Random Variable

Let (S, Σ, P ) be a probability space. Since P is a set function, it is not very easy to
handle. Also in many situations, one may not be interested in the sample space rather one
may be interested in some numerical characteristics of the sample space. For example,
when a coin is tossed n-times, which replication resulted in heads is not of much interest.
Rather, one is interested in the number of heads, and consequently, the number of tails,
that appear out of n tosses.

It is therefore desirable to introduce a point function on the sample space so that we

can use the theory of calculus or real analysis to study the properties of P .
Definition 1. A function X : S −→ R is called a random variable (RV) if X −1 (B) ∈ Σ,
for all B ∈ BR , that is, X −1 (B) = {w ∈ S : X(w) ∈ B} is an event.

Notations.

We will use the following notations throughout the course.

def def
• For B ∈ BR , {X ∈ B} = {w ∈ S : X(w) ∈ B} = X −1 (B);
def def
• {a < X ≤ b} = {w ∈ S : a < X(w) ≤ b} = X −1 ((a, b]);
def def
• {a ≤ X ≤ b} = {w ∈ S : a ≤ X(w) ≤ b} = X −1 ([a, b]);
def def
• {a < X < b} = {w ∈ S : a < X(w) < b} = X −1 ((a, b));
def def
• {a ≤ X < b} = {w ∈ S : a ≤ X(w) < b} = X −1 ([a, b));
def def
• {X = a} = {w ∈ S : X(w) = a} = X −1 ({a});
def def
• {X ≤ a} = {w ∈ S : X(w) ≤ a} = X −1 ((−∞, a]);
def def
• {X < a} = {w ∈ S : X(w) < a} = X −1 ((−∞, a));
def def
• {X ≥ a} = {w ∈ S : X(w) ≥ a} = X −1 ([a, ∞));
def def
• {X > a} = {w ∈ S : X(w) > a} = X −1 ((a, ∞)).
Remark 2. (1) X is a random variable if and only if for each x ∈ R, {X ≤ x} ∈ Σ.

(2) If Σ = P(S), then any function X : S −→ R is a random variable.

(3) Let (S, Σ, P ) be a probability space and X : S −→ R be a random variable. Then

the random variable X induces a probability space (R, BR , PX ), where PX (B) =
P ({w ∈ S : X(w) ∈ B}), for all B ∈ BR .
Example 3. Suppose that a fair coin is independently flipped thrice. Then
S = {HHH, HHT , HT H, T HH, T T H, T HT, HT T, T T T } and
P (E) = number of elements
8
in E
, for every E ∈ P(S). Define X : S −→ R by X(w) =
number of heads, i.e.,


 0, w = {T T T }

1, w ∈ {HT T, T T H, T HT }
X(w) =

 2, w ∈ {HHT, T HH, HT H}

3, w = {HHH}.


Clearly X is a random variable. The induced probability space is P (R, BR , PX ), where

PX ({0}) = PX ({3}) = 81 , PX ({1}) = PX ({2}) = 38 , and PX (B) = PX ({i}), for
i∈{0,1,2,3}∩B
all B ∈ BR .
1
Definition 4. Let (S, Σ, P ) be a probability space and X : S −→ R be a random variable.
The function FX : R −→ R, defined by,
FX (x) = P ({X ≤ x}), ∀ x ∈ R,

is called the cumulative distribution function (c.d.f ) or the distribution function

(d.f ) of the random variable X.
Theorem 5. Let FX be the cumulative distribution function of a random variable X.
Then

(1) FX is non-decreasing;
(2) FX is right continuous;
def def
(3) FX (−∞) = lim FX (x) = 0 and FX (∞) = lim FX (x) = 1.
x→−∞ x→∞

Proof. (1) Let x1 < x2 . Then (−∞, x1 ] ⊂ (−∞, x2 ]. Then by the properties of the
probability function, we have
FX (x1 ) = P ({X ≤ x1 }) ≤ P ({X ≤ x2 }) = FX (x2 ).

(2) Fix a ∈ R. Since FX is non-decreasing, FX (a+) = lim FX (x) exists. Therefore

x→a+

1 1
FX (a+) = lim FX (a + ) = lim P ({X ≤ a + }).
n→∞ n n→∞ n
Let En = {w ∈ S : X(w) ∈ (−∞, a + n1 ]}. Then En is an decreasing sequence of
events and Limn→∞ En = ∩∞ n=1 En = {w ∈ S : X(w) ∈ (−∞, a]}. Now by using
continuity of probability, we have
1
FX (a+) = lim P ({X ≤ a + })
n→∞ n
= lim P (En )
n→∞
= P (Limn→∞ En )
= P ({X ∈ (−∞, a]})
= P ({X ≤ a})
= FX (a)

(3) Let An = {w ∈ S : X(w) ∈ (−∞, −n]} and Bn = {w ∈ S : X(w) ∈ (−∞, n]}.

Then An and Bn are decreasing and increasing sequence of events, respectively.
Also Limn→∞ An = ∩∞ ∞
n=1 An = ∅} and Limn→∞ Bn = ∪n=1 Bn = {w ∈ S : X(w) ∈
R} = S. Therefore, by using continuity of probability, we have
FX (−∞) = lim FX (−n)
n→∞
= lim P ({X ∈ (−∞, −n]})
n→∞
= lim P (An )
n→∞
= P (Limn→∞ An )
= P (∅) = 0,
2
 and
FX (∞) = lim FX (n)
n→∞
= lim P ({X ∈ (−∞, n]})
n→∞
= lim P (Bn )
n→∞
= P (Limn→∞ Bn )
= P (S) = 1.


Remark 6. (1) Let En = {w ∈ S : X(w) ∈ (−∞, a − n1 ]} = {X ≤ a − n1 }. Then En
is an increasing sequence of events and Limn→∞ En = ∪∞n=1 En = {w ∈ S : X(w) ∈
(−∞, a)} = {X < a}. Now by using continuity of probability, we have
P ({X < a}) = P (Limn→∞ En )
= lim P (En )
n→∞
1
= lim P ({X ≤ a − })
n→∞ n
1
= lim FX (a − )
n→∞ n
= FX (a−).
Therefore, P ({X < a}) = FX (a−), ∀x ∈ R.

(2) For −∞ < a < b < ∞, we have

(a) P ({a < X ≤ b}) = P ({X ∈ ((−∞, b] − (−∞, a])}) = P ({X ≤ b}) − P ({X ≤
a}) = FX (b) − FX (a).

(b) P ({a < X < b}) = P ({X ∈ ((−∞, b) − (−∞, a])}) = P ({X < b}) − P ({X ≤
a}) = FX (b−) − FX (a).

(c) P ({a ≤ X < b}) = P ({X ∈ ((−∞, b) − (−∞, a))}) = P ({X < b}) − P ({X <
a}) = FX (b−) − FX (a−).

(d) P ({a ≤ X ≤ b}) = P ({X ∈ ((−∞, b] − (−∞, a))}) = P ({X ≤ b}) − P ({X <
a}) = FX (b) − FX (a−).

(3) For −∞ < a < ∞, we have

(a) P ({X ≥ a}) = P ({X ∈ (R − (−∞, a))}) = P ({X ∈ R}) − P ({X < a}) =
1 − FX (a−).

(b) P ({X > a}) = P ({X ∈ (R − (−∞, a]))}) = P ({X ∈ R}) − P ({X ≤ a}) =
1 − FX (a).

(4) The distribution function FX has atmost countable number of discontinuities.

Example 7. Let (S, Σ, P ) be a probability space. Define X : S −→ R by X(w) = c,
for all x ∈ S, where c is a fixed real number. Clearly, X is a random variable and the
cumulative distribution function of X is
(
0, x < c
FX (x) = P ({X ≤ x}) =
1, x ≥ c.
3
Example 8. Let (S, Σ, P ) be a probability space and E be an event. Define IE : S −→ R
by IE (w) = 1, if w ∈ E and IE (w) = 0, if w ∈
/ E. The function IE is called the indicator
function or characteristic function of E and is sometimes denoted by 1E or χE . We have

∅,
 a<0
−1 c
{IE ≤ a} = IE ((−∞, a]) = E , 0 ≤ a < 1

S, a ≥ 1.

Clearly, IE is a random variable and the cumulative distribution function of IE is


0,
 a<0
c
FIE (a) = P ({IE ≤ a}) = P (E ), 0 ≤ a < 1

1, a ≥ 1.
Example 9. Let S = {HHH, HHT , HT H, T HH, T T H, T HT, HT T, T T T } with P (E) =
number of elements in E
8
, for every E ∈ P(S). Let X : S −→ R be a random variable, defined
by X(w) = number of heads. Then the cumulative distribution function of X is


 0, x < 0
1
8, 0 ≤ x < 1


X 
FX (x) = P ({X ≤ x}) = P ({X = i}) = 21 , 1 ≤ x < 2
7, 2 ≤ x < 3

i∈{0,1,2,3}∩(−∞,x] 
8


1, x ≥ 3.
R∞
Example 10. Consider the probability space (R, BR , P ) with P (B) = e−t IB (t)dt, where
0
IB is the indicator function of B. Define X : R −→ R by

(
0, w≤0
X(w) = √
w, w > 0.
We have (
∅, x<0
{X ≤ x} = X −1 ((−∞, x]) = 2
(−∞, x ], x ≥ 0.
Thus X is a random variable. Now, the cumulative distribution function of X is
(
P (∅), x<0
FX (x) = P ({X ≤ x}) = 2
P ((−∞, x ]), x ≥ 0.
Thus 
0, x<0
 (
0, x<0
FX (x) = Rx2 −t = 2
 e dt,
 x≥0 1 − e−x , x ≥ 0.
0

Definition 11. A real-valued function F : R −→ R that is increasing, right continuous

and satisfies
F (−∞) = 0 and F (∞) = 1
is called a distribution function.
Theorem 12. Every distribution function is the distribution function of a random vari-
able on some probability space.