Moment Generating function

• Let f(x,y) denote the joint probability density

function of the two random variables X and Y. If
E[e t1 X t2Y ]
exists for –h1 < t1 < h1 ,-h2 < t2 < h2 ,
where h1 and h2 are positive, it is denoted by M(t1
, t2) and is called the moment-generating function
of the joint distribution of X and Y.
t X
Hence, the Marginal
M (t1 ,0)  Distribution
E[ e ] M (t1 ) of X and Y are
1

M (0, t 2 ) E[e t2Y ] M (t 2 )

and K m   k m

E ( X Y )  k m M (t1 , t 2 )
Also,  t1 t 2  t1 0,t2 0
 Independence of Two Random
Variables
• Two random variables X and Y, forming a
discrete random variable, are independent if
and only if: pij  p* j pi *
where pij is their joint probability mass
function and pi* and p* j are their marginal
probability mass functions.
 Independence of Two Random
Variables
• Two random variables X and Y, forming an
absolutely continuous random variable, are
independent if and only if: f ( x, y )  f X ( x) fY ( y )
where f ( x, y ) is their joint probability mass
function and f X (x) and fY ( y ) are their marginal
probability mass functions.
 Independence of Two Random
Variables
Let the stochastically independent random variables
X and Y have the marginal probability density
functions f X (x) and fY ( y ) , respectively. Then
E[ XY ] E[ X ]E[Y ]
• Let e  y , 0  x  y  
f ( x, y ) 
0, elsewhere
be the joint probability density function of X and
Y. Find the moment generating function of this
distribution.
Moment Generating Function of X and Y,
M (t1 , t 2 ) E[e t1 X t2Y ]
 y
e t1x t2 y f ( x, y )dxdy
0 0
 y
e t1x t2 y e  y dxdy
0 0
 y
e t1x (t2  1) y dxdy
0 0

 y
e 
t1 x
e ( t 2  1) y
  dy
0  t1  0
 y
 e  1
t1 y
e ( t 2  1) y
  dy
0  t1  0
 
1
 
M (t1 , t 2 )  e  (1 t1  t2 ) y  e  (1 t2 ) y dy
t1 0
1 1 1 
   
t1  1  t1  t 2 1  t 2 
1

(1  t1  t 2 )(1  t 2 )
Discrete Probability Distributions
 Binomial Distribution
If X is discrete random variable which can take
values 0,1,2,3,…,n such that P( X x) nC x p x q n x ,
x = 0,1,2,….,n where p  q 1 then X is said to
follow a Binomial distribution with parameters
n and p.
 Moment Generating Function of
Binomial Distribution
M (t )  E[e tX ]
n
 e tx p x
x 0
n
 e tx nC x p x q n  x
x 0
n n
 nC x ( pe t ) x q n  x ,  (a  b) n  nC r a r b n  r
x 0 r 0

( pe t  q )
Derivatives of MGF,
M (t ) n( pet  q ) n  1 pet
M (t ) np[( pet  q ) n  1 e t  (n  1)( pet  q ) n  2 pe 2t ]
 Mean and Variance of Binomial
Distribution
E ( X )  M (0) np
E ( X 2 )  M (0) np[1  (n  1) p ]
Var ( X )  E ( X 2 )  [ E ( X )]2 npq
Mean :np
Variance : npq
• In a large consignment of electric bulbs 10 % are
defective. A random sample of 20 is taken for
inspection. Find the probability that (i) All are
good bulbs (ii) at most 3 are defective bulbs (iii)
Exactly there are three defective bulbs.
• Let X be the event of defective bulbs,
p = 0.1 q = 0.9 n = 20
(i)P(X = 0) = 20C0 (0.1)0 (0.9)20 = 0.1216

(ii)P(X  3) = p(0) + p(1) + p(2) + p(3)

= 20C0 (0.1)0 (0.9)20 +20C1 (0.1)1 (0.9)19

+ 20C2 (0.1)3 (0.9)18 + 20C3 (0.1)3 (0.9)17

= 0.8666

• P(X = 3) = 20C3 (0.1)3 (0.9)17 = 0.19.

• Find the probability that in tossing 4 coins one
will get either all heads or all tails for the third
time on the seventh toss.
P(H H H H) = 1/16; P(T T T T) = 1/16
P(all head  all tail) = 1/16 + 1/16 = 1/8
 p = 1/8, q =7/8 ; x + r = 7, r = 3

P( X  x) ( x  r  1)C r  1 p r q x

3 4
 1  7
P( X 4) 7  1C3 1    
 8  8
3 4
 1  7
6C2    
 8  8
0.0169
• In a company 5% defective components are
produced. What is the probability that atleast 5
components are to be examined in order to get
three defectives?
Given, p = 0.05, q =0.95 ; x + r ≤ 5, r = 3
P( X  x) ( x  r  1)C r  1 p r q x

P( X  2) = 1 – P(X < 2)
= 1 – P(X = 0) – P(X = 1)
= 1 – 2C2 (0.05)3 (0.95)0 – 3C2 (0.05)3 (0.95)1
= 0.9995.
 Poisson Distribution
If X is a discrete random variable that can assume
the values 0,1,2,… such that its probability mass
function is given by
e   x
P( X  x) 
x! x = 0,1,2,….;  > 0.
Then X is said to follow a Poisson distribution with
parameter  .
 Poisson distribution is a limiting case of
binomial distribution under the following
assumptions.
• The number of trials ‘n’ should be indefinitely
large. i.e., n  .
• The probability of successes ‘p’ for each trial is
indefinitely small.
• np = , should be finite where  is a constant.
Moment Generating Function of
Poisson Distribution

M (t )  e tx p x
x 0
 x

e 
 e tx
x 0 x!
(e t  ) x
 
an
e  
,  e 
a

x 0 x! n 0 n!
  et 
e e
 ( e t  1)
e
 Mean and Variance of Poisson
Distribution
  et 
M (t ) e e e t
M (t ) e 
e (e
t et 
e )  e e
t et  t

  e0
E[ X ] M (0) e e e 0  
E[ X ] M (0) e
2 
e (e
0 e0 0

e )  e e e   e   e   2  
e0 0

Mean , E[ X ] 
Variance, Var [ X ] E[ X 2 ]  E[ X ]2 2    2 
• A random variable X follows Poisson distribution
and if P(X=1) = 2P(X=2),
find (i) P(X = 0)
(ii) S.D. of X.
 Given that P(X=1) = 2P(X=2)
e   1 e   2
2
1! 2!
  1
e   0
(i) P ( X 0)  e  1 0.3679
0!

(ii) S.D. of X  var X   1

Continuous Probability
Distribution
 Normal Distribution
Let X be a continuous random variable having
the probability density function
2
1  x a 
1  
2 b 

f ( x)  e ,    x  , b  0
b 2
Moment Generating Function of
Normal Distribution
The moment generating function is
b 2t 2
at 
M (t ) e 2

and
b 2t 2
at 
M (t ) (a  b 2 t )e 2

b 2t 2 b 2t 2
at  at 
M (t ) (a  b 2 t ) 2 e 2
 b2e 2

Mean,   E[ X ]  M (0) a
E[ X 2 ]  M (0) a 2  b 2
Variance,  2 Var [ X ]  E[ X 2 ]  E[ X ] 2 a 2  b 2  a 2 b 2
• If the random variable X is n(µ,2) then the
random variable W=(X - µ)/ is n(0,1)
Let G(w) and g(w) be the distribution and density
function of W and W=(X - µ)/.
G ( w)  P[W  w]
X  
P  w
  
 P[ X    w]
2
x  w 1  x  
1   


2  
 e dx
x   2

Let, y 
x 
i .e , x    y , dx  dy

when x   , y   ; x    w , y  w
Hence
y w 1 2
1  y
G ( w)  
y  2
e 2
 dy

1
1  w2
g ( w) G ( w)  e 2

2
2
1  w 0 
1  
2 1 

 e ,  w
2

which is n(0,1)