Introduction

Mojtaba Mahsuli
mahsuli@sharif.edu
Structural Reliability and Probabilistic Modeling
UNCERTAINTY
Nature of uncertainty
Sources of uncertainty
Describing uncertainty
Concept of probability
History of probability
Describing probability
Typical probabilities
Risk and reliability
2
Laplaces Demon
A Philosophical Essay on Probabilities
Pierre-Simon Laplace (1814)

We may regard the present state of the
universe as the effect of its past and the
cause of its future. An intellect which at a
certain moment would know all forces
that set nature in motion, and all
positions of all items of which nature is
composed, if this intellect were also vast
enough to submit these data to analysis,
it would embrace in a single formula the
movements of the greatest bodies of the
universe and those of the tiniest atom;
for such an intellect nothing would be
uncertain and the future just like the past
would be present before its eyes.
3


Solvay Conference, 1927
4
God does not
play dice!
Einstein! Stop
telling God what
to do.
2
p _
o o >
Describing Uncertainties
Probability
Most prevalent
A number between 0 and 1
Examples:
Probability that an earthquake occurs with M8.0
Probability that the drift of a structure exceeds 2%
Fuzzy approaches
Possibilistic methods


Entropy
A measure of the unpredictability of a phenomenon
Very low entropy: Almost unpredictable
Very high entropy: Very predictable

5
Meaning of Probability
A number between 0 and 1
0 No occurrence
1 Certain occurrence


Two schools of thought amongst Statisticians:

Frequentist (Objectivist)
Probability means the frequency of occurrence


Bayesian (Subjectivist)
Probability means the degree of belief in occurrence

-club in the structural reliability community
6
( ) lim
E
n
n
P E
n

=
( ) Degree of belief in occurrence of P E E =
History of Probability
1545: Gerolamo Cardano
1
st
paper on probabilities



1654: Blaise Pascal
First usage of probabilities to solve problems
in communication with Pierre de Fermat

7
Pascal to Farmat, 1654
Monsieur,
Impatience has seized me as well as it has you, and although I am still
abed, I cannot refrain from telling you that I received your letter in
regard to the problem of the points yesterday evening from the hands
of M. Carcavi, and that I admire it more than I can tell you. I do not have
the leisure to write at length, but, in a word, you have found the two
divisions of the points and of the dice with perfect justice. I am
thoroughly satisfied as I can no longer doubt that I was wrong, seeing
the admirable accord in which I find myself with you.




.
.
.
8
Pascal to Fermat, 1654
Let us suppose that the first of them has two (points) and the other one. They
now play one throw of which the chances are such that if the first wins, he
will win the entire wager that is at stake, that is to say 64 pistoles. If the other
wins, they will be two to two and in consequence, if they wish to separate, it
follows that each will take back his wager that is to say 32 pistoles. Consider
then, Monsieur, that if the first wins, 64 will belong to him. If he loses, 32 will
belong to him. Then if they do not wish to play this point, and separate
without doing it, the first should say I am sure of 32 pistoles, for even a loss
gives them to me. As for the 32 others, perhaps I will have them and perhaps
you will have them, the risk is equal. Therefore let us divide the 32 pistoles in
half, and give me the 32 of which I am certain besides. He will then have 48
pistoles and the other will have 16.
Now let us suppose that the first has two points and the other none, and that
they are beginning to play for a point. The chances are such that if the first
wins, he will win all of the wager, 64 pistoles. If the other wins, behold they
have come back to the preceding case in which the first has two points and
the other one. But we have already shown that in this case 48 pistoles will
belong to the one who has two points. Therefore if they do not wish to play
this point, he should say, If I win, I shall gain all, that is 64. If I lose, 48 will
legitimately belong to me. Therefore give me the 48 that are certain to be
mine, even if I lose, and let us divide the other 16 in half because there is as
much chance that you will gain them as that I will. Thus he will have 48 and
8, which is 56 pistoles.
9
Pascal to Fermat, 1654
10
2 1
64 0
32 32
Next game:
32+16=48 16
2 0
64 0
48 16
Next game:
48+8=56 8
48 16
History of Probability (Continued)
1657: Christiaan Huygens
1
st
book on probabilities
De ratiociniis in ludo aleae
(On Reasoning in Games of Chance)



1713: Jacob Bernoulli (one of six bothers!)
The Art of Guessing

11
History of Probability (Continued)
1760: Thomas Bayes
Bayes Theorem
An Essay towards solving a Problem
in the Doctrine of Chances

1814: Pierre-Simon Laplace
Analytical Theories of Probability


1933: Andrey Kolmogorov
Formulated the axioms of the probability theory
Foundations of the Theory of Probability 12
Representation of Probabilities
Probability itself!



Chance, described by percentage


Odds
Outcomes for : Outcomes against


Outcomes for in total outcomes

13
1
0.333 or
3
p p = =
33.3%
1: 2 (1 to 2)
: n m
in
a
a b p
b
=
1 in 3

n
p
n m
=
+
Typical Probabilities
Death/h (10
-9
) Exposure h/y Prob./y (10
-6
)
Alpine climbing 40,000 50 2,000
Smoking 2,500 400 1,000
Boating 1,500 80 120
Air travel 1,200 20 24
Car travel 700 300 210
Train travel 80 200 16
Coal mining (UK) 210 1500 315
Construction work 70-200 2,200 154-440
Building fires 1-3 8,000 8-24
Structure failures 0.02 6,000 0.12

14
SET THEORY
Definitions
Venn diagram
Operations
MECE
De Morgan rules
15
Set Theory
Set Theory is the official language for defining events
Kolmogorov employed this theory to formulate the axioms of
the Probability Theory
Set Theory Defines events
Probability Theory Computes the probability of events

16
Definitions
Sample space: Set of all possible events S
Discrete (finite or infinite)


Continuous (always infinite)
Sample point x
Event: Collection of sample points E



Complement of an event
Certain event S
Null event

17
{ }
No damage, Light damage, Heavy damage, Collapse S =
{ }
1, 2, 3, 4, S = .
{ }
| 0 S x x = >
E
|
{ } { }
{ }
1 2
3
1, 2 | 4 6
Heavy damage, Collapse
E E x x
E
= = < <
=
7 sets
Venn Diagram
18
S
x
E
Discrete
x
.
E
S
Continuous
Operations
Union





Intersection
19

or
E
1
E
2
2 1
E E

and
E
1
E
2
2 1
E E
1 2 1 2
E E E E =
1 2
E E
Operation Rules
Commutative rule


Associative rule


Distributive rule


Intersection operations take precedence over union
operations, unless specified otherwise by parenthesis
20
1 2 2 1
E E E E =
1 2 2 1
E E E E =
( ) ( )
3 2 1 3 2 1
E E E E E E = ( ) ( )
3 2 1 3 2 1
E E E E E E =
( )
1 2 3 1 3 2 3
E E E E E E E = ( ) ( )( )
3 2 3 1 3 2 1
E E E E E E E =
Mutually Exclusive & Collectively
Exhaustive (MECE) Events

Mutually exclusive events


Collectively exhaustive events

For n sets:
21
E
1
E
2
E
3

E
n
1 2
E E | =
1 2
E E S =
De Morgans Rules
The complement of a union is equal to the intersection of the
complements



The complement of an intersection is equal to the union of
the complements



De Morgans rules convert series system problems to
parallel system problems, and vice versa

22
1 2 1 2 n n
E E E E E E =
1 2 1 2
E E E E =
n n
E E E E E E
2 1 2 1
=
1 2 1 2
E E E E =
De Morgans Rules
Proof by Venn diagram
23
E
1
E
2
2 1
E E
2 1
E E
PROBABILITY THEORY
Axioms
Frequentist vs. Bayesian
Probability rules
Statistical dependence
24
Axioms
25
Formulated in 1933 by
Andrey Nikolaevich Kolmogorov
( )
1 2
1 2 1 2
For an event
0
For the certain event
1
For mutually exclusive events and
( ) ( )
( )
( )

E
P E
S
P S
E E
P E E P E P E
=
= +
>
1
2
3
Probability Rules
Probability of the complement

Union rule

Inclusion-exclusion rule

Conditional probability rule

Multiplication rule

Bayes rule

Rule of total probability

26
) ( 1 ) ( E P E P =
) ( ) ( ) ( ) (
2 1 2 1 2 1
E E P E P E P E E P + =
1 2 3 1 2 3 1 2
1 3 2 3 1 2 3
( ) ( ) ( ) ( ) ( )
( ) ( ) ( )
P E E E P E P E P E P E E
P E E P E E P E E E
= + +
+
) (
) (
) | (
2
2 1
2 1
E P
E E P
E E P =
) ( ) | ( ) (
2 2 1 2 1
E P E E P E E P =
) (
) (
) | (
) | ( A P
E P
A E P
E A P =

=
=
n
i
i i
E P E A P A P
1
) ( ) | ( ) (
Complement Probability
All rules are derived from the axioms
Consider the complementary events E and E
They are mutually exclusive and collectively exhaustive
From second axiom:


From third axiom:


Hence
27
1 ) ( ) ( = = E E P S P
) ( ) ( ) ( E P E P E E P + =
) ( 1 ) ( E P E P =
Union Rule
Consider a Venn diagram with two events





Subtract the area that we count twice when P(E
1
E
2
) is nonzero


The inclusion-exclusion rule extends the Union Rule to more
than two events


28
E
1
E
2
) (
2 1
E E P
) ( ) ( ) ( ) (
2 1 2 1 2 1
E E P E P E P E E P + =
1 2 3 1 2 3
1 2 1 3 2 3
1 2 3
( ) ( ) ( ) ( )
( ) ( ) ( )
( )
P E E E P E P E P E
P E E P E E P E E
P E E E
= + +

+
Conditional Probability Rule

This rule may be justified by the frequency notion of
probability


E
2
can be considered as the new outcome space





The Multiplication rule is a direct consequence of the
conditional probability rule:


29
) | (
) (
) (
2 1
2
12
2
12
2
2 1
E E P
n
n
n
n
n
n
E P
E E P
= = =
E
1
E
2
) (
) (
) | (
2
2 1
2 1
E P
E E P
E E P =
1 2
1 2
2
( )
( | )
( )
P E E
P E E
P E
=
) ( ) | ( ) (
2 2 1 2 1
E P E E P E E P =
Bayes Rule
Conditional probability rule


Multiplication rule


Combine to obtain Bayes rule
30
1 2
1 2
2
( )
( | )
( )
E P E
P E E
P E
=
2 1 1 1 2
( ) ( | ) ( ) P E P E E P E E =
2 1
1 2 1
2
( | )
( | ) ( )
( )
P E E
P E E P E
P E
=
Posterior
Likelihood
Normalization
Prior
Bayes Theorem
The Bayes theorem is describes the application of the Bayes
rule
The theorem expresses how a subjective degree of belief
should rationally change to account for evidence
This is Bayesian inference, which is fundamental to Bayesian
statistics


Suppose the probability distribution of the earthquake
occurrence rate is described by f()
Upon observing another earthquake in the time interval x:

31
( | )
( | ) ( )
( )
x f
f f
f x
x

=
2 1
1 2 1
2
( | )
( | ) ( )
( )
P E E
P E E P E
P E
=
Rule of Total Probability
Extensively employed in the performance-based engineering
framework put forward by PEER

For a set of MECE events, E
i
:

32
( )
( )

=
=
=
=
=
=
=
n
i
i i
n
i
i
n
n
E P E A P
AE P
AE AE AE P
E E E A P
AS P A P
1
1
2 1
2 1
) ( ) | (
) (

) (
) ( ) (


E
1
E
2
E
3

E
n
A

The Monty Hall Problem
33
The Monty Hall Problem
34
The Monty Hall Problem
35
Should you switch?
The Monty Hall Problem
36
Each door has a 1 in 3 chance of hiding the grand prize.
Suppose we begin by choosing door #1.
In this case Monty may
open either door #2 or
#3
In both of these cases,
Monty is forced to reveal
the only other zonk.
The Monty Hall Problem
37
So what happens when you switch?
In this case you were
right the first time.
You lose!
In both of these cases,
you switch to the correct
door.
You win!
The Monty Hall Problem
38
Generalized Monty Hall Problem
39