MAT 2230: Engineering Mathematics - IV

Probability
Addition rule

If A and B are two events of an experiment having sample space S, then

P( A  B) = P( A) + P( B) − P( A  B) .

The conditional probability of an event B, given that the event A already taken place is

P( A  B)
P( B / A) = , P( A)  0.
P( A)

Baye’s Theorem

Let B1 , B2 ,...Bk are the partitions of S with P ( Bi )  0, i = 1, 2,...k and A be any event of S, then

P( A / Bi ) P ( Bi ) .
P( Bi / A) = k

 P( A / B ) P( B )
i =1
i i

𝑃(𝐴)𝑃(𝐵|𝐴), 𝑖𝑓 𝑃(𝐴) ≠ 0
The multiplicative rule of probability : 𝑃(𝐴 ∩ 𝐵) = {
𝑃(𝐵)𝑃(𝐴|𝐵), 𝑖𝑓 𝑃(𝐵) ≠ 0
If 𝑃(𝐴 ∩ 𝐵) = 𝑃(𝐴)𝑃(𝐵), then A and B are independent.

Random Variable: Let S be the sample of space of a random experiment. Suppose with each
element s of S, a unique real number X is associated according to some rule then X is called random
variable. There are two types of random variable, i) Discrete and ii) Continuous.

Discrete Random Variable: A random variable X is said to be discrete, if the number of possible
values of X is finite or countably infinite. The probability distribution function (pdf) is named as
probability mass function (PMF). The Probability mass function is defined as, let X be a random
variable, hence the range space 𝑅𝑋 of consists of atmost a countably infinite number of values. The
probability mass function is defined as

𝑝(𝑥𝑖 ) = Pr {𝑋 = 𝑥𝑖 }, satisfying the conditions i) 𝑝(𝑥𝑖 ) ≥ 0 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑖

ii) ∑𝑘
𝑖=1 𝑝(𝑥𝑖 ) = 1.
Continuous Random Variable: A random variable X is said to be continuous if it can take all
possible values between certain limits, here the range space of X is infinite. Therefore the probability
distribution function named for such random variable is Probability density function (PDF), which is
defined as the pdf of X is a function 𝑓(𝑥) satisfying the following properties i) 𝑓(𝑥) ≥ 0
∞
ii) ∫−∞ 𝑓(𝑥)𝑑𝑥 =1
𝑏
iii) Pr{𝑎 ≤ 𝑋 ≤ 𝑏} = ∫𝑎 𝑓(𝑥)𝑑𝑥 for any a, b such that −∞ < 𝑎 < 𝑏 < ∞.
Note:

1. If X is a continuous random variable with pdf f(x), then

𝑏
𝑃(𝑎 < 𝑋 < 𝑏) = 𝑃(𝑎 ≤ 𝑋 < 𝑏) = 𝑃(𝑎 < 𝑋 ≤ 𝑏) = 𝑃(𝑎 ≤ 𝑋 ≤ 𝑏) = ∫𝑎 𝑓(𝑥)𝑑𝑥.

2. 𝑃(𝑋 = 𝑎) = 0, if X is a continuous random variable.

MAT 2260: Engineering Mathematics - IV 45

Cumulative distribution function: Let X be random variable (discrete or continuous), we define F to
be the cumulative distribution function of a random variable X given by 𝐹(𝑥) = Pr{𝑋 ≤ 𝑥}.

Case i) If X is discrete random variable then

𝐹(𝑡) = Pr{𝑋 ≤ 𝑡} = 𝑃(𝑥1 ) + 𝑃(𝑥2 ) + ⋯ + 𝑃(𝑡)
𝑥
Case ii) If x is a continuous random variable then 𝐹(𝑥) = Pr{𝑋 ≤ 𝑥} = ∫−∞ 𝑓(𝑥)𝑑𝑥.
Two dimensional random variable: Let E be an experiment and S be a sample space associated
with E. Let X=X(s) and Y=Y(s) be two functions each assigning a real number to each outcome s of S.
We call (X, Y) to be two dimensional random variable.

Discrete 2D: If the possible values of (X, Y) are finite or countably infinite then (X, Y) is called discrete
and it is defined as 𝑃(𝑥𝑖 , 𝑦𝑗 ) satisfying the following condition,

i) 𝑃(𝑥𝑖 , 𝑦𝑗 ) ≥ 0 and
ii) ∑∞ ∞
𝑗=1 ∑𝑖=1 𝑃(𝑥𝑖 , 𝑦𝑗 ) = 1. The function 𝑃(𝑥𝑖 , 𝑦𝑗 ) defined is called as Joint probability

distribution function (Jpdf).

Continuous 2D: If (X, Y) is a continuous random variable assuming all values in some region R of
the Euclidean plane, then the Joint probability density function 𝑓(𝑥, 𝑦) is a function satisfying the
following conditions

i) 𝑓(𝑥, 𝑦) ≥ 0 for all (x, y)𝜖𝑅

ii) ∬ 𝑓(𝑥, 𝑦)𝑑𝑥 𝑑𝑦 = 1 over the region R.

Marginal Probability distribution: The marginal probability distribution is defined as

Case i) In the discrete (X, Y), it is defined as𝑝(𝑥𝑖 ) = 𝑃{𝑋 = 𝑥𝑖 } = ∑∞

𝑗=1 𝑃(𝑥𝑖 , 𝑦𝑗 ) is the marginal
∞
probability distribution of X. Similarly 𝑞(𝑦𝑗 ) = 𝑃{𝑌 = 𝑦𝑗 } = ∑𝑖=1 𝑃(𝑥𝑖 , 𝑦𝑗 ) is the marginal
probability distribution of Y.

Case ii) In the continuous (X, Y), it is defined as the marginal probability function of X is defined as
∞ ∞
𝑔(𝑥) = ∫−∞ 𝑓(𝑥, 𝑦)𝑑𝑦 and the marginal probability function of Y is defined as ℎ(𝑦) = ∫−∞ 𝑓(𝑥, 𝑦)𝑑𝑥.
To calculate the conditional probability:
𝑃(𝑥𝑖 ,𝑦𝑗 )
Case i) Discrete: Probability of 𝑥𝑖 given 𝑦𝑗 is defined as = , 𝑞(𝑦𝑗 ) > 0
𝑞(𝑦𝑗 )

𝑃(𝑥𝑖 ,𝑦𝑗 )
Probability of 𝑦𝑗 given 𝑥𝑖 is defined as = , 𝑝(𝑥𝑖 ) > 0
𝑝(𝑥𝑖 )

𝑓(𝑥,𝑦)
Case ii) Continuous: The pdf of X for given Y=y is = , ℎ(𝑦) > 0
ℎ(𝑦)

𝑓(𝑥,𝑦)
The pdf off Y for given X=x is = , 𝑔(𝑥) > 0.
𝑔(𝑥)

Independent Random variable: If X and Y are independent random variable then two dimensional
random variable in case of discrete is defined as 𝑃(𝑥𝑖 , 𝑦𝑗 ) = 𝑝(𝑥𝑖 ). 𝑞(𝑦𝑗 ) for all the values of i and
j. In case of Continuous it is defined as 𝑓(𝑥, 𝑦) = 𝑔(𝑥). ℎ(𝑦).

Mathematical Expectation: If X is a discrete random variable with pmf p(x), then the expectation of X
is given by 𝐸(𝑋) = ∑𝑥 𝑥𝑝(𝑥), provided the series is absolutely convergent.

MAT 2260: Engineering Mathematics - IV 46

 If X is continuous with pdf f(x), then the expectation of X is given by 𝐸(𝑋) = ∫ 𝑥𝑓(𝑥)𝑑𝑥, provided
∫ |𝑥|𝑓(𝑥)𝑑𝑥 < ∞.
2
Variance of X is given by 𝑉(𝑋) = 𝐸(𝑋 − 𝐸(𝑋))2 = 𝐸(𝑋 2 ) − (𝐸(𝑋)) .

DISTRIBUTIONS
Distribution PMF/PDF Mean Variance
Binomial P( x) = Ck p (1 − p)
n k n−k
, k = 0,1, 2,..., n E ( x) = np V ( x) = np(1 − p)
distribution
X ~ B(n, p)

Poisson’s e −  k E ( x) =  = np V ( x) =  = np
Distribution P( x) = , k = 0,1, 2,...,   0
X ~ P( ) k!
Uniform  1 b+a (b − a ) 2
, a xb E ( x) = V ( x) =
f ( x) =  b − a
Distribution
2 12
X ~ U ( a, b) 
 0, otherwise
Normal −1 ( x −  )2 E ( x) =  V ( x) =  2
1
Distribution f ( x) = e2 2
, −   x ,   ,   0
X ~ N ( , 2 )  2
1
Exponential  e−  x , x0 E ( x) =
1 𝑉(𝑋) =
Distribution f ( x) =  𝜆2
X ~ E ( ) 
 0, otherwise
Gamma  x r −1e− x r E ( x) =
r
V ( x) =
r
, x  0,  , r  0
f ( x) =  (r )
Distribution
X ~ G (r ,  )  2
 0, elsewhere
Chi-square  n
−1 −
x E ( x) = n V ( x ) = 2n

2 2
Distribution x e
, x0
X ~  2 ( n) f ( x) =  n

 (n / 2)2 2
0,
 elsewhere

Uniform distribution on a two dimensional set: If R is a set in the two-dimensional plane, and R has a
finite area, then we may consider the density function equal to the reciprocal of the area of R inside R,
and equal to 0 otherwise:
1
; 𝑖𝑓 (𝑥, 𝑦) ∈ 𝑅
𝑓(𝑥, 𝑦) = {𝑎𝑟𝑒𝑎 𝑅 .
0 𝑂𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

Chebyshev’s inequality:

Let x be random variable with mean  and variance  2 then for any positive real number k(k>0)

2
p X −   k   (Upper bound)
k2

2
p X −   k   1 − (Lower bound)
k2

MAT 2260: Engineering Mathematics - IV 47

Note: some other forms

p X −   k   px −   k   1 (Upper bound)

1
1. and
k2
p X −    E ( x − c) 2 and px −    1
1
2.
2

Covariance:
Cov ( X , Y ) = E ( XY ) − E ( X ) E (Y )

Correlation coefficient:
E ( XY ) − E ( X ) E (Y )
 xy =  =
V ( X )V (Y )

Properties:
1. 𝐸(𝑐) = 𝑐, where c is a constant.
2. 𝑉(𝑐) = 0, where c is a constant.
3. If E ( XY ) = 0 then X and Y are orthogonal.

4. V ( AX + B ) = A 2V ( X ) when AX+B is linear function of X.

5. If  xy = 0 then X and Y are un correlated.

6. V ( AX + BY ) = A 2V ( X ) + B 2V (Y ) + 2 ABCOV ( X , Y )

Regression:
If the straight line chosen such that the sum of the squares of errors parallel to y axis is minimum then
it is called line of regression of y on x and it gives the best estimate of y for any given value of x. It’s
given by

y y
y−y= ( x − x) , b yx =  is the regression coefficient of Y on X
x x

Therefore y − y = b yx ( x − x)

Similarly line of regression of x on y is given by

x x
x−x =  ( y − y) , bxy =  is the regression coefficient of X on Y
y y

Therefore x − x = bxy ( y − y )

Note:

x x
2 2

 2
= −  = E ( x 2 ) − ( E ( x)) 2 is the variance of X
x
n  n 
 

MAT 2260: Engineering Mathematics - IV 48

Similarly

y y
2 2

 2
= −  = E ( y 2 ) − ( E ( y )) 2 is the variance of Y
y
n  n 
 
E ( xy) − E ( x) E ( y )
=
V ( x)V ( y )

Also,  =  bxy  b yx . The correlation coefficient is positive if the two regression coefficients are
positive, and negative if the two regression coefficients are negative.

FUNCTIONS OF ONE DIMENSIONAL RANDOM VARIABLES

Let 𝑆 be a sample space associated with a random experiment 𝐸, then it is known that a random variable
𝑋 on 𝑆 is a real valued function, i.e., : 𝑆 → 𝑅 , for each element s  S, there is a real number associated.

Let 𝑋 be a random variable defined on 𝑆. Let 𝑦 = 𝐻(𝑥) is a real valued function of 𝑥. Then 𝑌 = 𝐻(𝑋) is
a random variable on 𝑆. i.e., for each element s  S, there is a real number associated, say 𝑦 = 𝐻(𝑋(𝑠)).
Here 𝑌 is called a function of the random variable 𝑋.

Notations:

1. 𝑅𝑋 – the set of all possible values of the function 𝑋, called the range space of the random
variable 𝑋.
2. 𝑅𝑌 – the set of all possible values of the function 𝑌 = 𝐻(𝑋), called the range space of the
random variable 𝑌.

Equivalent Events: Let C be an event associated with the range space 𝑅𝑌 . Let 𝐵 ⊂ 𝑅𝑋 defined by 𝐵 =
{𝑥 ∈ 𝑅𝑋 ; 𝐻(𝑥) ∈ 𝐶}, then 𝐵 and 𝐶 are called equivalent events.

Distribution function of functions of random variables:

Case 1: Let 𝑋 be a discrete random variable with p.m.f. 𝑝(𝑥𝑖 ) = 𝑃(𝑋 = 𝑥𝑖 ) for 𝑖 = 1,2,3, … Let 𝑌 =
𝐻(𝑋) then 𝑌 is also a discrete random variable. If 𝑌 = 𝐻(𝑋) is a one to one function then the probability
distribution of 𝑌 is as follows:

For the possible values of 𝑦𝑖 = 𝐻(𝑥𝑖 ) for 𝑖 = 1,2,3, …. The p.m.f. of 𝑌 = 𝐻(𝑋) is 𝑞(𝑦𝑖 ) = 𝑃(𝑌 = 𝑦𝑖 ) =
𝑃(𝑋 = 𝑥𝑖 ) = 𝑝(𝑥𝑖 ) for 𝑖 = 1,2,3, ….

Case 2: Let 𝑋 be a discrete random variable with p.m.f. 𝑝(𝑥𝑖 ) = 𝑃(𝑋 = 𝑥𝑖 ) for 𝑖 = 1,2,3, … Let 𝑌 = 𝐻(𝑋)
then 𝑌 is also a discrete random variable. Suppose tat for one value of 𝑌 = 𝑦𝑖 there corresponds several
values of 𝑋 say 𝑥𝑖1 , 𝑥𝑖2 , … , 𝑥𝑖𝑗 , … then the p.m.f. of of 𝑌 = 𝐻(𝑋) is

𝑞(𝑦𝑖 ) = 𝑃(𝑌 = 𝑦𝑖 ) = 𝑝(𝑥𝑖 1 ) + 𝑝(𝑥𝑖2 ) + ⋯ + 𝑝 (𝑥𝑖𝑗 ) + ⋯

Case 3: Let 𝑋 be a continuous random variable with p.d.f. 𝑓(𝑥). Let 𝑌 = 𝐻(𝑋) be a discrete random
variable. Then if the set {𝑌 = 𝑦𝑖 } is equivalent to an event 𝐵 ⊆ 𝑅𝑋 then the p.m.f. of 𝑌 is

𝑞(𝑦𝑖 ) = 𝑃(𝑌 = 𝑦𝑖 ) = ∫ 𝑓(𝑥) 𝑑𝑥

𝐵

Case 4: Let 𝑋 be a continuous random variable with p.d.f. 𝑓(𝑥). Let 𝑌 = 𝐻(𝑋) be a continuous random
variable. Then the p.d.f. of 𝑌, say 𝑔 is obtained by the following procedure:

MAT 2260: Engineering Mathematics - IV 49

 Step 1: Obtain the c.d.f. of 𝑌, 𝐺(𝑦) = 𝑃(𝑌 = 𝑦), by finding the event

𝐴 ⊆ 𝑅𝑋 , which is equivalent to the event {𝑌 = 𝑦𝑖 }.

Step 2: Differentiate 𝐺(𝑦) with respect to 𝑦 to get 𝑔(𝑦).

Step 3: Determine those values of 𝑦 in 𝑅𝑌 for which 𝑔(𝑦) > 0.

Theorem: Let 𝑋 be a continuous random variable with p.d.f. 𝑓(𝑥) where 𝑓(𝑥) > 0 for 𝑎 < 𝑥 < 𝑏.
Suppose that 𝑌 = 𝐻(𝑋) is strictly monotonic function on [𝑎, 𝑏]. Then the p.d.f. of the random variable
𝑌 = 𝐻(𝑋) is given by

𝑑𝑥
𝑔(𝑦) = 𝑓(𝑥) | |
𝑑𝑦
If 𝑌 = 𝐻(𝑋) is strictly increasing then 𝑔(𝑦) > 0 for 𝐻(𝑎) < 𝑦 < 𝐻(𝑏).

If 𝑌 = 𝐻(𝑋) is strictly decreasing then 𝑔(𝑦) > 0 for 𝐻(𝑏) < 𝑦 < 𝐻(𝑎).

Theorem: Let 𝑋 be a continuous random variable with p.d.f. 𝑓(𝑥). Let 𝑌 = 𝑋 2 then the p.d.f. of 𝑌 is

1
𝑔(𝑦) = [𝑓(√𝑦) + 𝑓(−√𝑦) ]
2√𝑦

FUNCTIONS OF TWO DIMENSIONAL RANDOM VARIABLES

Let (𝑋, 𝑌) be a two dimensional continuous random variable. Let 𝑍 = 𝐻(𝑋, 𝑌) be a continuous function
of X and Y then 𝑍 = 𝐻(𝑋, 𝑌) is a continuous one dimensional random variable.

To find the p.d.f. of 𝑍, we introduce another suitable random variable say,

𝑊 = 𝐺(𝑋, 𝑌) and obtain the joint p.d.f. of the two dimensional random variable (𝑍, 𝑊), say 𝑘(𝑧, 𝑤). From
this distribution, the p.d.f. of 𝑍 can be obtained by integrating 𝑘 with respect to 𝑤.

Theorem: Suppose (𝑋, 𝑌) is a two dimensional continuous random variable with joint p.d.f. 𝑓(𝑥, 𝑦)
defined on a region 𝑅 of the XY-plane. Let 𝑍 = 𝐻1 (𝑋, 𝑌) and 𝑊 = 𝐻2 (𝑋, 𝑌). Suppose that 𝐻1 and 𝐻2
satisfies the following conditions;

(i) 𝑧 = 𝐻1 (𝑥, 𝑦) and 𝑤 = 𝐻2 (𝑥, 𝑦) may be uniquely solved for 𝑥, 𝑦 in terms of 𝑧 & 𝑤 say,
𝑥 = 𝐺1 (𝑧, 𝑤) and 𝑦 = 𝐺2 (𝑧, 𝑤).
𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦
(ii) The partial derivatives , , 𝜕𝑧 , 𝜕𝑤 exist and are continuous
𝜕𝑧 𝜕𝑤

Then the joint p.d.f. of (𝑍, 𝑊) say 𝑘(𝑧, 𝑤) is given by,

𝑘(𝑧, 𝑤) = 𝑓[𝐺1 (𝑧, 𝑤), 𝐺2 (𝑧, 𝑤)]|𝐽(𝑧, 𝑤)|

𝜕𝑥 𝜕𝑥
𝜕𝑧 𝜕𝑤
where 𝐽(𝑧, 𝑤) = | 𝜕𝑦 𝜕𝑦
| is called the Jacobian of the transformation
𝜕𝑧 𝜕𝑤

(𝑥, 𝑦) ↦ (𝑧, 𝑤). Also, 𝑘(𝑧, 𝑤) > 0 for those values of (𝑧, 𝑤) corresponding to the values of (𝑥, 𝑦) for
which 𝑓(𝑥, 𝑦) > 0.

MAT 2260: Engineering Mathematics - IV 50

MOMENT GENERATING FUNCTION (M.G.F.) OF ONE DIMENSIONAL RANDOM
VARIABLES
Let 𝑋 be any one dimensional random variable then the mathematical expectation 𝐸(𝑒 𝑡𝑋 ) if exists then
it is called the moment generating function (m.g.f.) of 𝑋.

i.e., 𝑀𝑋 (𝑡) = 𝐸(𝑒 𝑡𝑋 )

In particular, if 𝑋 is discrete then, 𝑀𝑋 (𝑡) = ∑𝑖=∞

𝑖=1 𝑒
𝑡𝑥𝑖
𝑃(𝑋 = 𝑥𝑖 ).
∞
If 𝑋 is continuous then, 𝑀𝑋 (𝑡) = ∫−∞ 𝑒 𝑡𝑥 𝑓(𝑥) 𝑑𝑥.

Properties of m.g.f.: Let 𝑋 be any one dimensional random variable and 𝑀𝑋 (𝑡) be the m.g.f. of 𝑋 then

1. 𝑀𝑋𝑛 (0) = 𝐸(𝑋 𝑛 ) where 𝑀𝑋𝑛 (0) is the nth derivative of 𝑀𝑋 (𝑡) at 𝑡 = 0.
i.e.; 𝑀𝑋′ (0) = 𝐸(𝑋)
𝑀𝑋′′ (0) = 𝐸(𝑋 2 )
2
2. 𝑉(𝑋) = 𝑀𝑋′′ (0) − (𝑀𝑋′ (0))
3. Let 𝑋 be any one dimensional random variable and 𝑀𝑋 (𝑡) be the m.g.f. of 𝑋. Let 𝑌 = 𝛼𝑋 + 𝛽.
Then the m.g.f. of 𝑌 is
𝑀𝑌 (𝑡) = 𝑒 𝛽𝑡 𝑀𝑋 (𝛼𝑡).
4. Suppose that 𝑋 and 𝑌 are independent random variables. Let 𝑍 = 𝑋 + 𝑌. Let 𝑀𝑋 (𝑡), 𝑀𝑌 (𝑡)
and 𝑀𝑍 (𝑡) be the m.g.f.’s of the random variables 𝑋, 𝑌 and 𝑍 respectively. Then
𝑀𝑍 (𝑡) = 𝑀𝑋 (𝑡)𝑀𝑌 (𝑡)
5. Let 𝑋1 , 𝑋2 , … , 𝑋𝑛 be 𝑛 independent random variables which follows a normal distribution
𝑁(𝜇𝑖 , 𝜎𝑖2 ) for 𝑖 = 1,2,3, . . , 𝑛. Let 𝑍 = 𝑋1 + 𝑋2 + ⋯ + 𝑋𝑛 then
𝑍 → 𝑁(𝜇1 + 𝜇2 + ⋯ + 𝜇𝑛 , 𝜎12 + 𝜎22 + ⋯ + 𝜎𝑛2 ).
6. Let 𝑋1 , 𝑋2 , … , 𝑋𝑛 be 𝑛 independent random variables which follows a Poisson distribution with
parameter 𝛼𝑖 for 𝑖 = 1,2,3, . . , 𝑛. Let 𝑍 = 𝑋1 + 𝑋2 + ⋯ + 𝑋𝑛 then 𝑍 has a Poisson distribution with
parameter
𝛼 = 𝛼1 + 𝛼2 + ⋯ + 𝛼𝑛 .
7. Let 𝑋1 , 𝑋2 , … , 𝑋𝑘 be 𝑘 independent random variables which follows a Chi-square distribution with
degrees of freedom 𝑛𝑖 for 𝑖 = 1,2,3, . . , 𝑘. Let 𝑍 = 𝑋1 + 𝑋2 + ⋯ + 𝑋𝑘 then 𝑍 has a Chi-square
distribution with degrees of freedom
𝑛 = 𝑛1 + 𝑛2 + ⋯ + 𝑛𝑘 .
8. Let 𝑋1 , 𝑋2 , … , 𝑋𝑘 be 𝑘 independent random variables, each having distribution 𝑁(0,1). Then 𝑆 =
𝑋12 + 𝑋22 + ⋯ + 𝑋𝑘2 has a Chi-square distribution with degrees of freedom 𝑘.
9. Let 𝑋1 , 𝑋2 , … , 𝑋𝑟 be 𝑟 independent random variables, each having exponential distribution with
the same parameter 𝛼. Let 𝑍 = 𝑋1 + 𝑋2 + ⋯ + 𝑋𝑟 then 𝑍 has a Gamma distribution with
parameters 𝛼 and 𝑟.

MAT 2260: Engineering Mathematics - IV 51

 10. Let 𝑋1 , 𝑋2 , … , 𝑋𝑛 , … be a sequence of random variable with c.d.f.’s 𝐹1 , 𝐹2 , … , 𝐹𝑛 , … and m.g.f.’s
𝑀1 , 𝑀2 , … , 𝑀𝑛 , … Suppose that lim 𝑀𝑛 (𝑡) = 𝑀(𝑡), where 𝑀(0) = 1. Then 𝑀(𝑡) is the m.g.f. of
𝑛→∞

the random variable 𝑋 whose c.d.f is 𝐹 = lim 𝐹𝑛 (𝑡).

𝑛→∞

MGF of some standard distributions:

1. Binomial Distributions: 𝑀𝑋 (𝑡) = 𝑀𝑋 (𝑡) = (𝑝𝑒 𝑡 + 𝑞)𝑛

𝑡 −1)
2. Poisson Distributions: 𝑀𝑋 (𝑡) = 𝑒 ∝(𝑒
𝜎2𝑡2
𝑡𝜇+
3. Normal Distributions: 𝑀𝑋 (𝑡) =𝑒 2

∝
4. Exponential Distributions: 𝑀𝑋 (𝑡) = ∝−𝑡
∝𝑟
5. Gamma Distributions: 𝑀𝑋 (𝑡) = (∝−𝑡)𝑟

6. Chi square Distributions: 𝑀𝑋 (𝑡) = (1 − 2𝑡)−𝑛⁄2

Sampling Theory
In statistical investigation, the characteristics of a large group of individuals (called population) is
studied. Sampling is a study of the relationship between a population and samples drawn from it.

The population mean and the population variance are denoted by 𝜇 and 𝜎 2 respectively.

Sample mean and sample variance: Let 𝑋 be the random variable which denotes the population
with mean 𝜇 and variance 𝜎 2 . Let (𝑋1 , 𝑋2 , … , 𝑋𝑛 ) be a random sample of size 𝑛 from 𝑋. Then,

∑ 𝑛
𝑋𝑖
Sample mean, 𝑋̅ = 𝑖=1 and
𝑛

∑𝑛 ̅ 2
𝑖=1(𝑋𝑖 −𝑋)
Sample variance, 𝑠 2 = 𝑛

• If 𝑋 → 𝑁(𝜇, 𝜎 2 ) then 𝑋̅ and 𝑠 2 are independent random variables.

• Let 𝑋 be the random variable with 𝐸(𝑋) = 𝜇 and 𝑉(𝑋) = 𝜎 2 . Let (𝑋1 , 𝑋2 , … , 𝑋𝑛 ) be a random
𝜎 2
sample of size 𝑛 from 𝑋. Then, 𝐸(𝑋̅) = 𝜇 and 𝑉(𝑋̅) = .
𝑛
𝜎2
• Let 𝑋 → 𝑁(𝜇, 𝜎 2 ) then 𝑋̅ → 𝑁(𝜇, ) and 𝑠 2 → 𝜒 2 (𝑛 − 1).
𝑛

Central Limit Theorem: Let 𝑋1 , 𝑋2 , … , 𝑋𝑛 be 𝑛 independent random variables all of which have the
same distribution. Let 𝜇 = 𝐸(𝑋𝑖 ) and 𝜎 2 = 𝑉(𝑋𝑖 ) be the common expectation and variance. Let 𝑆 =
𝑆−𝐸(𝑆)
∑𝑛𝑖=1 𝑋𝑖 then 𝐸(𝑆) = 𝑛𝜇 and 𝑉(𝑆) = 𝑛𝜎 2 then for large values of 𝑛, the random variable 𝑇𝑛 = has
√𝑉(𝑆)
approximately the distribution 𝑁(0,1).

Curve fitting
Let y = f(x) be the curve to be fit with n set of points and Y = F(x) is actual values, then the estimated
error is 𝑒𝑖 = 𝑦𝑖 − 𝑌𝑖 .

i) Fitting a straight line: For a straight line 𝑦 = 𝑎𝑥 + 𝑏, where a, b are the parameters, the normal
equations are to find a and b is given by

MAT 2260: Engineering Mathematics - IV 52

 ∑ 𝑌 = 𝑎 ∑ 𝑋 + 𝑛𝑏
∑ 𝑋𝑌 = 𝑎 ∑ 𝑋 2 + 𝑏 ∑ 𝑋

ii) Fitting a Parabola: For a parabola 𝑦 = 𝑎𝑥 2 + 𝑏𝑥 + 𝑐, the normal equations are given by

∑ 𝑌 = 𝑎 ∑ 𝑋 2 + 𝑏 ∑ 𝑋 + 𝑛𝑐
∑ 𝑋𝑌 = 𝑎 ∑ 𝑋 3 + 𝑏 ∑ 𝑋 2 + 𝑐 ∑ 𝑋

∑ 𝑋 2𝑌 = 𝑎 ∑ 𝑋 4 + 𝑏 ∑ 𝑋 3 + 𝑐 ∑ 𝑋 2

iii) Fitting a curve of the form 𝑦 = 𝑎𝑒 𝑏𝑥 , the normal equations are given by taking log on both
sides, 𝑙𝑜𝑔𝑒 𝑦 = 𝑙𝑜𝑔𝑒 (𝑎𝑒 𝑏𝑥 )
𝑙𝑜𝑔𝑒 𝑦 = 𝑙𝑜𝑔𝑒 𝑎 + 𝑏𝑥, Now consider 𝑌 = 𝑙𝑜𝑔𝑒 𝑦, 𝐴 = 𝑙𝑜𝑔𝑒 𝑎, 𝑏 = 𝐵, 𝑋 = 𝑥.
Therefore equation is 𝑌 = 𝐴𝑋 + 𝐵.

Special functions
𝑥 1
1. Bessel function 𝐽𝑛 (𝑥) = ∑∞ 𝑟
𝑟=0(−1) (2)
𝑛+2𝑟
𝑟! 𝛾(𝑛+𝑟+1)

2. Recurrence formula for 𝐽𝑛 (𝑥)

𝑑
(i) (𝑥 𝑛 𝐽𝑛 (𝑥)) = 𝑥 𝑛 𝐽𝑛−1 (𝑥)
𝑑𝑥
𝑑
(ii) (𝑥 −𝑛 𝐽𝑛 (𝑥)) = −𝑥 −𝑛 𝐽𝑛+1 (𝑥)
𝑑𝑥
𝑥
(iii) 𝐽𝑛 (𝑥) = 2𝑛 { 𝐽𝑛−1 (𝑥) + 𝐽𝑛+1 (𝑥)}
1
(iv) 𝐽𝑛 ′ (𝑥) = { 𝐽𝑛−1 (𝑥) − 𝐽𝑛+1 (𝑥)}
2
′ 𝑛
(v) 𝐽𝑛 (𝑥) = 𝑥 𝐽𝑛 (𝑥) − 𝐽𝑛+1 (𝑥)
2𝑛
(vi) 𝐽𝑛+1 (𝑥) = 𝐽𝑛 (𝑥) − 𝐽𝑛−1 (𝑥)
𝑥

3. Orthogonality of Bessel’s function:

 0, 

1

0 xJ n ( x)J n ( x)dx =  1 [ J n+1 ( )]2  =  , where  ,  are the roots of J n ( x) = 0

 2
4. The Bessel’s function of various orders can be derived as coefficients of different powers of t
1 t −1 
= t
x( ) n
in the expansion e 2 t
n =−
J n ( x ) , called as generating function of Bessel functions.

d2y dy
5. The differential equation of the form (1 − x ) 2 − 2 x + n(n + 1) y = 0
2
is the
dx dx
Legendre’s equation whose general solution when n is a positive integer, is polynomial solution
and an infinite series solution. The polynomial solution is called Legendre’s polynomial of order
n and infinite series solution is called Legendre’s function of the second kind.

MAT 2260: Engineering Mathematics - IV 53

 6. The Legendre’s polynomials are given by the Rodrigues formula as
1 dn 2
Pn ( x) = n n
( x − 1) n
n !2 dx
−
1 
7. Generating function of (1 − 2 xt + t 2 ) 2
=  t n Pn ( x)
n =0

8. Recurrence formulae for Pn ( x ) :

(i) (n + 1) Pn +1 ( x) = (2n + 1) xPn ( x) − nPn −1 ( x)

(ii) (n) Pn ( x) = xP 'n ( x) − P 'n −1 ( x)

(iii) ( (2n + 1) Pn ( x) = P 'n +1 ( x) − P 'n −1 ( x)

(iv) P 'n ( x) = xP 'n −1 ( x) + nPn −1 ( x)

(v) (1 − x2 ) P 'n ( x) = (n)[ Pn−1 ( x) − xPn ( x)]

MAT 2260: Engineering Mathematics - IV 54