-state probability that the process is in state k, and let the
parameter ρ = λ / µ . Then from local balance we obtain the following results:
λ
λp 0 = µp1 ⇒ p1 = p = ρp 0
µ 0
λ
λp1 = µp 2 ⇒ p 2 = p1 = ρ 2 p 0
µ
In general it can be shown that p k = ρ k p 0 , k = 0, 1, Κ , 6. Assuming that ρ < 1, then
from the law of total probability we have that
6 6
(
p0 1 − ρ 7 ) 1− ρ
∑ pk = 1 = p0 ∑ ρ k =
k =0 k =0 1− ρ
⇒ p0 =
1− ρ 7
Thus, the probability that there are three customers waiting in the shop is the probability
that the process is in state k = 4 (that is, there are 3 waiting customers and a customer is
currently receiving a haircut), which is given by
ρ 4 (1 − ρ )
p4 = ρ 4 p0 =
1− ρ7
(c) The probability that an arriving customer leaves without receiving a haircut is the
probability that there is no available chair when the customer arrives, which is the
probability that the process is in state k = 6 and is given by

39
 ρ 6 (1 − ρ )
p6 = ρ 6 p0 =
1− ρ 7
(d) The probability that an arriving customer does not have to wait is the probability that the
customer found the place empty, which is the probability that the process is in state k = 0
and is given by
1− ρ
p0 =
1− ρ7

12.50 A small company has two servers: A and B. The time to failure for server A is
exponentially distributed with a mean of 1 / λ A hours, and the time to failure for server B
is exponentially distributed with a mean of 1 / λ B hours. The servers also have different
repair times. The time to repair server A when it fails is exponentially distributed with a
mean if 1 / µ A hours and the time to repair server B when it fails is exponentially
distributed with a mean of 1 / µ B hours. There is only one repair person available to work
on both servers when failure occurs, and each server is considered as good as new after it
has been repaired.
(a) Give the state-transition-rate diagram of the process.
(b) What is the probability that both servers are down?
(c) What is the probability that server A is the first to fail given that both servers have
failed?
(d) What is the probability that both servers are up?
Solution: Let the state of the process be denoted by the pair (a, b), where a = 1 if server A is up
and a = 0 otherwise, and b = 1 if server B is up and b = 0 otherwise. Also, let the state (0 A , 0)
be the state in which both servers are down but server A failed first and was being repaired when
server B failed. Similarly, let the state (0, 0 B ) be the state in which both servers are down but
server B failed first and was being repaired when server A failed.
(a) The state-transition-rate diagram of the process is given by

λB
0, 1 0A, 0
λA
µA
µA
1, 1
λB
µB

µB 1, 0 0, 0 B
λA
40
(b) Let p k denote the limiting-state probability that the process is in state k. If we define
ρ A = λ A / µ A and ρ B = λ B / µ B , then from local balance we have that
λB
µ A p 0 A0 = λ B p 01 ⇒ p 0 A0 = p 01
µA
λ
µ B p 00 B = λ A p10 ⇒ p 00 B = A p10
µB
λB
(λ A + µ B ) p10 = λ B p11 + µ A p0 A0 = λ B p11 + λ B p01 ⇒ p10 = {p +p } (1)
λ A + µ B 11 01

λA
(λ B + µ A ) p01 = λ A p11 + µ B p00 B = λ A p11 + λ A p10 ⇒ p 01 = {p + p10 } ( 2)
λ B + µ A 11

Substituting for p10 in the last equation we obtain

λ A {λ A + λ B + µ B }
p 01 = p (3)
λ A µ A + λ B µ B + µ A µ B 11

Substituting the value obtained for p 01 in equation (1) we obtain

λ B {λ A + λ B + µ A }
p10 = p ( 4)
λ A µ A + λ B µ B + µ A µ B 11

Also we have that

λB ρ λ {λ + λ B + µ B }
p 0 A0 = p 01 = A B A p (5)
µA λ A µ A + λ B µ B + µ A µ B 11
λ ρ λ {λ + λ B + µ A }
p 00 B = A p10 = B A A p11 (6)
µB λ A µ A + λB µ B + µ A µ B
From the law of total probability we obtain

1 = p11 + p10 + p 01 + p 0 A0 + p 00 B
 {λ + ρ A λ B }{λ A + λ B + µ B } + {λ B + ρ B λ A }{λ A + λ B + µ A }
= p11 1 + A 
 λ A µ A + λB µ B + µ A µ B 
1
p11 =
1+ A
{λ + ρ A λ B }{λ A + λ B + µ B } + {λ B + ρ B λ A }{λ A + λ B + µ A }
λ A µ A + λB µ B + µ A µ B
λ A µ A + λB µ B + µ A µ B
=
λ A µ A + λ B µ B + µ A µ B + {λ A + ρ A λ B }{λ A + λ B + µ B } + {λ B + ρ B λ A }{λ A + λ B + µ A }

Thus, the probability that both servers are down is given by

41
  ρ λ {λ + λ B + µ B } ρ λ {λ + λ B + µ A } 
p 0 A0 + p 00 B =  A B A + B A A  p11
 λ A µ A + λB µ B + µ A µ B λ A µ A + λB µ B + µ A µ B 
ρ λ {λ + λ B + µ B } + ρ B λ A {λ A + λ B + µ A }
= A B A p11
λ A µ A + λB µ B + µ A µ B
(c) The probability that PC A is the first to fail, given that both PCs have failed, is the
probability that the process is in state (0 A , 0) and is given by
p 0 A0 ρ A λ B {λ A + λ B + µ B }
=
p 0 A0 + p 00 B ρ A λ B {λ A + λ B + µ B } + ρ B λ A {λ A + λ B + µ A }

(d) The probability that both servers are up is the probability that the process is in state (1, 1)
and is given by
λ A µ A + λB µ B + µ A µ B
p11 =
λ A µ A + λ B µ B + µ A µ B + {λ A + ρ A λ B }{λ A + λ B + µ B } + {λ B + ρ B λ A }{λ A + λ B + µ A }

12.51 Lazy Lou has three identical lightbulbs in his living room that he keeps on all the time.
But because of his laziness Lou does not replace a lightbulb when it fails. (Maybe Lou
does not even notice that the bulb has failed!) However, when all three bulbs have failed,
Lou replaces them at the same time. The lifetime of each bulb is exponentially distributed
with a mean of 1 / λ , and the time to replace all three bulbs is exponentially distributed
with a mean of 1 / µ .
(a) Give the state-transition-rate diagram of the process.
(b) What is the probability that only one lightbulb is working?
(c) What is the probability that all three lightbulbs are working?
Solution: Let the state k denote the number of lightbulbs that have not failed.
(a) The state-transition-rate diagram of the process is given by
3λ 2λ λ

µ
(b) Let p k denote the limiting-state probability that the process is in state k, k = 0, 1, 2, 3.
Then from global balance we obtain
3λ
3λp3 = µp 0 ⇒ p 0 = p3
µ
3
3λp3 = 2λp 2 ⇒ p 2 = p3 = 1.5 p3
2
2λp 2 = λp 1 ⇒ p1 = 2 p 2 = 3 p3

From the law of total probability we have that

42
  3λ 
1 = p3 + p 2 + p1 + p 0 = p3 1 + 1.5 + 3 +  = p3 {5.5 + 3ρ }
 µ

This gives

1
p3 =
5.5 + 3ρ

where ρ = λ / µ. Thus, the probability that only one lightbulb is working is

3
p1 = 3 p3 =
5.5 + 3ρ

(c) The probability that all 3 lightbulbs are working is p3 = 1 /(5.5 + 3ρ ).

12.52 A switchboard has two outgoing lines serving four customers who never call each other.
When a customer is not talking on the phone, he or she generates calls according to a
Poisson process with rate λ calls/minute. Call lengths are exponentially distributed with
a mean of 1 / µ minutes. If a customer finds the switchboard blocked (i.e., both lines are
busy) when attempting to make a call, he or she never tries to make that particular call
again; that is, the call is lost.
(a) Give the state-transition-rate diagram of the process.
(b) What is the fraction of time that the switchboard is blocked?
Solution: Let k denote the state in which k lines are busy, k = 0, 1, 2.
(a) The state-transition-rate diagram of the process is given by
4λ 3λ

µ 2µ

(b) The fraction of time that the switchboard is blocked is the limiting-state probability that
the process is in state 2, which can be obtained as follows. Let p k denote the limiting-
state probability that the process is in state k. If we define ρ = λ / µ , then from local
balance we have that
4λ
4λp 0 = µp1 ⇒ p1 = p 0 = 4 ρp 0
µ
3λ
3λp1 = 2 µp 2 ⇒ p 2 = p1 = 1.5 ρp1 = 6 ρ 2 p 0
2µ

{ }
1 = p 0 + p1 + p 2 = p 0 1 + 4 ρ + 6 ρ 2 ⇒ p 0 =
1
1 + 4ρ + 6ρ 2
43
 Thus, the fraction of time that the switchboard is blocked is
6ρ 2
p 2 = 6 ρ 2 p0 =
1 + 4ρ + 6ρ 2

12.53 A service facility can hold up to six customers who arrive according to a Poisson process
with a rate of λ customers per hour. Customers who arrive when the facility is full are
lost and never make an attempt to return to the facility. Whenever there are two or fewer
customers in the facility, there is only one attendant serving them. The time to service
each customer is exponentially distributed with a mean of 1 / µ hours. Whenever there
are three or more customers, the attendant is joined by a colleague, and the service time is
still the same for each customer. When the number of customers in the facility goes down
to 2, the last attendant to complete service will stop serving. Thus, whenever there are 2
or less customers in the facility, only one attendant can serve.
(a) Give the state-transition-rate diagram of the process.
(b) What is the probability that both attendants are busy attending to customers?
(c) What is the probability that neither attendant is busy?
Solution: Let the state k denote the number of customers in the service facility, where
k = 0, 1, 2, Κ , 6.
(a) The state-transition-rate diagram of the process is given by
λ λ λ λ λ λ

µ µ 2µ
2µ 2µ 2µ

(b) Let p k denote the limiting-state probability that the process is in state k. If we define
ρ = λ / µ , then from local balance we obtain
λ
λp 0 = µp1 ⇒ p1 = p 0 = ρp 0
µ
λ
λp1 = µp 2 ⇒ p 2 = p1 = ρ 2 p 0
µ
λ ρ3
λp 2 = 2µp3 ⇒ p3 = p2 = p0
2µ 2
λ ρ4
λp3 = 2 µp 4 ⇒ p 4 = p3 = p0
2µ 4
λ ρ5
λp 4 = 2µp5 ⇒ p5 = p4 = p0
2µ 8
λ ρ6
λp5 = 2 µp 6 ⇒ p 6 = p5 = p0
2µ 16
44
 From the law of total probability we have that
6
 ρ3 ρ4 ρ5 ρ6 
1 = ∑ p k = p 0 1 + ρ + ρ 2 + + + + 
k =0  2 4 8 16 
1 16
p0 = =
1+ ρ + ρ 2 +
ρ 3
+
ρ 4
+
ρ 5
+
ρ 6 2
{ }
16 1 + ρ + ρ + 8 ρ 3 + 4 ρ 4 + 2 ρ 5 + ρ 6
2 4 8 16

Thus, the probability q that both attendants are busy attending to customers is 1 minus the
probability that at least one attendant is idle, which is given by

q = 1 − {p 0 + p1 + p 2 } = 1 −
16 1 + ρ + ρ 2{ }
{ }
16 1 + ρ + ρ 2 + 8 ρ 3 + 4 ρ 4 + 2 ρ 5 + ρ 6
8ρ + 4 ρ + 2 ρ + ρ
3 4 5 6
=
{ }
16 1 + ρ + ρ 2 + 8 ρ 3 + 4 ρ 4 + 2 ρ 5 + ρ 6

(c) The probability that neither attendant is busy is given by p 0 , which is given above.

12.54 A taxicab company has a small fleet of three taxis that operate from the company’s
station. The time it takes a taxi to take a customer to his or her destination and return to
the station is exponentially distributed with a mean of 1 / µ hours. Customers arrive
according to a Poisson process with average rate of λ customers per hour. If a potential
customer arrives at the station and finds that no taxi is available, he or she goes to another
taxicab company. The taxis always return to the station after dropping off a customer
without picking up any new customers on their way back.
(a) Give the state-transition-rate diagram of the process.
(b) What is the probability that an arriving customer sees exactly one taxi at the station?
(c) What is the probability that an arriving customer goes to another taxicab company?
Solution: Let the state k denote the number of taxis waiting at the station, where k = 0, 1, 2, 3.
(a) The state-transition-rate diagram of the process is given by
λ λ λ

µ 2µ 3µ
(b) Let p k denote the limiting-state probability that the process is in state k. If we define
ρ = λ / µ , then from local balance we obtain

45
 λ
λp3 = µp 2 ⇒ p 2 = p = ρp 3
µ 3
λ ρ2
λp 2 = 2µp1 ⇒ p1 = p2 = p3
2µ 2
λ ρ3
λp1 = 3µp 0 ⇒ p 0 = p1 = p3
3µ 6

From the law of total probability we have that

 ρ2 ρ3 
1 = p3 + p 2 + p1 + p 0 = p 3 1 + ρ + + 
 2 6 
From this we obtain
1 6
p3 = =
ρ 2
ρ 3
6 + 6 ρ + 3ρ 2 + ρ 3
1+ ρ + +
2 6
Thus, the probability that an arriving customer sees exactly one taxi at the station is the
limiting-state probability that the process is in state 1, which is given by
ρ2 ( ρ 2 / 2) 6 3ρ 2
p1 = p3 = =
2 6 + 6 ρ + 3ρ 2 + ρ 3 6 + 6 ρ + 3ρ 2 + ρ 3
(c) The probability that an arriving customer goes to another taxicab company is the
probability that the process is in state 0, which is given by
ρ3 ( ρ 3 / 6) 6 ρ3
p0 = p3 = =
6 6 + 6 ρ + 3ρ 2 + ρ 3 6 + 6 ρ + 3ρ 2 + ρ 3

12.55 Consider a collection of particles that act independently in giving rise to succeeding
generations of particles. Suppose that each particle, from the time it appears, waits a
length of time that is exponentially distributed with a mean of 1 / λ and then either splits
into two identical particles with probability p or disappears with probability 1 − p. Let
X (t ), 0 ≤ t < ∞, denote the number of particles that are present at time t.
(a) Find the birth and death rates of the process.
(b) Give the state-transition-rate diagram of the process.
Solution: Let X (t ) = k denote the state at time t.
(a) When k = 1, the particle splits with rate λp and disappears with rate λ (1 − p ). When
k = n, there are n particles, each of which is acting independently. Therefore, the birth
and death rates of the process are given, respectively, by

bk = kλp k = 1, 2, Κ
d k = kλ (1 − p ) k = 1, 2, Κ
(b) The state-transition-rate diagram of the process is given by
46
 λp 2λ p 3λ p 4λ p 5λp
λ(1− p)
Λ
2λ (1 − p) 3λ(1− p) 4λ(1 − p) 5λ(1− p) 6 λ (1 − p )

47