Chapter 3

Discrete Random
Variables
Section 3.1

Definitions
Discrete Random Variables
• In this chapter and for most of the remainder of this book, we examine probability
models that assign numbers to the outcomes in the sample space.

• When we observe one of these numbers, we refer to the observation as a random

variable.

• In our notation, the name of a random variable is always a capital letter, for example,
X.

• The set of possible values of X is the range of X.

• Since we often consider more than one random variable at a time, we denote the
range of a random variable by the letter S with a subscript that is the name of the
random variable.

• Thus SX is the range of random variable X, SY is the range of random variable Y ,

and so forth.
Example 3.1

The experiment is to attach a photo detector to an optical fiber and count

the number of photons arriving in a one-microsecond time interval. Each
observation is a random variable X. The range of X is SX = {0, 1, 2, . . .}.
In this case, SX , the range of X, and the sample space S is identical.
Example 3.2

The experiment is to test six integrated circuits and after each test ob-
serve whether the circuit is accepted (a) or rejected (r). Each observation
is a sequence of six letters where each letter is either a or r. For example,
s8 = aaraaa. The sample space S consists of the 64 possible sequences.
A random variable related to this experiment is N , the number of ac-
cepted circuits. For outcome s8, N = 5 circuits are accepted. The range
of N is SN = {0, 1, . . . , 6}.
Example 3.3

In Example 3.2, the net revenue R obtained for a batch of six integrated
circuits is $5 for each circuit accepted minus $7 for each circuit rejected.
(This is because for each bad circuit that goes out of the factory, it will
cost the company $7 to deal with the customer’s complaint and supply a
good replacement circuit.) When N circuits are accepted, 6 − N circuits
are rejected so that the net revenue R is related to N by the function

R = g(N ) = 5N − 7(6 − N ) = 12N − 42 dollars. (1)

Since SN = {0, . . . , 6}, the range of R is

SR = {−42, −30, −18, −6, 6, 18, 30} . (2)

The revenue associated with s8 = aaraaa and all other outcomes for
which N = 5 is

g(5) = 12 × 5 − 42 = 18 dollars (3)

Definition 3.2 Discrete Random Variable

X is a discrete random variable if the range of X is a countable set

SX = {x1, x2, . . .} .
Quiz 3.1

A student takes two courses. In each course, the student will earn either
a B or a C. To calculate a grade point average (GPA), a B is worth 3
points and a C is worth 2 points. The student’s GPA G2 is the sum of the
points earned for each course divided by 2. Make a table of the sample
space of the experiment and the corresponding values of the GPA, G2.
Quiz 3.1 Solution

The sample space, probabilities and corresponding grades for the experi-
ment are
Outcomes BB BC CB CC
G2 3.0 2.5 2.5 2.0
Section 3.2

Probability Mass Function

 Probability Mass Function
Definition 3.3 (PMF)

The probability mass function (PMF) of the discrete random variable X

is

PX (x) = P [X = x]
Example 3.5 Problem

When the basketball player Wilt Chamberlain shot two free throws, each
shot was equally likely either to be good (g) or bad (b). Each shot that
was good was worth 1 point. What is the PMF of X, the number of
points that he scored?
Example 3.5 Solution

There are four outcomes of this experiment: gg, gb, bg, and bb. A simple
tree diagram indicates that each outcome has probability 1/4. The sample
space and probabilities of the experiment and the corresponding values
of X are given in the table:
Outcomes bb bg gb gg
P[·] 1/4 1/4 1/4 1/4
X 0 1 1 2
The random variable X has three possible values corresponding to three
events:

{X = 0} = {bb} , {X = 1} = {gb, bg} , {X = 2} = {gg} . (1)

Since each outcome has probability 1/4, these three events have proba-
bilities

P [X = 0] = 1/4, P [X = 1] = 1/2, P [X = 2] = 1/4. (2)

[Continued]
Example 3.5 Solution (Continued 2)

We can express the probabilities of these events in terms of the probability

mass function



 1/4 x = 0,

1/2

x = 1,
PX (x) = (1)


 1/4 x = 2,

0 otherwise.



It is often useful or convenient to depict PX(x) in two other display for-

mats: as a bar plot or as a table.
0.5
PX(x)
x 0 1 2
0
PX (x) 1/4 1/2 1/4
−1 0 1 2 x 3

Each PMF display format has its uses. The function definition (3.8) is
best when PX(x) is given in terms of algebraic functions of x for various
subsets of SX . The bar plot is best for visualizing the probability masses.
The table can be a convenient compact representation when the PMF is
a long list of sample values and corresponding probabilities.
Theorem 3.1

For a discrete random variable X with PMF PX(x) and range SX :

(a) For any x, PX(x) ≥ 0.

P
(b) x∈SX PX(x) = 1.

(c) For any event B ⊂ SX , the probability that X is in the set B is

X
P [B ] = PX (x) .
x∈B
Proof: Theorem 3.1

All three properties are consequences of the axioms of probability (Sec-

tion 1.3). First, PX(x) ≥ 0 since PX(x) = P[X = x]. Next, we observe
that every outcome s ∈ S is associated with a number x ∈ SX . There-
fore, P[x ∈ SX ] = x∈SX PX(x) = P[s ∈ S] = P[S] = 1. Since the events
P

{X = x} and {X = y} are mutually exclusive when x 6= y, B can be writ-

ten as the union of mutually exclusive events B = x∈B {X = x}. Thus
S

we can use Axiom 3 (if B is countably infinite) or Theorem 1.3 (if B is

finite) to write
X X
P [B ] = P [ X = x] = PX (x) . (1)
x∈B x∈B
Quiz 3.2

The random variable N has PMF


c/n n = 1, 2, 3,
PN (n) = (1)
0 otherwise.
Find

(a) The value of the constant c

(b) P[N = 1]

(c) P[N ≥ 2]

(d) P[N > 3]

Quiz 3.2 Solution
(a) To find c, we recall that the PMF must sum to 1. That is,
3
1 1
X  
PN (n) = c 1 + + = 1. (1)
n=1 2 3
This implies c = 6/11. Now that we have found c, the remaining
parts are straightforward.
(b) P[N = 1] = PN(1) = c = 6/11.
(c) P [N ≥ 2] = PN (2) + PN (3)
= c/2 + c/3 = 5/11.
(d) P[N > 3] = ∞
P
n=4 PN(n) = 0.
Section 3.3

Families of Discrete Random

Variables
Families of Random Variables
• In practical applications, certain families of random variables appear over and over
again in many experiments.

• In each family, the probability mass functions of all the random variables have the
same mathematical form.

• They differ only in the values of one or two parameters.

• Depending on the family, the PMF formula contains one or two parameters.

• By assigning numerical values to the parameters, we obtain a specific random

variable.

• Our nomenclature for a family consists of the family name followed by one or two
parameters in parentheses.

• For example, binomial (n, p) refers in general to the family of binomial random
variables.

• Binomial (7, 0.1) refers to the binomial random variable with parameters n = 7 and
p = 0.1.
Example 3.6
Consider the following experiments:

• Flip a coin and let it land on a table. Observe whether the side facing up is heads
or tails. Let X be the number of heads observed.

• Select a student at random and find out her telephone number. Let X = 0 if the
last digit is even. Otherwise, let X = 1.

• Observe one bit transmitted by a modem that is downloading a file from the Inter-
net. Let X be the value of the bit (0 or 1).

All three experiments lead to the probability mass function


1/2 x = 0,

PX (x) = 1/2 x = 1, (1)

0 otherwise.
Definition 3.4 Bernoulli (p) Random Variable

X is a Bernoulli (p) random variable if the PMF of X has the form


1 − p x = 0,


PX (x) = p x = 1,



0 otherwise,
where the parameter p is in the range 0 < p < 1.
Example 3.7 Problem

Test one circuit and observe X, the number of rejects. What is PX(x)
the PMF of random variable X?
Example 3.7 Solution

Because there are only two outcomes in the sample space, X = 1 with
probability p and X = 0 with probability 1 − p,

1 − p x = 0,


PX (x) = p x = 1, (1)



0 otherwise.
Therefore, the number of circuits rejected in one test is a Bernoulli (p)
random variable.
Example 3.8

If there is a 0.2 probability of a reject, the PMF of the Bernoulli (0.2)

random variable is
1

PX(x)
0.8 x = 0,

0.5

PX (x) = 0.2 x = 1, (1)



0 otherwise.
0
−1 0 1 2 x
Example 3.9 Problem

In a sequence of independent tests of integrated circuits, each circuit is

rejected with probability p. Let Y equal the number of tests up to and
including the first test that results in a reject. What is the PMF of Y ?
Example 3.9 Solution

The procedure is to keep testing circuits until a reject appears. Using a

to denote an accepted circuit and r to denote a reject, the tree is

p r •Y =1 p r •Y =2 p r •Y =3

  
  
  
  
 a  a  a ...
1−p 1−p 1−p

From the tree, we see that P[Y = 1] = p, P[Y = 2] = p(1−p), P[Y = 3] =

p(1 − p)2, and, in general, P[Y = y] = p(1 − p)y−1. Therefore,

p(1 − p)y−1 y = 1, 2, . . .
P Y (y ) = (1)
0 otherwise.
Y is referred to as a geometric random variable because the probabilities
in the PMF constitute a geometric series.
Definition 3.5 Geometric (p) Random Variable

X is a geometric (p) random variable if the PMF of X has the form


p(1 − p)x−1 x = 1, 2, . . .
P X ( x) =
0 otherwise.
where the parameter p is in the range 0 < p < 1.
Example 3.10

If there is a 0.2 probability of a reject, the PMF of the geometric (0.2)

random variable is
0.2
PY (y) 
0.1 (0.2)(0.8)y−1 y = 1, 2, . . .
P Y (y ) =
0 otherwise.
0
0 10 20 y
Example 3.11 Problem

In a sequence of n independent tests of integrated circuits, each circuit

is rejected with probability p. Let K equal the number of rejects in the n
tests. Find the PMF PK(k).
Example 3.11 Solution

Adopting the vocabulary of Section 2.3, we call each discovery of a defec-

tive circuit a success, and each test is an independent trial with success
probability p. The event K = k corresponds to k successes in n trials.
We refer to Theorem 2.8 to determine that the PMF of K is
n
P K (k ) = pk (1 − p)n−k . (1)
k
K is an example of a binomial random variable.
 Binomial (n, p) Random
Definition 3.6 Variable

X is a binomial (n, p) random variable if the PMF of X has the form

n
PX (x) = px(1 − p)n−x
x
where 0 < p < 1 and n is an integer such that n ≥ 1.
Example 3.12

If there is a 0.2 probability of a reject and we perform 10 tests, the PMF

of the binomial (10,0.2) random variable is
0.4
PK(k)
0.2 10
P K (k ) = (0.2)k (0.8)10−k . (1)
k
0
0 5 10 k
Example 3.13 Problem

Perform independent tests of integrated circuits in which each circuit is

rejected with probability p. Observe L, the number of tests performed
until there are k rejects. What is the PMF of L?
Example 3.13 Solution
For large values of k, it is not practical to draw the tree. In this case, L = l if and only
if there are k − 1 successes in the first l − 1 trials and there is a success on trial l so that
 

P [L = l] = P k − 1 rejects in l − 1 attempts, reject on attempt l (1)

| {z } | {z }
A B

The events A and B are independent since the outcome of attempt l is not affected by
the previous l − 1 attempts. Note that P[A] is the binomial probability of k − 1 successes
(i.e., rejects) in l − 1 trials so that
l − 1
P [A] = pk−1 (1 − p)l−1−(k−1)
. (2)
k−1
Finally, since P[B] = p,
l − 1
PL (l) = P [A] P [B] = pk (1 − p)l−k (3)
k−1
L is an example of a Pascal random variable.
Definition 3.7 Pascal (k, p) Random Variable

X is a Pascal (k, p) random variable if the PMF of X has the form

x − 1
PX (x) = pk (1 − p)x−k
k−1
where 0 < p < 1 and k is an integer such that k ≥ 1.
Example 3.14

If there is a 0.2 probability of a reject and we seek four defective circuits,

the random variable L is the number of tests necessary to find the four
circuits. The PMF of the Pascal(4,0.2) random variable is
0.1
PL(l)
0.05 l − 1
PL (l) = (0.2)4(0.8)l−4.
3

0
0 20 40 l
Example 3.15

In an experiment with equiprobable outcomes, the random variable N

has the range SN = {k, k + 1, k + 2, · · · , l}, where k and l are integers
with k < l. The range contains l − k + 1 numbers, each with probability
1/(l − k + 1). Therefore, the PMF of N is

1/(l − k + 1) n = k, k + 1, k + 2, . . . , l
PN (n) = (1)
0 otherwise
N is an example of a discrete uniform random variable.
 Discrete Uniform (k, l) Random
Definition 3.8 Variable

X is a discrete uniform (k, l) random variable if the PMF of X has the

form

1/(l − k + 1) x = k, k + 1, k + 2, . . . , l
PX (x) =
0 otherwise
where the parameters k and l are integers such that k < l.
Example 3.16

Roll a fair die. The random variable N is the number of spots on the side
facing up. Therefore, N is a discrete uniform (1, 6) random variable with
PMF
0.2

PN(n) 1/6 n = 1, 2, 3, 4, 5, 6,
0.1 PN (n) =
0 otherwise.
(1)
0
0 5 n
Definition 3.9 Poisson (α) Random Variable

X is a Poisson (α) random variable if the PMF of X has the form


αxe−α/x! x = 0, 1, 2, . . . ,
PX (x) =
0 otherwise,
where the parameter α is in the range α > 0.

Poisson models are often used to model occurrence (called arrival) of phenomenon of
interest. In these applications, we specify an average rate: λ arrivals per second and a time
interval T. In this time interval, α = λT
Example 3.17 Problem

The number of hits at a website in any time interval is a Poisson random

variable. A particular site has on average λ = 2 hits per second. What
is the probability that there are no hits in an interval of 0.25 seconds?
What is the probability that there are no more than two hits in an interval
of one second?
Example 3.17 Solution
In an interval of 0.25 seconds, the number of hits H is a Poisson random variable with
α = λT = (2 hits/s) × (0.25 s) = 0.5 hits. The PMF of H is
1
PH(h)  h −0.5
0.5 0.5 e /h! h = 0, 1, 2, . . .
PH (h) =
0 otherwise.

0
0 2 4 h
The probability of no hits is
P [H = 0] = PH (0) = (0.5)0 e−0.5 /0! = 0.607. (1)
In an interval of 1 second, α = λT = (2 hits/s) × (1 s) = 2 hits. Letting J denote the
number of hits in one second, the PMF of J is

PJ(j) 0.2
2j e−2 /j!

j = 0, 1, 2, . . .
0.1 PJ (j) =
0 otherwise.
0
0 2 4 6 8 j
[Continued]
Example 3.17 Solution (Continued 2)

To find the probability of no more than two hits, we note that

{J ≤ 2} = {J = 0} ∪ {J = 1} ∪ {J = 2} (1)
is the union of three mutually exclusive events. Therefore,

P [J ≤ 2] = P [J = 0] + P [J = 1] + P [J = 2]
= PJ (0) + PJ (1) + PJ (2)
= e−2 + 21e−2/1! + 22e−2/2! = 0.677. (2)
Example 3.18 Problem

The number of database queries processed by a computer in any 10-

second interval is a Poisson random variable, K, with α = 5 queries.
What is the probability that there will be no queries processed in a 10-
second interval? What is the probability that at least two queries will be
processed in a 2-second interval?
Example 3.18 Solution

The PMF of K is
0.2
PK(k) 
0.1 5k e−5/k! k = 0, 1, 2, . . .
PK (k) =
0 otherwise.
0
0 5 10 15 k
Therefore, P[K = 0] = PK(0) = e−5 = 0.0067. To answer the question
about the 2-second interval, we note in the problem definition that α = 5
queries = λT with T = 10 seconds. Therefore, λ = 0.5 queries per
second. If N is the number of queries processed in a 2-second interval,
α = 2λ = 1 and N is the Poisson (1) random variable with PMF

e−1/n! n = 0, 1, 2, . . .
PN (n) = (1)
0 otherwise.
Therefore,

P [N ≥ 2] = 1 − PN (0) − PN (1) = 1 − e−1 − e−1 = 0.264. (2)

Example 3.19

Calls arrive at random times at a telephone switching office with an

average of λ = 0.25 calls/second. The PMF of the number of calls that
arrive in a T = 2-second interval is the Poisson (0.5) random variable
with PMF
1
PJ(j) 
0.5 (0.5)j e−0.5/j! j = 0, 1, . . . ,
PJ (j ) =
0 otherwise.
0
0 2 4 j
Note that we obtain the same PMF if we define the arrival rate as λ =
60 · 0.25 = 15 calls per minute and derive the PMF of the number of calls
that arrive in 2/60 = 1/30 minutes.
Example 3.20

Calls arrive at random times at a telephone switching office with an

average of λ = 0.25 calls per second. The PMF of the number of calls
that arrive in any T = 20-second interval is the Poisson (5) random
variable with PMF
0.2
PJ(j) 
0.1 5j e−5/j! j = 0, 1, . . . ,
P J (j ) =
0 otherwise.

0
0 5 10 15 j
Quiz 3.3
Each time a modem transmits one bit, the receiving modem analyzes the
signal that arrives and decides whether the transmitted bit is 0 or 1. It
makes an error with probability p, independent of whether any other bit
is received correctly.
(a) If the transmission continues until the receiving modem makes its first
error, what is the PMF of X, the number of bits transmitted?
(b) If p = 0.1, what is the probability that X = 10? What is the probability
that X ≥ 10?
(c) If the modem transmits 100 bits, what is the PMF of Y , the number
of errors?
(d) If p = 0.01 and the modem transmits 100 bits, what is the probability
of Y = 2 errors at the receiver? What is the probability that Y ≤ 2?
(e) If the transmission continues until the receiving modem makes three
errors, what is the PMF of Z, the number of bits transmitted?
(f) If p = 0.25, what is the probability of Z = 12 bits transmitted until
the modem makes three errors?
Quiz 3.3 Solution
Decoding each transmitted bit is an independent trial where we call a bit error a “suc-
cess.” Each bit is in error, that is, the trial is a success, with probability p. Now we can
interpret each experiment in the generic context of independent trials.
(a) The random variable X is the number of trials up to and including the first success.
Similar to Example 3.9, X has the geometric PMF
p(1 − p)x−1 x = 1, 2, . . .

PX(x) =
0 otherwise.
(b) If p = 0.1, then the probability exactly 10 bits are sent is
PX (10) = (0.1)(0.9)9 = 0.0387. (1)
The probability that at least 10 bits are sent is
∞
X
P [X ≥ 10] = PX (x) . (2)
x=10
This sum is not too hard to calculate. However, its even easier to observe that
X ≥ 10 if the first 10 bits are transmitted correctly. That is,
P [X ≥ 10] = P [first 10 bits correct] = (1 − p)10 . (3)
For p = 0.1,
P [X ≥ 10] = 0.910 = 0.3487. (4)
(c) The random variable Y is the number of successes in 100 independent trials. Just
as in Example 3.11, Y has the binomial PMF
100
PY (y) = py (1 − p)100−y . (5)
y
Quiz 3.3 Solution (Continued 2)
If p = 0.01, the probability of exactly 2 errors is
100
PY (2) = (0.01)2 (0.99)98 = 0.1849. (6)
2
(d) The probability of no more than 2 errors is
P [Y ≤ 2] = PY (0) + PY (1) + PY (2)
100
100 99
= (0.99) + 100(0.01)(0.99) + (0.01)2 (0.99)98
2
= 0.9207. (7)
(e) Random variable Z is the number of trials up to and including the third success.
Thus Z has the Pascal PMF (see Example 3.13)
z − 1
PZ (z) = p3 (1 − p)z−3 . (8)
2
Note that PZ(z) > 0 for z = 3, 4, 5, . . ..
(f) If p = 0.25, the probability that the third error occurs on bit 12 is
11
PZ (12) = (0.25)3 (0.75)9 = 0.0645. (9)
2
Section 3.4

Cumulative Distribution Function

(CDF)
 Cumulative Distribution
Definition 3.10 Function (CDF)

The cumulative distribution function (CDF) of random variable X is

FX (x) = P [X ≤ x] .

PMF: PX(x)=P[X=x]
Theorem 3.2

For any discrete random variable X with range SX = {x1, x2, . . .} satisfying
x1 ≤ x2 ≤ . . .,

(a) FX(−∞) = 0 and FX(∞) = 1.

(b) For all x0 ≥ x, FX(x0) ≥ FX(x).

(c) For xi ∈ SX and , an arbitrarily small positive number,

FX (xi) − FX (xi − ) = PX (xi) .

(d) FX(x) = FX(xi) for all x such that xi ≤ x < xi+1.

3.4 Comment: Theorem 3.2

Each property of Theorem 3.2 has an equivalent statement in words:

(a) Going from left to right on the x-axis, FX(x) starts at zero and ends
at one.

(b) The CDF never decreases as it goes from left to right.

(c) For a discrete random variable X, there is a jump (discontinuity) at

each value of xi ∈ SX . The height of the jump at xi is PX(xi).

(d) Between jumps, the graph of the CDF of the discrete random variable
X is a horizontal line.
Theorem 3.3

For all b ≥ a,

FX (b) − FX (a) = P [a < X ≤ b] .

FX(b)=P[X≤ b]=P[X≤ a or a < X ≤ b ]=P[X≤ a]+P[a < X ≤ b ]

FX(a)=P[X ≤ a ]
Proof: Theorem 3.3

To prove this theorem, express the event Eab = {a < X ≤ b} as a part of

a union of mutually exclusive events. Start with the event Eb = {X ≤ b}.
Note that Eb can be written as the union

Eb = {X ≤ b} = {X ≤ a} ∪ {a < X ≤ b} = Ea ∪ Eab (1)

Note also that Ea and Eab are mutually exclusive so that P[Eb] = P[Ea] +
P[Eab]. Since P[Eb] = FX(b) and P[Ea] = FX(a), we can write FX(b) =
FX(a) + P[a < X ≤ b]. Therefore, P[a < X ≤ b] = FX(b) − FX(a).
Example 3.21 Problem

In Example 3.5, random variable X has PMF

0.5 
 1/4 x = 0,
PX(x) 


1/2

x = 1,
PX (x) = (1)


 1/4 x = 2,

0 0

otherwise.
−1 0 1 2 3 x
Find and sketch the CDF of random variable X.
Example 3.21 Solution

Referring to the PMF PX(x), we derive the CDF of random variable X:

1 
 0 x < 0,
FX(x) 


0.5 1/4 0 ≤ x < 1,

FX (x) = P [X ≤ x] =


 3/4 1≤x<2

0 1

x ≥ 2.
−1 0 1 2 3 x
Keep in mind that at the discontinuities x = 0, x = 1 and x = 2, the
values of FX(x) are the upper values: FX(0) = 1/4, FX(1) = 3/4 and
FX(2) = 1. Math texts call this the right hand limit of FX(x).
Example 3.22 Problem

In Example 3.9, let the probability that a circuit is rejected equal p = 1/4.
The PMF of Y , the number of tests up to and including the first reject,
is the geometric (1/4) random variable with PMF

(1/4)(3/4)y−1 y = 1, 2, . . .
PY (y ) = (1)
0 otherwise.
What is the CDF of Y ?
Example 3.22 Solution

Random variable Y has nonzero probabilities for all positive integers. For
any integer n ≥ 1, the CDF is
n n
1 3 j−1
X X  
FY (n) = P Y (j ) = . (1)
j=1 j=1 4 4

Equation (3.30) is a geometric series. Familiarity with the geometric

series is essential for calculating probabilities involving geometric random
variables. Appendix B summarizes the most important facts. In partic-
Pn
ular, Math Fact B.4 implies (1 − x) j=1 xj−1 = 1 − xn. Substituting
x = 3/4, we obtain
 n
3
FY (n) = 1 − . (2)
4
The complete expression for the CDF of Y must show FY (y) for all integer
and noninteger values of y. [Continued]
Example 3.22 Solution (Continued 2)

For an integer-valued random variable Y , we can do this in a simple way

using the floor function byc, which is the largest integer less than or equal to
y. In particular, if n ≤ y < n + 1 for some integer n, then byc = n and

FY (y ) = P [Y ≤ y ] = P [Y ≤ n] = FY (n) = FY (byc) . (1)

In terms of the floor function, we can express the CDF of Y as
1
FY (y) 
0.5 0 y < 1,
F Y (y ) = (2)
1 − (3/4)byc y ≥ 1.
0
0 5 10 y
To find the probability that Y takes a value in the set {4, 5, 6, 7, 8}, we
refer to Theorem 3.3 and compute

P [3 < Y ≤ 8] = FY (8) − FY (3) = (3/4)3 − (3/4)8 = 0.322. (3)

Quiz 3.4

Use the CDF FY (y) to find the following probabilities:

(a) P[Y < 1]
1
0.8 (b) P[Y ≤ 1]
FY (y) 0.6 (c) P[Y > 2]
0.4
(d) P[Y ≥ 2]
0.2
0 (e) P[Y = 1]
0 1 2 3 4 5 y (f) P[Y = 3]
Quiz 3.4 Solution
Each of these probabilities can be read from the graph of the CDF FY (y).
However, we must keep in mind that when FY (y) has a discontinuity at
y0, FY (y) takes the upper value FY (y0+).
(a) P[Y < 1] = FY (1−) = 0.
(b) P[Y ≤ 1] = FY (1) = 0.6.
(c) P[Y > 2] = 1 − P[Y ≤ 2] = 1 − FY (2) = 1 − 0.8 = 0.2.
(d) P[Y ≥ 2] = 1 − P[Y < 2] = 1 − FY (2−) = 1 − 0.6 = 0.4.
(e) P[Y = 1] = P[Y ≤ 1] − P[Y < 1] = FY (1+) − FY (1−) = 0.6.
(f) P[Y = 3] = P[Y ≤ 3] − P[Y < 3] = FY (3+) − FY (3−) = 0.8 − 0.8 = 0.
Section 3.5

Averages and Expected Value

Definition 3.13 Expected Value

The expected value of X is

X
E [X ] = µX = xPX (x) .
x∈SX
Example 3.24 Problem

Random variable X in Example 3.5 has PMF

0.5 
 1/4 x = 0,
PX(x) 


1/2

x = 1,
PX (x) = (1)


 1/4 x = 2,

0 
0 otherwise.
−1 0 1 2 3 x
What is E[X]?
Example 3.24 Solution

E [X ] = µX = 0 · PX (0) + 1 · PX (1) + 2 · PX (2)

= 0(1/4) + 1(1/2) + 2(1/4) = 1. (1)
Theorem 3.4

The Bernoulli (p) random variable X has expected value E[X] = p.

Proof: Theorem 3.4

E[X] = 0 · PX(0) + 1PX(1) = 0(1 − p) + 1(p) = p.

Theorem 3.5

The geometric (p) random variable X has expected value E[X] = 1/p.
Proof: Theorem 3.5

Let q = 1 − p. The PMF of X becomes


pq x−1 x = 1, 2, . . .
PX (x) = (1)
0 otherwise.
The expected value E[X] is the infinite sum
∞ ∞
xpq x−1
X X
E [X ] = xPX (x) = . (2)
x=1 x=1
Applying the identity of Math Fact B.7, we have
∞ ∞
p p q p 1
xq x−1 = xq x =
X X
E [X ] = p 2
= 2
= . (3)
x=1 q x=1 q1 − q p p
Theorem 3.6

The Poisson (α) random variable in Definition 3.9 has expected value
E[X] = α.
Proof: Theorem 3.6
∞ ∞
X X αx −α
E [X] = xPX (x) = x e . (1)
x=0 x=0
x!

We observe that x/x! = 1/(x − 1)! and also that the x = 0 term in the sum is zero. In
addition, we substitute αx = α · αx−1 to factor α from the sum to obtain
∞
X αx−1 −α
E [X] = α e . (2)
x=1
(x − 1)!
Next we substitute l = x − 1, with the result
∞
X αl
E [X] = α e−α = α. (3)
l!
l=0
| {z }
1

We can
P∞conclude that the sum in this formula equals 1 either by referring to the identity
α l
e = l=0 α /l! or by applying Theorem 3.1(b) to the fact that the sum is the sum of
the PMF of a Poisson random variable L over all values in SL and P[SL ] = 1.
Theorem 3.7

(a) For the binomial (n, p) random variable X of Definition 3.6,

E [X ] = np.

(b) For the Pascal (k, p) random variable X of Definition 3.7,

E [X ] = k/p.

(c) For the discrete uniform (k, l) random variable X of Definition 3.8,

E [X ] = (k + l)/2.
Quiz 3.5

In a pay-as-you go cellphone plan, the cost of sending an SMS text

message is 10 cents and the cost of receiving a text is 5 cents. For
a certain subscriber, the probability of sending a text is 1/3 and the
probability of receiving a text is 2/3. Let C equal the cost (in cents) of
one text message and find

(a) The PMF PC(c)

(b) The expected value E[C]

(c) The probability that the subscriber receives four texts before sending
a text.

(d) The expected number of texts received by the subscriber before the
subscriber sends a text.
Quiz 3.5 Solution
(a) With probability 1/3, the subscriber sends a text and the cost is C = 10 cents.
Otherwise, with probability 2/3, the subscriber receives a text and the cost is
C = 5 cents. This corresponds to the PMF

2/3 c = 5,

PC (c) = 1/3 c = 10, (1)

0 otherwise.
(b) The expected value of C is
E [C] = (2/3)(5) + (1/3)(10) = 6.67 cents. (2)
(c) For the next two parts we think of each text as a Bernoulli trial such that the trial
is a “success” if the subscriber sends a text. The success probability is p = 1/3.
Let R denote the number of texts received before sending a text. In terms of
Bernoulli trials, R is the number of failures before the first success. R is similar to
a geometric random variable except R = 0 is possible if the first text is sent rather
than received. In general R = r if the first r trials are failures (i.e. the first r texts
are received) and trial r + 1 is a success. Thus R has PMF

(1 − p)r p r = 0, 1, 2 . . .

PR (r) = (3)
0 otherwise.
[Continued]
Quiz 3.5 Solution (Continued 2)
The probability of receiving four texts before sending a text is
PR (4) = (1 − p)4 p. (4)
(d) The expected number of texts received before sending a text is
∞
X ∞
X
E [R] = rPR (r) = r(1 − p)r p. (5)
r=0 r=0

Letting q = 1 − p and observing that the r = 0 term in the sum is zero,

∞
X
E [R] = p rq r . (6)
r=1

Using Math Fact B.7, we have

q 1−p
E [R] = p = = 2. (7)
(1 − q)2 p
Section 3.6

Functions of a Random Variable

Definition 3.14 Derived Random Variable

Each sample value y of a derived random variable Y is a mathematical

function g(x) of a sample value x of another random variable X. We
adopt the notation Y = g(X) to describe the relationship of the two
random variables.
Example 3.25 Problem

A parcel shipping company offers a charging plan: $1.00 for the first
pound, $0.90 for the second pound, etc., down to $0.60 for the fifth
pound, with rounding up for a fraction of a pound. For all packages
between 6 and 10 pounds, the shipper will charge $5.00 per package. (It
will not accept shipments over 10 pounds.) Find a function Y = g(X)
for the charge in cents for sending one package.
Example 3.25 Solution

When the package weight is an integer X ∈ {1, 2, . . . , 10} that specifies

the number of pounds with rounding up for a fraction of a pound, the
function

105X − 5X 2 X = 1, 2, 3, 4, 5
Y = g(X) = (1)
500 X = 6, 7, 8, 9, 10.
corresponds to the charging plan.
Theorem 3.9

For a discrete random variable X, the PMF of Y = g(X) is

X
P Y (y ) = PX (x) .
x:g(x)=y
Figure 3.1

X=1
 Y =100
.15



.15

 
X=2
 Y =190
 


 .15



X=3
 Y =270

.15 X=4

 
X



H X
ZHXXX
@ Y =340
ZHH XXX
@Z H XXX
@Z HH
@Z
.10
H
X
X=5 Y =400
@ZZ HHH
.10
@ Z H
X=6XXXXX
.10
@ Z XX
@ ZZ XXX

.10@ Z X=7  
X
X

 Y =500
@
 
X=8
@@  

The derived random variable Y = g(X) for Example 3.27.

Example 3.26 Problem

In Example 3.25, suppose all packages weigh 1, 2, 3, or 4 pounds with

equal probability. Find the PMF and expected value of Y , the shipping
charge for a package.
Example 3.26 Solution

From the problem statement, the weight X has PMF


1/4 x = 1, 2, 3, 4,
P X ( x) = (1)
0 otherwise.
The charge for a shipment, Y , has range SY = {100, 190, 270, 340} cor-
responding to SX = {1, . . . , 4}. The experiment can be described by the
following tree. Here each value of Y derives from a unique value of X.
Hence, we can use Equation (3.55) to find PY (y).

1/4 y = 100, 190, 270, 340,
1/4 
X=1•Y =100
 PY (y ) =
 0 otherwise.

1/4

  X=2•Y =190


The expected shipping bill is




X=3•Y =270
 


X
1/4
XXX
XXX
XXX 1
1/4
XXX
X=4•Y =340
X E [Y ] = (100 + 190 + 270 + 340)
4
= 225 cents.
Example 3.27 Problem

Suppose the probability model for the weight in pounds X of a package

in Example 3.25 is
0.2

PX(x)
0.15 x = 1, 2, 3, 4,

0.1

P X ( x) = 0.1 x = 5, 6, 7, 8,



0 otherwise.
0
0 5 10 x
For the pricing plan given in Example 3.25, what is the PMF and expected
value of Y , the cost of shipping a package?
Example 3.27 Solution

Now we have three values of X, specifically (6, 7, 8), transformed by g(·)

into Y = 500. For this situation we need the more general view of the
PMF of Y , given by Theorem 3.9. In particular, y6 = 500, and we have
to add the probabilities of the outcomes X = 6, X = 7, and X = 8 to
find PY (500). That is,

PY (500) = PX (6) + PX (7) + PX (8) = 0.30. (1)

The steps in the procedure are illustrated in the diagram of Figure 3.1.
Applying Theorem 3.9, we have

 0.15 y = 100, 190, 270, 340,
PY (y) 0.2



0.10

y = 400,
PY (y ) =


 0.30 y = 500,
0

0

otherwise.
100 270 500 y
For this probability model, the expected cost of shipping a package is

E [Y ] = 0.15(100 + 190 + 270 + 340) + 0.10(400) + 0.30(500) = 325 cents.

Example 3.28 Problem

The amplitude V (volts) of a sinusoidal signal is a random variable with

PMF
0.2
PV (v) 
0.1 1/7 v = −3, −2, . . . , 3,
PV (v ) =
0 otherwise.
0
−5 0 5 v
Let Y = V 2/2 watts denote the power of the transmitted signal. Find
PY (y).
Example 3.28 Solution

The possible values of Y are SY = {0, 0.5, 2, 4.5}. Since Y = y when V =

√ √
2y or V = − 2y, we see that PY (0) = PV (0) = 1/7. For y = 0.5, 2, 4.5,
√ √
PY (y) = PV ( 2y) + PV (− 2y) = 2/7. Therefore,


PY (y) 0.2 1/7 y = 0,


P Y (y ) = 2/7 y = 0.5, 2, 4.5, (1)


0

0 otherwise.
0 1 2 3 4 5 y
Quiz 3.6

Monitor three customers purchasing smartphones at the Phonesmart

store and observe whether each buys an Apricot phone for $450 or a
Banana phone for $300. The random variable N is the number of cus-
tomers purchasing an Apricot phone. Assume N has PMF

0.4 n = 0,


PN (n) = 0.2 n = 1, 2, 3, (1)



0 otherwise.
M dollars is the amount of money paid by three customers.

(a) Express M as a function of N .

(b) Find PM(m) and E[M ].

Quiz 3.6 Solution
(a) As a function of N , the money spent by the tree customers is
M = 450N + 300(3 − N ) = 900 + 150N.
(b) To find the PMF of M , we can draw the following tree and map the outcomes to
values of M :

 N =0 •M =900
0.4

  N =1 •M =1050
 0.2
 
  



 
XXX
H
HHXXX
H XXX
HH XX
H 0.2 XXX N =2 •M =1200
HH
H
0.2HHH
H N =3 •M =1350

From this tree,


0.4
 m = 900,
PM (m) = 0.2 m = 1050, 1200, 1350 (1)

0 otherwise.
From the PMF PM(m), the expected value of M is
E [M ] = 900PM (900) + 1050PM (1050)
+ 1200PM (1200) + 1350PM (1350) (2)
= (900)(0.4) + (1050 + 1200 + 1350)(0.2) = 1080. (3)
Section 3.7

Expected Value of a Derived

Random Variable
Theorem 3.10

Given a random variable X with PMF PX(x) and the derived random
variable Y = g(X), the expected value of Y is
X
E [Y ] = µY = g(x)PX (x) .
x∈SX
Proof: Theorem 3.10

From the definition of E[Y ] and Theorem 3.9, we can write

X X X X X
E [Y ] = yPY (y ) = y P X ( x) = g(x)PX (x) ,
y∈SY y∈SY x:g(x)=y y∈SY x:g(x)=y
(1)
where the last double summation follows because g(x) = y for each x in
the inner sum. Since g(x) transforms each possible outcome x ∈ SX to a
value y ∈ SY , the preceding double summation can be written as a single
sum over all possible values x ∈ SX . That is,
X
E [Y ] = g(x)PX (x) . (2)
x∈SX
Example 3.29 Problem

In Example 3.26,

1/4 x = 1, 2, 3, 4,
PX (x) =
0 otherwise,

105X − 5X 2 1 ≤ X ≤ 5,
Y = g(X) =
500 6 ≤ X ≤ 10.
What is E[Y ]?
Example 3.29 Solution

Applying Theorem 3.10 we have

4
X
E [Y ] = PX (x) g(x)
x=1
= (1/4)[(105)(1) − (5)(1)2] + (1/4)[(105)(2) − (5)(2)2]
+ (1/4)[(105)(3) − (5)(3)2] + (1/4)[(105)(4) − (5)(4)2]
= (1/4)[100 + 190 + 270 + 340] = 225 cents. (1)
Theorem 3.11

For any random variable X,

E [X − µX ] = 0.
Proof: Theorem 3.11

Defining g(X) = X − µX and applying Theorem 3.10 yields

X X X
E [g(X)] = (x − µX )PX (x) = xPX (x) − µX P X ( x) . (1)
x∈SX x∈SX x∈SX
The first term on the right side is µX by definition. In the second term,
P
x∈SX PX(x) = 1, so both terms on the right side are µX and the differ-
ence is zero.
Theorem 3.12

For any random variable X,

E [aX + b] = a E [X ] + b.
Example 3.30 Problem

Recall from Examples 3.5 and 3.24 that X has PMF

0.5 
 1/4 x = 0,
PX(x) 


1/2

x = 1,
PX (x) = (1)


 1/4 x = 2,

0 0

otherwise.
−1 0 1 2 3 x
What is the expected value of V = g(X) = 4X + 7?
Example 3.30 Solution

From Theorem 3.12,

E [V ] = E [g(X)] = E [4X + 7] = 4 E [X ] + 7 = 4(1) + 7 = 11. (1)

We can verify this result by applying Theorem 3.10:

E [V ] = g(0)PX (0) + g(1)PX (1) + g(2)PX (2)

= 7(1/4) + 11(1/2) + 15(1/4) = 11. (2)
Example 3.31 Problem

Continuing Example 3.30, let W = h(X) = X 2. What is E[W ]?

Example 3.31 Solution

Theorem 3.10 gives

h(x)PX (x) = (1/4)02 + (1/2)12 + (1/4)22 = 1.5.

X
E [W ] = (1)
Note that this is not the same as h(E[W ]) = (1)2 = 1.
Quiz 3.7
The number of memory chips M needed in a personal computer depends
on how many application programs, A, the owner wants to run simulta-
neously. The number of chips M and the number of application programs
A are described by



4 chips for 1 program, 

4

chips for 2 programs, 0.1(5 − a) a = 1, 2, 3, 4,
M = PA (a) =


6 chips for 3 programs, 0 otherwise.

8

chips for 4 programs,
(1)
(a) What is the expected number of programs µA = E[A]?
(b) Express M , the number of memory chips, as a function M = g(A) of
the number of application programs A.
(c) Find E[M ] = E[g(A)]. Does E[M ] = g(E[A])?
Quiz 3.7 Solution
(a) Using Definition 3.13, the expected number of applications is
4
X
E [A] = aPA (a)
a=1
= 1(0.4) + 2(0.3) + 3(0.2) + 4(0.1)
= 2. (1)
(b) The number of memory chips is

4
 A = 1, 2,
M = g(A) = 6 A = 3, (2)

8 A = 4.
(c) By Theorem 3.10, the expected number of memory chips is
4
X
E [M ] = g(A)PA (a)
a=1
= 4(0.4) + 4(0.3) + 6(0.2) + 8(0.1)
= 4.8. (3)
Since E[A] = 2,
g(E[A]) = g(2) = 4.
However, E[M ] = 4.8 6= g(E[A]). The two quantities are different because g(A) is
not of the form αA + β.
Section 3.8

Variance and Standard Deviation

Definition 3.15 Variance

The variance of random variable X is

h i
2
Var[X] = E (X − µX ) .
Definition 3.16 Standard Deviation

The standard deviation of random variable X is

q
σX = Var [X ].
Theorem 3.14

h i h i
2
Var [X ] = E X − µX = E X − (E [X ])2 .
2 2
Proof: Theorem 3.14

Expanding the square in (3.75), we have

x2PX(x) − µ2
X X X
Var[X] = 2µX xPX(x) + X PX(x)
x∈SX x∈SX x∈SX
= E[X 2] − 2µX xPX(x) + µ2
X X
X PX(x)
x∈SX x∈SX
= E[X 2] − 2µ2
X + µ 2.
X (1)
Definition 3.17 Moments

For random variable X:

(a) The nth moment is E[X n].

(b) The nth central moment is E[(X − µX )n].

Example 3.32 Problem

Continuing Examples 3.5, 3.24, and 3.30, we recall that X has PMF
0.5 
 1/4 x = 0,
PX(x) 


1/2

x = 1,
PX (x) = (1)


 1/4 x = 2,

0 0

otherwise,
−1 0 1 2 3 x
and expected value E[X] = 1. What is the variance of X?
Example 3.32 Solution
In order of increasing simplicity, we present three ways to compute Var[X].
• From Definition 3.15, define
W = (X − µX )2 = (X − 1)2 . (1)
We observe that W = 0 if and only if X = 1; otherwise, if X = 0 or X = 2, then
W = 1. Thus P[W = 0] = PX(1) = 1/2 and P[W = 1] = PX(0) + PX(2) = 1/2. The
PMF of W is

1/2 w = 0, 1,
PW (w) = (2)
0 otherwise.
Then
Var [X] = E [W ] = (1/2)(0) + (1/2)(1) = 1/2. (3)
• Recall that Theorem 3.10 produces the same result without requiring the derivation
of PW(w).
Var[X] = E (X − µX )2
 

= (0 − 1)2 PX (0) + (1 − 1)2 PX (1) + (2 − 1)2 PX (2)

= 1/2. (4)
• To apply Theorem 3.14, we find that
E X 2 = 02 PX (0) + 12 PX (1) + 22 PX (2) = 1.5.
 
(5)
Thus Theorem 3.14 yields
Var [X] = E X 2 − µ2X = 1.5 − 12 = 1/2.
 
(6)
Theorem 3.15

Var [aX + b] = a2 Var [X ] .

Proof: Theorem 3.15

We let Y = aX + b and apply Theorem 3.14. We first expand the second

moment to obtain
h i h i h i
2 2 2 2 2 2
E Y = E a X + 2abX + b = a E X + 2abµX + b2. (1)
Expanding the right side of Theorem 3.12 yields

µ2
Y = a2 µ2 + 2abµ + b2 .
X x (2)
Because Var[Y ] = E[Y 2] − µ2
Y , Equations (3.85) and (3.86) imply that
h i h i
2
Var [Y ] = a E X − a µX = a (E X − µ2
2 2 2 2 2
X ) = a2 Var X .
[ ] (3)
Example 3.33 Problem

A printer automatically prints an initial cover page that precedes the

regular printing of an X page document. Using this printer, the number
of printed pages is Y = X + 1. Express the expected value and variance
of Y as functions of E[X] and Var[X].
Example 3.33 Solution

The expected number of transmitted pages is E[Y ] = E[X] + 1. The

variance of the number of pages sent is Var[Y ] = Var[X].
Example 3.34 Problem

In Example 3.28, the amplitude V in volts has PMF


1/7 v = −3, −2, . . . , 3,
PV (v ) = (1)
0 otherwise.
A new voltmeter records the amplitude U in millivolts. Find the variance
and standard deviation of U .
Example 3.34 Solution

Note that U = 1000V . To use Theorem 3.15, we first find the variance
of V . The expected value of the amplitude is

µV = 1/7[−3 + (−2) + (−1) + 0 + 1 + 2 + 3] = 0 volts. (1)

The second moment is
h i
2
E V = 1/7[(−3)2 + (−2)2 + (−1)2 + 02 + 12 + 22 + 32] = 4 volts2.
(2)
Therefore the variance is Var[V ] = E[V 2] − µ2 2
V = 4 volts . By Theo-
rem 3.15,

Var [U ] = 10002 Var[V ] = 4,000,000 millivolts2, (3)

and thus σU = 2000 millivolts.
Theorem 3.16
(a) If X is Bernoulli (p), then (d) If X is Pascal (k, p), then

Var[X] = p(1 − p). Var[X] = k(1 − p)/p2.

(b) If X is geometric (p), then (e) If X is Poisson (α), then

Var[X] = (1 − p)/p2. Var[X] = α.

(c) If X is binomial (n, p), then (f) If X is discrete uniform (k, l),

Var[X] = np(1 − p). Var[X] = (l − k)(l − k + 2)/12.

Quiz 3.8

In an experiment with three customers entering the Phonesmart store,

the observation is N , the number of phones purchased. The PMF of N
is

(4 − n)/10 n = 0, 1, 2, 3
PN (n) = (1)
0 otherwise.
Find
(a) The expected value E[N ]
(b) The second moment E[N 2]
(c) The variance Var[N ]
(d) The standard deviation σN
Quiz 3.8 Solution
For this problem, it is helpful to wrote out the PMF of N in the table
n 0 1 2 3
PN (n) 0.4 0.3 0.2 0.1
The PMF PN(n) allows us to calculate each of the desired quantities.
(a) The expected value is
3
X
E [N ] = nPN (n)
n=0
= 0(0.4) + 1(0.3) + 2(0.2) + 3(0.1) = 1. (1)
(b) The second moment of N is
3
X
 2
E N = n2 PN (n)
n=0
= 0 (0.4) + 12 (0.3) + 22 (0.2) + 32 (0.1) = 2.
2
(2)
(c) The variance of N is
Var[N ] = E N 2 − (E [N ])2 = 2 − 12 = 1.
 
(3)
p
(d) The standard deviation is σN = Var[N ] = 1.
Section 3.9

Matlab
Matlab: Finite Random Variables

• In Matlab, we
h representi0 SX , the sample space of X, by the column
vector s = s1 · · · sn and the corresponding probabilities by the
h i0
vector p = p 1 · · · p n .∗

• The function y=finitepmf(sx,px,x) generates

h the probabilities
i0 of the
elements of the m-dimensional vector x = x1 · · · xm .

h i0
• The output is y = y1 · · · ym where yi = PX(xi).

• That is, for each requested xi, finitepmf returns the value PX(xi).

• If xi is not in the sample space of X, yi = 0.

∗ Although
column vectors are supposed to appear as columns, we generally write a
 0
column vector x in the form of a transposed row vector x1 · · · xm to save space.
Example 3.35 Problem

In Example 3.27, the random variable X, the weight of a package, has

PMF

0.15 x = 1, 2, 3, 4,


PX (x) = 0.1 x = 5, 6, 7, 8, (1)



0 otherwise.
Write a Matlab function that calculates PX(x). Calculate the probability
of an xi pound package for x1 = 2, x2 = 2.5, and x3 = 6.
Example 3.35 Solution

The Matlab function shipweightpmf(x) implements PX(x). We can then

use shipweightpmf to calculate the desired probabilities:

function y=shipweightpmf(x) >> shipweightpmf([2 2.5 6])’

s=(1:8)’; ans =
p=[0.15*ones(4,1); 0.1*ones(4,1)]; 0.1500 0 0.1000
y=finitepmf(s,p,x);
Example 3.36 Problem

Write a Matlab function geometricpmf(p,x) to calculate, for the sample

values in vector x, PX(x) for a geometric (p) random variable.
Example 3.36 Solution

function pmf=geometricpmf(p,x) In geometricpmf.m, the last line ensures

%geometric(p) rv X
%out: pmf(i)=Prob[X=x(i)] that values xi 6∈ SX are assigned zero prob-
x=x(:); ability. Because x=x(:) reshapes x to be
pmf= p*((1-p).^(x-1));
pmf= (x>0).*(x==floor(x)).*pmf;
a column vector, the output pmf is always
a column vector.
Example 3.37 Problem

Write a Matlab function that calculates the Poisson (α) PMF.

Example 3.37 Solution
For an integer x, we could calculate PX(x) by the direct calculation

px= ((alpha^x)*exp(-alpha*x))/factorial(x)

This will yield the right answer as long as the argument x for the factorial function is
not too large. In Matlab version 6, factorial(171) causes an overflow. In addition,
for a > 1, calculating the ratio ax /x! for large x can cause numerical problems because
both ax and x! will be very large numbers, possibly with a small quotient. Another
shortcoming of the direct calculation is apparent if you want to calculate PX(x) for
the set of possible values x = [0, 1, . . . , n]. Calculating factorials is a lot of work for
a computer and the direct approach fails to exploit the fact that if we have already
calculated (x − 1)!, we can easily compute x! = x · (x − 1)!. [Continued]
Example 3.37 Solution (Continued 2)
A more efficient calculation makes use of the observation
ax e−a a
PX (x) = = PX (x − 1) . (1)
x! x
The poissonpmf.m function uses Equation (3.94) to calculate PX(x). Even this code is
not perfect because Matlab has limited range.

function pmf=poissonpmf(alpha,x) In Matlab, exp(-alpha) returns zero for alpha >

%output: pmf(i)=P[X=x(i)] 745.13. For these large values of alpha,
x=x(:); k=(1:max(x))’; poissonpmf(alpha,x)
ip=[1;((alpha*ones(size(k)))./k)]; returns zero for all x. Problem 3.9.9 out-
pb=exp(-alpha)*cumprod(ip); lines a solution that is used in the version of
%pb= [P(X=0)...P(X=n)] poissonpmf.m on the companion website.
pmf=pb(x+1); %pb(1)=P[X=0]
pmf=(x>=0).*(x==floor(x)).*pmf;
%pmf(i)=0 for zero-prob x(i)
Example 3.38 Problem

Write a Matlab function that calculates the CDF of a Poisson random

variable.
Example 3.38 Solution

function cdf=poissoncdf(alpha,x) Here we present the Matlab code for the

%output cdf(i)=Prob[X<=x(i)]
x=floor(x(:)); Poisson CDF. Since the sample values
sx=0:max(x); of a Poisson random variable X are in-
cdf=cumsum(poissonpmf(alpha,sx));
%cdf from 0 to max(x)
tegers, we observe that FX(x) = FX(bxc)
okx=(x>=0);%x(i)<0 -> cdf=0 where bxc, equivalent to the Matlab func-
x=(okx.*x);%set negative x(i)=0
cdf= okx.*cdf(x+1);
tion floor(x), denotes the largest inte-
%cdf=0 for x(i)<0 ger less than or equal to x.
Example 3.39 Problem

In Example 3.17 a website has on average λ = 2 hits per second. What

is the probability of no more than 130 hits in one minute? What is the
probability of more than 110 hits in one minute?
Example 3.39 Solution

Let M equal the number of hits in one minute (60 seconds). Note that
M is a Poisson (α) random variable with α = 2 × 60 = 120 hits. The
PMF of M is

(120)me−120/m! m = 0, 1, 2, . . .
PM (m) = (1)
0 otherwise.

>> poissoncdf(120,130) The Matlab solution shown on the left executes

ans =
the following math calculations:
0.8315
>> 1-poissoncdf(120,110) 130
X
ans = P [M ≤ 130] = P M (m) , (2)
0.8061
m=0
P [M > 110] = 1 − P [M ≤ 110]
110
X
=1− PM (m) . (3)
m=0
Example 3.40 Problem

Write a function that generates m samples of a binomial (n, p) random

variable.
Example 3.40 Solution

function x=binomialrv(n,p,m)For vectors x and y, c=count(x,y) returns

% m binomial(n,p) samples
a vector c such that c(i) is the number
r=rand(m,1); of elements of x that are less than or equal
cdf=binomialcdf(n,p,0:n);
x=count(cdf,r);
to y(i). In terms of our earlier pseudocode,
k∗ = count(cdf,r).
If count(cdf,r) = 0, then r ≤ PX(0) and k∗ = 0.
Figure 3.2

Rel. Freq.

Rel. Freq.
PY(y)

0.2 0.2 0.2

0 0 0
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
y y y
PMF PY (y) Sample Run 1 Sample Run 2

The PMF of Y and the relative frequencies found in two sample runs of
voltpower(100). Note that in each run, the relative frequencies are close
to (but not exactly equal to) the corresponding PMF.
Example 3.41 Problem

Simulate n = 100 trials of the experiment producing the power measure-

ment Y in Example 3.28. Compare the relative frequency of each y ∈ SY
to PY (y).
Example 3.41 Solution

function voltpower(n) In voltpower.m, we calculate Y = V 2/2 for each

v=duniformrv(-3,3,n);
y=(v.^2)/2; of n samples of the voltage V . As in Exam-
yrange=0:max(y); ple 2.26, the function hist(y,yrange) produces
yfreq=(hist(y,yrange)/n)’;
pmfplot(yrange,yfreq);
a vector with jth element equal to the number
of occurrences of yrange(j) in the vector y.
The function pmfplot.m is a utility for producing PMF bar plots in the
style of this text. Figure 3.2 shows PY (y) along with the results of two
runs of voltpower(100).
Matlab: Derived Random Variables
• For derived random variables, we exploit a feature of finitepmf(sx,px,x) that allows
the elements of sx to be repeated.

• Essentially, we use ( sx, px), or equivalently (sX , pX ), to represent a random variable

X described by the following experimental procedure:

Finite sample space

Roll an n-sided die such that side i has probability pi .

If side j appears, set X = xj .

A consequence of this approach is that if x2 = 3 and x5 = 3, then the probability

of observing X = 3 is PX(3) = p2 + p5 .
Example 3.42

>> sx=[1 3 5 7 3]; finitepmf() accounts for multiple occur-

>> px=[0.1 0.2 0.2 0.3 0.2];
>> pmfx=finitepmf(sx,px,1:7); rences of a sample value. In the example
>> pmfx’ on the left,
ans =
0.10 0 0.40 0 0.20 0 0.30 pmfx(3)=px(2)+px(5)=0.4.
Example 3.43 Problem

Recall that in Example 3.27 the weight in pounds X of a package and

the cost Y = g(X) of shipping a package were described by

0.15 x = 1, 2, 3, 4, 
105X − 5X 2

1 ≤ X ≤ 5,

PX (x) = 0.1 x = 5, 6, 7, 8, Y =

 500 6 ≤ X ≤ 10.

0 otherwise,
Write a function y=shipcostrv(m) that outputs m samples of the shipping
cost Y .
Example 3.43 Solution

function y=shipcostrv(m) The vector gx is the mapping g(x) for each

sx=(1:8)’;
x ∈ SX . In gx, the element 500 appears three
px=[0.15*ones(4,1); ...
0.1*ones(4,1)]; times, corresponding to x = 6, x = 7, and x = 8.
gx=(sx<=5).* ... The function y=finiterv(gx,px,m)) produces m
(105*sx-5*(sx.^2))...
+ ((sx>5).*500); samples of the shipping cost Y .
y=finiterv(gx,px,m);

>> shipcostrv(9)’
ans =
270 190 500 270 500 190 190 100 500
Quiz 3.9

In Section 3.5, it was argued that the average

n
1 X
mn = x(i) (1)
n i=1
of samples x(1), x(2), . . . , x(n) of a random variable X will converge to
E[X] as n becomes large. For a discrete uniform (0, 10) random variable
X, use Matlab to examine this convergence.
(a) For 100 sample values of X, plot the sequence m1, m2, . . . , m100. Re-
peat this experiment five times, plotting all five mn curves on common
axes.
(b) Repeat part (a) for 1000 sample values of X.
Quiz 3.9 Solution

The function samplemean(k) generates and plots five mn sequences for n =

1, 2, . . . , k. The ith column M(:,i) of M holds a sequence m1, m2, . . . , mk .

function M=samplemean(k);
K=(1:k)’;
M=zeros(k,5);
for i=1:5,
X=duniformrv(0,10,k);
M(:,i)=cumsum(X)./K;
end;
plot(K,M);

Here are two examples of samplemean: [Continued]

Quiz 3.9 Solution (Continued 2)

10 10

8 8

6 6

4 4

2 2

0 0
0 50 100 0 500 1000

(a) samplemean(100) (b) samplemean(1000)

Each time samplemean(k) is called produces a random output. What

is observed in these figures is that for small n, mn is fairly random but
as n gets large, mn gets close to E[X] = 5. Although each sequence
m1, m2, . . . that we generate is random, the sequences always converges
to E[X]. This random convergence is analyzed in Chapter 10.