Lecture8 - Infinite Coin Toss

EE5110: Probability Foundations for Electrical Engineers
July-November 2015
Lecture 8: The Infinite Coin Toss Model

Lecturer: Dr. Krishna Jagannathan
Scribe: Subrahmanya Swamy P
In this lecture, we will discuss the random experiment where each trial consists of tossing a coin infinite
times. We will describe the sample space, an appropriate -algebra, and a probability measure that intuitively
corresponds to fair coin tosses. If we denote Heads/Tails with 0/1, the sample space of this experiment turns
out to be = {0, 1} , and each elementary outcome is some infinite binary string. As we have seen before,
this is an uncountable sample space, so defining a useful -algebra on takes some effort.
8.1
A -algebra on = {0, 1}
Let Fn be the collection of subsets of whose occurrences can be decided by looking at the result of the
first n tosses. More formally, the elements of Fn can be described as follows: A Fn if and only if there
n
exists some A(n) {0, 1} such that A = { |(1 , 2 , ..., n ) A(n) }.
Examples:
1 Let A1 be the set of all elements of such that there are exactly 2 heads during the first 4 coin tosses.
Clearly, A1 F4 .
2 Let A2 be the set of all elements of such that the third toss is a Head. Then, A2 F3 .
Also note that the following relation holds:

Fn Fn+1 n N.
(8.1)
Although Fn is a -algebra, it has the drawback that it allows us to describe only those subsets which can
be decided in n tosses. For example, the singleton set containing all Heads is not an element of Fn for any
n.
S
In order to overcome this drawback, we define F0 =
Fi . In words, F0 is the collection of all subsets of
iN
that can be decided in finitely many coin tosses, since an element of F0 must be an element of Fi for some
i N.
Proposition 8.1 We claim the following:
(i) F0 is an algebra.
(ii) F0 is not a -algebra.
8-1
8-2
Proof:
(i) This is just definition chasing! (Left as an exercise).
(ii) Consider the following example: Let E = { | every odd toss results in Heads}. Clearly, E
/ F0
since we cannot decide the occurrence of E in finitely many tosses. On the other hand, E can be
expressed as a countable intersection of elements in F0 :
E=
A2i1 ,
i=1
where Ai F0 is the set of all binary strings with Heads in the ith toss.
Next, consider the smallest -algebra containing all the elements of F0 , i.e., define
F = (F0 ).
8.2
A Probability Measure on ( = {0, 1} , F)
Now, we shall define a uniform probability measure on F that corresponds to a fair coin toss model. We shall
first define a finitely additive function P0 on F0 that also satisfies P0 () = 1. Then, we shall subsequently
extend P0 to a probability measure P on F.
If A F0 , then by the definition of F0 , n such that A Fn . By the definition of Fn , we know that for
every A Fn , there exists a corresponding A(n) {0, 1}n . We will use this A(n) in the definition of P0 . We
define P0 : F0 [0, 1] as follows:
P0 (A) =
|A(n) |
2n .
Having defined P0 this way, we need to verify that this definition is consistent. In particular, we note that
if A Fn , A Fn+1 , which is trivially true because Fn0 s are nested increasing. We therefore need to prove
that when we apply the definition P0 (A) for different choices of n, we obtain the same value. We leave it to
the reader to supply a formal proof for the consistency of P0 . However, we illustrate this consistency using
the examples provided in Section 8.1.
(i) The occurrence of the event A2 can be decided in the first 3 tosses. So, A2 F3 . The elements in
A(3) {0, 1}3 corresponding to the event A2 are {(0, 0, 0), (0, 1, 0), (1, 0, 0), (1, 1, 0)}. So |A(3) | = 4.
So, P0 (A2 ) = 243 = 12 .
The event A2 can also be looked as an event in F4 since, Fn Fn+1 . The elements in the corresponding A(4) will be {(0, 0, 0, 0), (0, 1, 0, 0), (1, 0, 0, 0), (1, 1, 0, 0), (0, 0, 0, 1), (0, 1, 0, 1), (1, 0, 0, 1), (1, 1, 0, 1)}.
So |A(4) | = 8. So, P0 (A2 ) = 284 = 12 .
(ii) A1 can be decided by looking at the outcome of the first four tosses. So, A1 F4 . It is easy to see that
the number of elements in A(4) {0, 1}4 corresponding to the event A1 that has exactly two heads is

(42)
4
.
Hence,
P
(A
)
=
0
1
2
24 . Next, can you compute P0 (A1 ) by considering A1 as an element of, say F5 ?
From the above examples, we can observe that
8-3
(a) The definition of P0 is consistent over different choices on n namely n = 3 and n = 4 for a given set
A2 .
(b) The definition of P0 is also consistent with the intuition of a fair coin toss model with probability of
heads being 12 .
It can be easily verified that P0 () = 1 and P0 is finitely additive. It also turns out that P0 is countably
additive on F0 (the proof of this fact is non-trivial and is omitted here). This allows us to invoke the
Caratheodory extension theorem and extend P0 to P, a legitimate probability measure on (, F) which
agrees with P0 on F0 . In other words, there exists a unique probability measure P on (, F).
As an example, let us consider the event E that is defined above (i.e. the set of strings in which all the odd
tosses are heads). As E
/ F0 , P0 is not defined for the event E. However, it is clear that E F, so that P
is defined for E. Let us calculate the probability of the event E. Recall that
E=
A2i1 , where Ai
{ | i = 0}.
i=1
Tm
Let us define the event Em = i=1 A2i1 . In other words, Em is set of outcomes in which the first 2m tosses
have the property of all odd tosses being heads. We can easily verify that P(Em ) = P0 (Em ) = 21m . Note
that {Em , m 1} is a sequence of nested decreasing events i.e., Em Em+1 , m
1. It can be easily
T
verified that E can be expressed in terms of these decreasing nested events as E = m=1 Em .
Thus,
P(E)
!
Em
m=1
(a)
=
=
lim P (Em )
lim
1
2m
0,
where the equality (a) follows from the continuity of probability measures.
8.3
Exercises
1. Show that Fn (defined in equation 8.1) is a -algebra n N.

2. Recall the infinite coin toss model with = {0, 1} ; where 0 denotes heads and 1 denotes tails.
Define Fn as the collection of subsets of whose occurrence can be decided by looking at the results of the
first n tosses. Exercise:
(a) Show that Fn is a -algebra.
It turns out that the -algebra Fn for any fixed n is too small; after all, it can only serve to model the
first n tosses. Let us define
[
F0 =
Fn .
(8.2)
i=1
8-4
(b) Give a verbal description of the collection F0 .

(c) Show that F0 is an algebra on .
(d) Consider the subset A consisting of sequences in which Tails occurs infinitely many times. Does
A F0 ?
(e) Is Ac countable?
(f) Let B be the set of all infinite sequences for which n = 0 for every odd n; i.e., every odd numbered
toss is Heads. Show that B can be written as a countable intersection of subsets in F0 , but B
/ F0 .
Therefore F0 is not a -algebra.
Define F = (F0 ), the -algebra generated by F0 .
(g) Show that every singleton {} is F measurable. Show that the uniform measure on (, F) defined in
class assigns zero probability measure to singletons.
(h) Let Ai be the set of all outcomes such that the ith toss is Tails. Note that Ai F0 . Show that A in
part (e) can be written as
[
\
Ai
(8.3)
A=
n=1 i=n
Hence show that A is F measurable.What is P {A} under the uniform measure?

(i) Let T be the set of all coin toss sequences in which the fraction of Tails is exactly 1/2: More
precisely,
Pn
1
i=1 i
T = { : lim
= }
(8.4)
n
n
2
The set T is called the strong-law truth set, for reasons that will become clear later. Does T F0 ?
(j) Show that T can be expressed as
T =
[
\
Pn
{ |
k=1 m=1 n=m
i=1
1
1
|< }
2
k
(8.5)
Argue that the subset inside the nested union and intersection above belongs to F0 : Hence show that
T is F-measurable. Hint: Dont get intimidated by the multiple unions and intersections! Write-out
the limit in the definition of T as the set of all such that for all k 1; there exists an m for
which for all n > m; we have
Pn
i
1
1
| i=1
|<
(8.6)
n
2
k

Lecture8 - Infinite Coin Toss

Uploaded by

Copyright:

Available Formats

Lecture8 - Infinite Coin Toss

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lecture8 - Infinite Coin Toss

Uploaded by

Copyright:

Available Formats

EE5110: Probability Foundations for Electrical Engineers

Lecture 8: The Infinite Coin Toss Model

Scribe: Subrahmanya Swamy P

Also note that the following relation holds:

Lecture 8: The Infinite Coin Toss Model

A Probability Measure on ( = {0, 1} , F)

Lecture 8: The Infinite Coin Toss Model

1. Show that Fn (defined in equation 8.1) is a -algebra n N.

Lecture 8: The Infinite Coin Toss Model

(b) Give a verbal description of the collection F0 .

Hence show that A is F measurable.What is P {A} under the uniform measure?

k=1 m=1 n=m

You might also like