NPTEL

Course On

STRUCTURAL
RELIABILITY
Module # 02
Lecture 3

Course Format: Web

Instructor:
Dr. Arunasis Chakraborty
Department of Civil Engineering
Indian Institute of Technology Guwahati
3. Lecture 03: Theory of Probability
(Contd.)

Random Variables

Random variable is a variable whose values are always associated with a probability of
occurrence, i.e. the numerical value of the random variable cannot be predicted with certainty
before experiment.

Random variable ‘X’ in sample space ‘S’ is a set of real number. It may be discrete or
continuous.

Ex. : Let a die is thrown. The outcome is S  1,2,3,4,5,6. Draw the probability mass function
(pmf) and cumulative probability distribution function (CDF).

Discrete Random Variables

pmf,

PX  x  p X x  2.3.1

It satisfies three axioms of probability –

0  p X x   1 2.3.2

 p x   1
all xi
X i 2.3.3
all xi b
Pa  X  b   p x  X i 2.3.4
all xi  a

Jointly Distributed Discrete Random Variables

Two or more random variables

Course Instructor: Dr. Arunasis Chakraborty

1
 Lecture 03: Theory of Probability (Contd.)

Joint pmf,
 
p XY xi , y j   P X  xi , Y  y j  P X  x   Y  y  2.3.5

Joint CDF,

for Discrete RV

FXY x, y     p XY xi , y j  2.3.6

xi  x y j  y

for Continuous RV

FXY xi , y j   P X  x   Y  y  2.3.7

Note: p XY x, y   0 &   p x , y   1

all xi all yj
XY i j

Ex.: Five beams are tested in the laboratory and the load corresponding to first crack at SLS and
ULS is determined. The joint pmf is given in following table

Y
X Total
0 1 2 3 4 5
0 0.000 0.020 0.000 0.000 0.000 0.000 0.020
1 0.010 0.050 0.000 0.000 0.000 0.000 0.060
2 0.020 0.010 0.100 0.010 0.000 0.000 0.140
3 0.030 0.015 0.010 0.150 0.015 0.000 0.220
4 0.040 0.020 0.015 0.010 0.250 0.010 0.345
5 0.050 0.020 0.020 0.015 0.010 0.100 0.215
Total 0.150 0.135 0.145 0.185 0.275 0.110 1.000

Plot the pmf first and evaluate the probability of an event E in which same number of beams fail
in both SLS and ULS.

Marginal Distribution

Course Instructor: Dr. Arunasis Chakraborty

2
 Lecture 03: Theory of Probability (Contd.)

p X x   PX  x   p x , y 
all yj
XY i j 2.3.8

Note: It is possible to get distribution for the individual variable i.e. marginal distribution from
the joint distribution. Similar expression can be written for pY  y  .

Ex.: Evaluate marginal distribution from the previous example. (Task for the reader)

Conditional Distribution

Let us assume two discrete random variables ‘X’ and ‘Y’ with values x and y1. Then the
conditional mass function is given by

p XY x, y1 
p X |Y x | y1  
pY  y1 
2.3.9

The conditional pmf to be proper, following conditions are to be satisfied –

0  p X |Y x, y   1 2.3.10

 p x , y  1
all xi
X |Y i 2.3.11

pY  y1   0 2.3.12

Continuous RV

CDF

x
FX x   PX  x   f x dx
X for all x 2.3.13


Course Instructor: Dr. Arunasis Chakraborty

3
 Lecture 03: Theory of Probability (Contd.)
In the above expression f X x  is called the probability distribution function or pdf. If ‘X’ is
positive and is only defined in 0 to  , then CDF is

x
FX x   PX  x   f X x dx for all x in 0 to  2.3.14
0

FX x  to be a proper distribution function, the following conditions are to be satisfied –

FX    0 2.3.15

FX    1 2.3.16



 f x dx  1

X 2.3.17

f X x   0 2.3.18

FX x   0 and is a non-decreasing function in x 2.3.19

Note: For continuous random variable

b a
Pa  X  b   f X x dx   f x dx  F b  F a 
X X X 2.3.20
 

Ex: The strength of a concrete cube is known to randomly vary from 20 to 40kN/m2 with pdf

 y 
k 1   20  x  40
f X  x    40 
0 elsewhere

Where, ‘k’ is constant. Find out the probability of failure under the uniform load of 30kN/m2.

Jointly Distributed Continuous Random Variables

Let us assume two continuous random variables ‘X’ and ‘Y’. Then the joint probability density
function is given by –

Course Instructor: Dr. Arunasis Chakraborty

4
 Lecture 03: Theory of Probability (Contd.)
a2 b2

Pa1  X  a 2 , b1  Y  b2     f x, y dxdy

XY 2.3.21
a1 b1

Note: RHS is the volume under the pdf f XY x, y  .

The joint pdf f XY x, y  between ‘X’ and ‘Y’ satisfies the following conditions –

f XY x, y   0 for all x and y 2.3.22



  f x, y dxdy  1
  
XY 2.3.23

2
Note: f XY x, y   FXY x, y  & FXY x, y   PX  x, Y  y
xy

Marginal Distribution Function

Marginal pdf,


f X x    f x, y dy
XY 2.3.24


Marginal CDF,


FX x   PX  x   f x dx  F x, 
X XY 2.3.25


Conditional Distribution

Conditional pdf of ‘X’ such that ‘Y’ has taken a value of y1 is given by –

f XY x, y1 
f X |Y x, y1  
f Y  y1 
2.3.26

Course Instructor: Dr. Arunasis Chakraborty

5
 Lecture 03: Theory of Probability (Contd.)

x
FX |Y x, y1    f x, y dx
X |Y 1 2.3.27


Ex: If the joint pdf of two random variables ‘X’ and ‘Y’ is given by FXY x, y   xy for 0  x  1
and 0  y  2 , evaluate joint CDF FXY x, y  , marginal density function of ‘X’ and conditional
pdf of ‘Y’ f Y | X  y | x  .

Note: if two random variables ‘X’ and ‘Y’ are independent, then

p XY x, y   p X x . pY  y   p X |Y x | y   p X x  for DRV 2.3.28

f XY  x, y   f X  x . f Y  y 
FXY x, y   FX  x .FY  y 
f X |Y  x | y   f X  x  for CRV 2.3.29
fY|X  y | x  fY  y 
FX |Y  x | y   FX x 

DRV: Discrete random variable, CRV: Continuous random variable

Functions of Random Variables

In science and engineering, very often we encounter two random variables one of which depends
on another. For example, lateral pressure on a wall and water level in a tank. Let ‘Z’ be a
dependent random variable that depends on the independent random variable ‘X’ such that

Z  gX  & x  g 1
z   dx 

d g 1 z  
dz 2.3.30
dz

where, ‘Z’ is a single valued function of ‘X’ and ‘g’ is a monotonically increasing function then

PZ  z   PX  x  P X  g 1 z    2.3.31

For discrete random variable

Course Instructor: Dr. Arunasis Chakraborty

6
 Lecture 03: Theory of Probability (Contd.)


p Z z   p X g 1 z   2.3.32

FZ z    p x  X j 2.3.33
all x j  g 1  z 

For continuous random variable

FZ z   PZ  z   PX  x  FX g 1 z    2.3.34

 f g z  dz g z dz
x z
FZ z   f X x dx 
d

1 1
X 2.3.35
 


f Z z   f X g 1 z   dzd g 1
z   f X x 
dx
dz
2.3.36

 f Z z  
dx
dz

.nf X g 1 z   2.3.37

Note: Modulus is for monotonically decreasing function and ‘n’ is for a function where z
corresponds to ‘n’ values of ‘x’.

Ex: If T  d  h , where ‘T’ is the dependent random variable and ‘h’ is the independent random
variable whose pdf is given by

f H h   exp  h 2 2
1
for h  0
2

Evaluate pdf of ‘T’ i.e. f T t  .

Ans.: h  t  d    g 1 t  Using the transformation given in eq 2.3.37, one can get

Course Instructor: Dr. Arunasis Chakraborty

7
 Lecture 03: Theory of Probability (Contd.)

d t  d   1  t  d 2 
 
f T t    . f H h   .
1 1
 exp  .   td
dt     2  2   
 

Course Instructor: Dr. Arunasis Chakraborty

8