.

UNIT 5 SAMPLE SPACE OF A RANDOM

EXPERIMENT
Structure
5.1 Introduction
Objectives
5.2 Random Experiments
5.3 Sample Space
5.4 Events
5.5 Algebra of Events
5.6 Summary
5.7 Solutions and Answers

INTRODUCTION

Many situations arise in our everyday life as well as in scientific, administrative or

organisational work, where we cannot predict'the outcome of our actions or of the
experiment we are conducting. Such experiments, whose outcome cannot be predicted, are
called random experiments. We give a wide variety of examples in Sec. 5.2 to explain the
concept of a random experiment. The set of all possible outcomes of an experiment is called
its sample space. We have illustrated the different types of sample spaces that we generally
come across in Sec. 5.3. Section 5.4 deals with the study of events associated with a random
experiment whose sample space is either finite or countably infinite. In Sec. 5.5 we discuss
methods of combining events to generate new events. Here is a list of what you should be
able to do by the end of this unit.

Objectives
After studying this unit you should be able to :
distinguish between random and non-random experiments,
specify the sample space of a random experiment and classify it as discrete or
continuous,
identify events with subsets ofthe sample space,
examine and identify relations between events,
generate new events out of a given collection of events.

RANDOM EXPERIMENTS

We give below some examples of a random experiment :

A physicist performs an experiment to discover laws governing the flow of an electrical
current or the propagation of sound, heat or light etc.
A chemist studies the reactions of chemicals and tries to understand the chemical
properties of matter.
A physician compares two or more d r u to
~ findout the' most effective one by trying
them out on experimental animals or on patients.
To describe the relationship between the price of a commodity and its demand and
* supply, an economist observes the values assumed by these variables by conducting a
market survey over a period of time.
'.I With a little imagination, we can construct many more examples of such experiments.

Experimentation is not necessarily restricted to a laboratory or to a university or a college. It

fnrms an irnnnrtant nart nf nllr Pvervdav life When vn11huv a dress nr a shirt when vnll
Probability on Discrete Sample vote for a candidate at an ele~~.,~.,r..lLii J O U r l l b p t X t a few grains of rice to decide whether
Spaces the rice is cookedor not, when you decide to register for this course, you are performing an
experiment. Thus, experimentation constitutes an integral part of our lives as well as our
learning processes. In this unit we shall develop methods of describing the results of an
experiment. Once we can describe the results we'll be able io talk about the chances of their
occurrence.

Consider the following simple experiments :

Experiment 1 :A stone is allowed to fall freely from height and we observe whether or not
the stone hits the ground.

Experiment 2 :Water in a pot is heated for a sufficiently long time to a temperature greater
than 100°C.We observe whether the water turns into steam.

In these experiments, we have no doubt about the final outcome. The stone will eventually
hit the ground. The water in the pot will ultimately turn into steam. These experiments have
only one possible outcome. Even if these experiments are repeated again and again, every
such repetition will yield the same result.

On the other hand, in the following experiments there are two or more possible results.

Experiment 3 :A coin is tossed to deoide which of the two teams A and I3 would bat first in
a game of cricket. The coin may tum up a head or a tail.

Experiment 4 :A person coming out of a polling centre is requested to disclose the name,of
the candidate in whose favour he/she has voted. Helshe may refuse to tell us or give the
name of any one of the candidate.

Experiment 5 :Three consecutive items produced by a machine are inspected and classified
as good or bad (defective). We may get 0, 1,2, or 3 defective items as a result of this
inspection.

Experiment 6 :A newly invented vaccine against a disease is given to 30 healthy people.

These thirty people as well as another group of 20 similar people who are not vaccinated,
are watched over the next six months to see whether they develop the disease. The total
number of affected people may vary between 0 and 50.

Experiment 7 :A small town has 100 telephones. The number of busy telephones between
9 and 10 a.m. is noted for each day of a week. The number of busy telephones may be any
number between 0 to 100.

Experiment 8 :A group of ten persons is classified according to their blood groups 0,A, B
and AB. The number of persons in each group may vary between 0 and 10, subject to the
frequencies of all four classes adding up to 10.

Experiment 9 :The number of accidents along the Bombay-Bangalore national highway

during the month is noted.

Experiment 10 :A radio-active substance emits particles called a-particles. The number of

such~particlesreaching an observation screen during one hour is noted.

Experiment 11 :Thirteen cards are selected without replacement from a well-shuffled paclc
of 52 playing cards.

The nine experiments, 3- 11, have two common.features.

i) Each of these experiments h v e more than olie possibie outcome.
ii) It is impossible to predict the outcome of the experiment.
For example, we cannot predict whether a coin, when it is tossed, will turn up a head or a
tail (Experiment 3). Can we predict without error the number of busy telephones
(Experiment 7)? It is impossible to predict the 13 cards we shall obtain from a well-shuffled
pack (Experiment 11).

Do you agree that all the experiments 3-1 1 have the above-mentioned features (i) and (ii)?
Go through them carefully again, and convince yourself.

This discussion leads us to the fnlln-. : inition.

 Definition 1 :An experiment with more than one possible outcome and whose result cannot Sample Space of a Random
be predicted, is called a random experiment. Experiment

So, Experiments 3 to 11 are random experiments, while in Experiments 1 and 2 the outcome
of the experiment can be predicted. Therefore, Experiments 1 and 2 do not qualify as
random experiments. You will meet many more illustrations of random experiments in this
and subsequent units.

You may now try this exercise.

El) Classify the experiments described below as random or non-random experiments.

-
In the dictionary- YOU will find that
a) A spark of electricity is intiaducedin a cylinder containing a mixture of hydrogen that is rendom, happens
and oxygen. The end product is observed. or is chosen without a definite plan.
pattern or purpbse.
b) A lake contains two types of fish. Ten fish are caught and the number of fish of
each type is noted.
c) The time taken by a powerful radio impulse to travel from the earth to the moon
and for its echo to return to the sender is observed.
d) Two cards are drawn from a well-shuffled pack of 52 playing cards and the suits
(Club, Diamond, Heart and Spade) to which they belong are noted.

In the next'section we shall talk about the set of all possible outcomes of a random
experiment.

5.3 SAMPLE SPACE

In the previous section you have seen a number of examples of random experiments. The
first step we take in the study of such experiments is to specify the set of all possible
outcomes of the experiment under consideration.
When a coin is tossed (Experiment 3), either a head turns up or a tail turns up. We do not
consider the possibility of the coin standing on its edge or that of its rolling away out of
I
sight. Thus, the set SZ of all possible outcomes consists of two elemends, Head and Tail.
Therefore, we write SZ = (Head, Tail) or, more simply, SZ = (H, T). !2 is the Greek letter capital 'omega'
In Experiment 4, the person coming out of the polling centre may give us the name of the
candidate for whom helshe voted, or may refuse to disclose hisher choice. If there are 5
candidates C1,C2, C3. C4 and C5, seeking election, then there are six possible outcomes,
five corresponding to the five candidates and the sixth one corresponding to the refusal R of
the interviewed person to disclose hisher choice. The set of all possible outcomes is thus,
{Cl,C2, C3*C4, C57 R ) .

Note that here we have ignored certain possibilities, like the possibility of the person not
voting at all or voting in such a manner that hisher ballot paper becomes invalid.

Experiment 5 is comparatively simple, if we agree that it is possible to classify each item as

Good (G) or Bad (B) without error. Then R = {GGG,GGB, GBG, BGG, BBG, BGB, GBB,
BBB ) where. for example, GBG denotes the outcome when the first and third units are good
and the second one is bad.

The situation in Experiment 6 is a little more complicated. To test the efficacy of the
vaccine, we will have to look at the number of vaccinated persons who were affected (x) q d
the number of non-vaccinated ones who were affected (y). Here x can be any integer
between 0 and 30 and y can be any integer between 0 and 20. The set SZ of all possible
outcomes is
SZ={(x,y)Ix=O,l, ..., 3O,y=O,l,2,..., 20).

This specification of f2 is valid only if we assume that we are able to observe all the 50
persons for the entire period of six months. In particular, we assume that none of them
becomes untraceable because of hisher leaving the town or because of hisher death due to
some other cause.
Probability on Discrete Sample In the illustrations discussed so far, do you notice that the number of points in fl is finite in
Spaces each case? It is 2 for Experiment 3 , 6 for ExIjeriment 4,31 x 21 = 651 for Experiment 6. But
this is not always true.

Consider, for example, Experiments 9 and 10. The number of accidents along the
Bombay-Bangalore highway during the month of observation can be zero, one, two, . . .or
some other positive integer. Similarly, the number of a-particles emitted by the radio-active
substance can be any positive integer. Can we say that the number of accidents or
a-particles would not exceed a specified limit? No. Because of this, and also in order to
simplify our mathematics, we usually postulate that in both these examples the set of all
possible outcomes is R = (0, 1,2, . . .},i.e., it is the set of all non-negative integers.
We are ngw in a position to introduce certain terms in a formal manner.

Definition 2 : The set R of all possible outcomes of an experiment E is called the sample
space of the experiment. Each individual outcome of E is called a point, a sample point or
an element of R.
You would also notice that in every experiment th,at was discussed, we made certain
assumptions like the coin not being able to stand on its edge or not rolling away, all the fifty
persons being available for the entire period of six months for observation, etc. Such
assumptions are necessary to simplify our problems as well as our mathematics.

In all the examples discussed so far, the sample space is either a finite set, i.e., a set
containing a finite number of points or is an infinite set whose elements can be arranged in
an unending sequence, i.e., has a countable infinity of elements. We have a special name for
such spaces.
Definition 3 :A sample space containing a finite number of points or a countable infinity of
points is called a discrete sample space.
In this block we shall be concerned only with discrete sample spaces. However, there are
m h y situations where we have to deal with sample spaces which are not discrete. For
example, consider the age of a person. Although there are limitations to the accuracy with
which we can measure the age of a person, in the idealised situation we can think of age
being any number between 0 and w. Of course, no one has met a person.with infinite age of
for that matter who is more than 150 years old. Nevertheless, most of the actuarial and
demographic studies are carried out assuming that there is no upper bound on age. Thus, we
may say that the sample space of the experiment of finding out the age of an arbitrarily
selected person is the interval 10, w[. Since the elements of the interval 10, w[ cannot be
arranged in a sequence, such a sample space is not a discrete sample space.

Some other examples where non-discrete sample spaces are appropriate are (i) the price of
wheat, (ii) the amount of ozone in a volume of space, (iii) the length of a telephone
conversation, (iv) the duration one spends in a queue, (v) the yield of rice in our country in
one year.

In all these examples, it is necessary to deal with non-discrete sample spaces, However,
we'll defer the study of probability theory for such experiments to the next block.

Now see if you can solve this exercise.

E2) Write down the sample spaces of all those experiments from 3 to 11 which we have
not discussed earlier. Indicate in each case the assumptions made by you.

Now that we have seen how to specify the elements of a sample space, we can talk about the
events associated with it.

5.4 EVENTS

We have described a number of random experiments till now. We have also identified the
sample spaces associated with them. In the study of random experiments, we are interested
not only in the individual outcomes but also in certain events. As you will see later, events
are subsets of the sample space. In this section we shall formalise the intuitive concept of an
event associated with a random ex~erimentwhich has a discrete sam~lemace. We shall also
study methods of generating new eveqts from specified ones and study iheit Sample Space o l a Random
inter-relationships. Experiment

Consider the experiment of inspecting three items (Experiment 5). The sample space has the
eight points,
GGG, GGB, GBG, BGG, BBG, BGB, GBB, BBB.
We label these points o l , y,. . . , us, respectively.

Suppose we are interested in those outcomes which correspond to the event of obtaining
exactly one good item in the three inspected items. The corresponding'sample points are
o5= BBG, o6= BGB and a, = GBB. Thus, the subset (05, 06, 9)of the sample space o is the lower case Greek letter
'omega'.
corresponds to the "event" A that only one of the inspected items is good.
On the other hand, consider the subset C =.(a5, 06, 9 , 081 consisting of the points BBG,
BGB, GBB, BBB. We can identify the subset C with the event "There are at least two bad
items."

This discussion suggests that we can associate a subset of the sample space with an event
and an event with a subset. This leads us to the following definition.

Definition 4 :When the sample space of an experiment is discrete, any subset of the sample
space is called an event.
Thus, we alsq consider the empty set as an event.
You will soon find that the two extreme events, cp and R, consisting, respectively, of no
points and all the points of R are most uninteresting. But we need them to complete our
description of the class' of all events. Ih fact, cp is called the impossible event and R is called
the sure event, for reasons which will be obvious in the next unit. Also, note that an
individual outcome o , when identified with the singleton (a), constitutes an event.
The following example will help you in understanding events.

Example 1 : Suppose we toss a coin twice. The sample space of this experiment is
R = (HH, HT, TH, TT}, where HT stands for a head followed by a tail, and other points are
similarly defined. Let's list all the events associated with this experiment. There are 16 such
events. These are :
cp,(HHl, (HTl, (THI,

Since we have identified an event with a subset of R, the class of all events is the class of all
the subsets of R. If R has N points, for a fixed r, we sets consisting of r points,
where r = 0, 1, . . . , N. The total number of events is, thirefore,
By binomial theorem

In Example 1, N = 4. Therefore, we have 24 = 16 events. If N = 10, we shall 21° = 1024

events. The number of events thus increases rapidly with N. It is infinite i'f the sample space
is infinite.
Let us now clarify the meaning of the phrase "The event A has occurred."

We continue with Experiment 5. Let A denote the event (/w5,05, 9)= (BBG, BGB,
GBB 1. If, after performing the experiment, our outcome b o5= BBG, which is a point of
the set A, we say that the event A has occurred. If, on the other hand, the outcome is wg =
BBB, which is not a point of A, then we say that A has not occurred. In other words, given
the outcome o of the experiment, we say that A has occurred if o E A and that A has not
occurred if o e A.
n n the n t h ~hand
r we
if we nnlv know that A has ncc~~rred.'all know is thatthe outcome of
Robablllty on Dkrete Sample the experiment is one of the points of A. It IS the11not possible to decide which individual
spPces outcome has resulted unless A is a singleton.

In the next section we shall talk about some ways of combining events.

5.5 ALGEBRA OF EVENTS

In this section we shall study different ways in which we can combine two or more events.
We shall also study relations ktween them. Since we are dealing with discrete sample
spaces and since any subset of the sample space is an event, we shall use the terms event and
subset interchangeable.

In what follows, events and sets are denoted by capital letters A, B, C, . . . , with or without
suffixes. We shall assume that they all consist of points chosen from the same sample
space Q.

Let f2 = (GGG, GGB, GBG, BGG, BBG, BGB, GBB, BBB ) be the sample space
co~~esponding to Experiment 5. Let A = (BBG, BGB, GBB) be the event that only one of
the three inspected items is good. Here the point BGB is an element of the set A and the
~ecall.(unit 1,MTE-04)that the point BBB is not an element of A. We express this by writing BGB E A and BBB Q A.
complernent. A' = {oE I o e A].
Then $' = C2 and ff = $. Fig. 1 Suppose, now, that the outcome of the experiment is BBB. Obviously, the event A has not
shows a Venn diagram representing occurred. But, we may say the event "not A" hasaoccurred.In probability theory, the event
the sets A and A'.
"not A" is called the event complementary to A and is denoted by A'.
.
Let's try to understand this concept by looking back at Experiments 3-11.

Example 2
i) For Experiment 5 , if A = (BBG, BGB, GBB ) ,then
AC= (GGG, GGB, BGG, GBG, BBB J.
ii) In Experiment 6, let A denote the event that the number of infected persons is at most
40. Then

iii) In Experiment 11, if B denotes the event that none of the 13 cards is a spade, B'
consists of all hands of 13 cards, each one of which has at least one spade.
Suppose now that Al and A2 are two events associated with an experiment. We can get two
new events, Al n.A2 (Al intersection A2) and A, U A2 (Al union A2) from these two. With
your knowledge of set theofy (MTE-04). you would expect the event Al n A2 to correspond
to the set whose elements belong to both Al and A2. Thus,

Similarly, the event Al U A2corresponds to the set whose elements belong to at least one of
A1 and A2.

Fig. 2 (a) and (b) show the Venn diagrams representing Al n A2 and Al U A2.
 Sample Space of a Random
We'll try to clarify this concept with some examples. Experiment

Example 3 :In many games of chance, a small'cube (or die) with equal sides, bearing
numbers 1,2,3,4,5,6, or dots 1-6 on its six faces (Fig. 3), is used. When such a symmetric ,
die is thrown, one of its six faces would be uppermost. The number (or number of dots) on
the uppermost faces is called the score obtained on the throw or roll of a die. The
appropriate sample space for the experiment of throwing a die is then R = { 1,2,3,4,5,6).
Let Al be the event that the score exceeds three and A2 be the event that the score is even.
Then

Therefore, Al fA2
l= ( 4 , 6] and

Suppose now that the score is 6. We can say that Al has occurred. But then A2 has also
occurred. In other words, both Al and A2 have occurred. Thus, the simultaneous occurrence
of Al and A2 corresponds to the occurrence of the event Al fl A2.
1

When the outcome is 5, Al has occurred but A2 has not occurred. Further, when the
outcome is 2, A2 has occurred and Al has not. When the outcome is 4, both Al and A2 have
occurred. In case of each of these outcomes, 2,5 or 4, we notice that at least one of Al and
A2 has occurred. Note, further, that Al U A2 has also occurred. Thus, the occurrence of at
least one of the two events Al and A2 corresponds to the occurrence of Al U A2.

Example 4 : Suppose the die in Example 3 is thrown twice. Then R is the set {(x,y) I x, y =
1,2,3, . . . , 6 ] consisting of thirty-six points (x, y), where x is the score obtained on the first
throw and y, that obtained on the second throw. If BI is the event that the score on the first
Fig. 3
throw is six and B2 the event that the sum of the two scores is at least 11, then

PI
$1 and
B2 = { (5,6), ( 6 % (6,6) I.
What are B I fl B2 and B U B2? You can check that

Bl n B2= ( ( 6 9 , (6,6)]

The union and intersection of two setscan be utilised to define union and intersection.of
three or more sets.
- ,
L L

't"
L So, if AI, A2, . . . ,An are n events, then we define
I: j n

and
n
U Aj = { o I oAjforatleastonej=l,
~ ..., n].
j= I
n
Note that the occurrence of fl Aj corresponds to the simultaneous occprrence of all the n
j= I
n
events and the occurrence of U Aj corresponds to that of at least one of the n events
j= 1
A*,. . . ,A,. We can similarly define the union and intersection of an infinite number of
itt .
events, A1, A2, . . . 'An, . . . .
Probability on Dlscrete Sample Another set operation with which you are familiar is a combination of complementation and
SP- intersection. Let A and B be two sets. Then the set A fl BCis usually called the difference of
A and B and is denotedby A - B. It consists of all points which belong to A but not to B.
Thus, in Example 4,

and
B 2 - B 1 = ((5,6)1
In this notation, ACis the set $2-A. You can see the Venn diagram for A - B in Fig. 4.
Now, suppose A l , A2 and A3 are three arbitmy events. What does the occurrence of
Al n A: r l A; signify?

This event occurs iff only Al out of Al, A2 and A3 occurs, that is, iff A l occurs but neither
A2 nor A3 occur. '

If you have followed this, you should be able to do this exercise quite easily.

E3) If A,, A2 and Aj are three arbitrary events, what does the occurrence of the following
events signify?
Fig. 4 :The shaded portion a) E l = A l n A 2 f l A 3
-
represents A B.
b) E 2 = A i n A 5 n A ?

The set operations like formation of intersection, ur~ionand complementation of two or more
sets that we have listed above and their combinations are sufficient for constructing new
events out of old ones. However, we need to express in a precise way commonly used
expressions like (i) if the event A has occurred, B could not have occurred and (ii) the
occurrence of A implies that of B. We'll explain this by taking an example first.
Example 5 :Let us consider the following experiments.
i) In the experiment of tossing a die twice, let A be the event that the total score is 8 and
B that the absolute difference of the two scores is 3. Then

and B = ( ( x , y ) I I x - y ( = 3 , x , y = 1 , 2 , 3 ,..., 6.)

ii) Consider Experiment 1 1, where we select 13 cards without replacement from a pack
of cards. Let
event A : all the 13 cards are black and
event B : there are 6 diamonds and 7 hearts.
Note that in both the-casesthere is no point which is common to both A and B. Or in
other words, A n B is the empty set. Therefore, in both i) and ii) we conclude that if
A occurs, B cannot occur and conversely, if B occurs A cannot occur.
Now let us find an example for the sifuation :the occurrence of A implies that of B.
.
Take the experiment of tossing a die twice. Let A = ((x, y) 1 x + y = 12) be the event
that the total score is 12, and B = ((x, y) 1 x - y = 0 ) be the event of having the same
score on both the throws. Then
A = {(6,6)]and

so that whenever A occurs, B does. Note that A c B.

 You were already familiar with the various operations on sets. In Sec. 5.4 we had . Sample Space of a Random
Experiment
identified events with subsets of the sample space. What we have done in.this section -
is to apply set operations to events, and to interpret the combined events. I
I

You can check whether you have grasped these ideas by doing the following exercises.
9

E4) Let Al, A2, A3 and A4 be arbitrary events. Find expressions for the events that
correspond to occurrence of
a) only Al and A2,
b) none of Al, A2, A, and A4,
c) one and only one of A1, A,, A3, A,,
d) not more than one of Al, A2, A3. A4,
e) at least two of Al, A2, A3. A4
E5) Express in words the following events :
a) A; n A 2 fl A3
I
b) (A; nA; nA;) U ( A ~nA;nA;)

C) (A1 U A2) - (A3 U A4)

d) (A, u A,) n

Now, before ending this unit-let us go over its main points.

5.6 SUMMARY

In this introductory unit to the study of probability, we have made the following points :
1) There are many situations in real life as well as in scientific work which can be
regarded as experiments having more than one possible oatcome. We cannot predict
the outcome that we will obtain at the conclusion of the experiment. Such experiments
are called random experiments.
2) The study of random experiments begins with a specification of its all possible
outcomes. In this specification, we have to make certain assumptions to avoid
complexities. The set of all possible outcome is called the sample space of the
experiment. A sample space with a finite number or a countable infinity of points is a
/
discrete sample space.
'
3) When we are dealing with a discrete sample space, we can identify events with sets of
points in the sample space. Thus, an event can be formally regarded as a subset of the
sample space. This definition works only when the sample space is discrete.
4) We can use operations like complementation, intersection,,union and difference to
generate new events.
5) Some complex events can be described in terms of simpler events by using the
above-mentioned set operations.

5.7 SOLUTIONS AND 'ANSWERS

El) a) This is a non-random experiment as the electrical spark would ignite the hydrogen
and it would combine with oxygen to produce water.
b) This is a random experiment as onecannot predict the number of fish of each type
that would be caught. /
-. . - - .- -
Probnbillty on Discrete hrnple a physical constant. The time for the radio impulse to reach the moon ; ~ l r t l
SpPcrs for its echo to return can be predicted without error. Hence this is a non-random
experiment.
d) A random experiment.
E2) The sample space for the random experiment described in
i) Experiment 7 i s Q = (0, 1,. . . , 1001
ii) Experiment 8, is
n = r (x0, XA, XB,xAB)I x0 + X A + XB + XAB = 10 I.

~ the number of persons with blood-groups 0, A, B

where xo, xA, xB and x A are
and AB, respectively,in the group of 10 persons.
iii) Experiment 11, is the set of all possible, i.e.,

(:)= 2.476552 x suits of 13 cards that can fomed out of 52 cards.

E3) a) The event El = Al flA2 r l A3 occurs if all the three occur.

b) None of the three events Al, A2. A3 occurs iff A: n A", A; occurs.

c) The event E3 occurs if exactly two of the three events occur.

d) El U E3corresponds to occurrence of at least two of the three events
A,, A2 and A3.

E4) a) Al n A 2 , 9 A $ n A i

C) ( A ~ ~ A $ ~ A ; ~ A ~ ) u ( A ; ~ A ~ ~ A $ ~ A ~ ) u ( A ~ ~ A $
u A; n A; n.A; 0 A4) = E2, say.

d) El U E ,
e) (El U QC.
E5) These events corresponds to occurrence of
a) A2 and A3 but not Al, .
b) None of A2 and Ag,

C) At least one of Al and A2, but none of A3 and A4,

d) A3 and at least one of A, and A2,

e) At least two out of Alp A2 and A3.