Lecturenotes5 6 Probability
Lecturenotes5 6 Probability
K. Suresh Kumar
Department of Mathematics
Indian Institute of Technology Bombay
August 5, 2017
2
LECTURES 5 - 6
Proof: The proof of first two are simple exercises. But don’t forget to write
down the solutions.
For a ∈ R,
The proof of the above identity, only uses ”between any two real numbers,
there is always a rational” Now
1
{X + Y ≤ a} = ∩∞
n=1 {X + Y < a + }.
n
Hence {X + Y ≤ a} ∈ F for all a ∈ R. Therefore X + Y is a random
variable.
For a ∈ R,
∅∈F if a < 0
{X 2 ≤ a} = {X = 0} = X −1 ({0}) ∈ F if a = 0
√ √ √ √
{− a ≤ X ≤ a} = X −1 ([− a, a]) ∈ F if a > 0 .
√ √
(In the above {0} and [− a, a] are closed sets and hence Borel sets.)
Hence, X 2 is a random variable.
Note that
1
XY = [(X + Y )2 − X 2 − Y 2 ] .
2
Since 2 (X + Y ) , − 2 X , − 12 Y 2 are random variables and XY is their sum,
1 2 1 2
XY is a random variable.
Proof:
(i) Set Z = min{X, Y }. For a ∈ R,
{Z ≤ a} = {X ≤ a} ∪ {Y ≤ a} ∈ F .
{X ≤ a} = ∩∞
n=1 {Xn ≤ a} .
is a random variable.
Proof. For a ∈ R.
ω ∈ {X ≤ a} ⇒ for each m ≥ 1, there exists n such that
1 1
Xk (ω) ≤ X(ω) + m ≤ a+ m for all k ≥ n
⇒ ω ∈ ∪∞ ∞
n=1 ∩k=n {Xk ≤ a +
1
m} , m ≥1
⇒ ω ∈ ∩∞ ∞ ∞
m=1 ∪n=1 ∩k=n {Xk ≤ a +
1
m} .
Hence
1
{X ≤ a} ⊆ ∩∞ ∞ ∞
m=1 ∪n=1 ∩k=n {Xk ≤ a + }.
m
Now suppose
1
ω ∈ ∩∞ ∞ ∞
m=1 ∪n=1 ∩k=n {Xk ≤ a + }.
m
If ω ∈
/ {X ≤ a}, then there exists there exists m0 and n0 such that
1
Xk (ω) > a + for all k ≥ n0 . (0.1)
m0
4
⇒ Xk (ω) ≤ a + m10
for all k ≥ n1 for some n1
ω ∈ {X ≤ a} .
Hence we have
1
{X ≤ a} = ∩∞ ∞ ∞
m=1 ∪n=1 ∩k=n {Xk ≤ a + }. (0.2)
m
1
Since {Xk ∈ a + m } ∈ F and F is a σ-field, using(0.2) it follows from the
definition of σ-field that
{X ≤ a} ∈ F .
Therefore X is a random variable.
x > c ⇒ there exists y ∈ Da such that c < y < x ⇒ f (x) ≥ f (y) > a ⇒ x ∈ Da .
Proof. For B ∈ BR ,
PA (B) = P (B|A), B ∈ F .
PA (A) = 1, PA (B) = 0 if B ⊆ Ac .
(exercise)
Remark 0.1 The above probability space may look useless but it is use-
ful for processing the thought experiments of the following type. Note that
(Ω, F, PA ) corresponds to a random experiment which comes from adding the
information that the event A has occurred. Hence by understanding directly
this random experiment gives the probabilities PA (B) directly without using
the formula for PA (B). We will see an illustration of this in a moment.
7
26 13
P (A♣,♦,♥,♠ |A♣,♦,♥ ) = , P (A♣,♦,♥,♠ |A♣,♦,♥ ) = .
50 49
Hence
39 × 26 × 13
P (A♣,♦,♥,♠ ) = = 0.11.
51 × 50 × 49
3
To compute the conditional probabilities, we will not use the definition of conditional
probability instead we use understand an underlying random experiment which gives raise
to the probability space of conditional probabilities, as told in Remark 0.1. Note that
given the information that the club ace is distributed to one player, we can think about
the random experiment as an urn problem with 51 urns numbered 1 to 51 and a ball
(identified with diamond ace) and our event is distributing the ball into one of the first
39 urns.
8
The above example illustrates it is some times more easy to compute (or
natural to specify) conditional probabilities and use them to specify the
underlying probabilities, a reverse procedure!
Example 0.6 (Probability in the game show ”Let’s make a deal”) Here we
look at a version of the game show ”Let’s make a deal” which made its debut
on NBC Television network on December 30, 1963. Description of the game
is the following. A prize is placed behind one of the three doors and are
closed. Contestants of the show are aware of this. Contestant is asked to
select a door (but is not going to open at the moment). Once the choice is
made, the moderator of the show (Monty Hall) opens on of the remaining
doors and display what is in it (He will only open a door which has no prize).
Contestant now is given a chance to change the earlier choice. Qustion is,
will the contestant stay with the earlier choice or not?
Without any loss of generality, assume that the contestant chose door
no.1 (label chosen door as no.1).
Here take sample space as
(Here ♦ab denotes price ♦ behind the first door and the ’worthless’ a and
b behind the doors 2 and 3 respectively. Other sample points are similraly
interpreted.)
Question can be answered if we know the probability of ’♦ behind door
1’ given the additional information of the object behind one of the doors 2
or 3.
Let us denote the event ’♦ behind door 1’ by ’♦ ∈ 1’. Also Monty Hall
revealing object behind door 2 means behind door 2, object is either a or b.
Hence occurrence of the event A, i.e. revealing door 2 means occurrence of
{♦ab, ♦ba, ab♦, ba♦}. i.e.
Hence it is better to change the option, since you have 2/3rd chance of
winning the prize by switching the door.
WARNING! Sometimes one make the mistake of calculating
1
P (♦ ∈ 1|A) + P (♦ ∈ 1|B) =
2
and gives ’wrong’ advise.
(ii) ∪N
i=1 Ai = Ω .
Here N may be ∞. If N < ∞, then partition is said to be finite partition
and if N = ∞, it is called a countable partition.
Proof.
N N
X X P (BAi )
P (B|Ai ) P (Ai ) = P (Ai )
P (Ai )
i=1 i=1
XN
= P (BAi )
i=1
= P (B(∪ni=1 Ai )) = P (B) .
The second last equality uses the countable additivity (for N = ∞) of prob-
ability to get convergence of the series.
P (B|A)P (A)
P (A|B) = .
P (B|A) + P (B|Ac )
10
Proof.
P (BA)P (A) P (B|A)P (A)
P (A|B) = = .
P (A) P (B) P (B)
Now use Law of total probability to complete the proof.
Remark 0.2 The total probability law is another formula for using condi-
tioning argument.