Upload
hoangtruc
View
229
Download
2
Embed Size (px)
Citation preview
11
Stochastic Structural Dynamics
Lecture-2
Dr C S ManoharDepartment of Civil Engineering
Professor of Structural EngineeringIndian Institute of ScienceBangalore 560 012 India
Scalar random variables-1
Recall
• Uncertainty modeling using theories of probability and random processes
• Definitions of probability– Classical definition P(A)=m/n
– Relative frequency definition
– Axiomatic definition2
nmAP
n lim)(
3
ExperimentTrialOutcome
Randomexperiment
SampleSpace
Event e
Event space
0 1
R
E
B
Undefinednotions
0
1
if
iP A
P
P A B P A P BA B
Axioms
P e
•Sample point: element of sample space•Events are subsets of sample space on which we assign probability•Axiomatic definition does not prescribe how to assign probability
Axiomatic definition of probability
Recall (continued)
4
Recall (continued)
•Conditional Probability
•Stochastic independence
•Total probability theorem•Bayes theorem
.0;BAP
occurred has B given thatA event ofy Probabilit|Definition
BPBP
BAP
BPAPBAPBA tindependen areB and A:Notation
5
ExperimentTrialOutcome
Randomexperiment
SampleSpace
Event e
Event space
0 1
R
R X
0)(: (2)
event,an is )(: ,every for (1)such that line real into
space sample fromfunction a is variableRandom
XPxXRx
Random variable E
B
Undefinednotions
0
1
if
iP A
P
P A B P A P BA B
Axioms
P e
66
iX
xωX
i 10 Define654321
tossing.die of experiment heConsider t:Example
of Meaning
X is a subset of and hence
an element of and hence an event on which we assign probabilities.
x
B
Observation
54325020
1005
432140
XXXX
. writeWe xXxX
defined. are etc., ,covariance deviation, standardmean, assuch qunatitieson which basis theForms (c)
ies.uncertaint oftion quantifica Enables (b)y valued.numericallnot is which with deal tous Enables (a)
variable random a of notion the gintroducinfor Need
8
Definition
, ;
Notation, (the dependence on is not explicitly displayed)
X
X
X X
P x
P x P X x x
P x P x
Probability Distribution Function ︵ PDF ︶
9
2 1 2 1
2 1
2 1 1 2
1 1 2
2 1 1 2
2 1
( ) 0 ( ) 1( ) ( ) ( ) 1( ) ( ) ( ) 0( ) ( ) ( )PDF is monotone nondecreasing.Let .
X
X
X
X X
a P xb P P X Pc P P X Pd x x P x P x
x xX x X x x X x
X x x X x
P X x P X x P x X x
P X x P X x
Properties
0 0
( ) lim
PDF is right continuous.
X Xe P x P x
1010
x Xx PXx0 0-100 0600 15 010 1 1/620 1,2 2/622 1,2 2/635 1,2,3 3/653 1,2,3,4,5 5/6
: Die tossing: 1 2 3 4 5 6 ; 10iX i Example
1111
.continuous becomes T of PDF the,N As.1,,2,1 , at itiesdiscontinu have wouldT of PDF the,NFor
.1,,2,1; if T takeWe
.,,2,1,0; with
segments timediscrete into interval thedivide usLet
.arrival of time thedenoteswhich variablerandom thebe Let
likely.equally be interval in theinstant any timeLet
time.arrival TLet platform. on the
train a of arrival of timeheConsider t
1
1211
21
21
21
Nnt
Nntttt
NnttNntt,tt
N,tt
T,ttΩ
,tt
n
nnn
nnn
Example
12
lycontinuous and jumpsboth with proceeds PDF :Mixedjumps.any without proceeds PDF :Continuous
jumps.gh only throu proceeds PDF :Discretevariables Random
PDF of aMixed RV
xPxP XX
00
lim
1313
Probability density function (pdf)
x
XX
XX
duupxP
dxxdPxp
)()(
)()(
Definition
)(
)(
)(1)()(
dxxXxPdxxp
duupbXaP
duupPXPP
X
b
aX
XX
Properties
xpX
14
15
(a) ; 0,1,2, is also knwon as probability mass function (PMF).(b) PDF is also known as cumulative probability distribution function (CDF).
kP X k p k Notes :
Heaveside’s step function
16
ax
axaxaxU
21
1 0
x
axU
0,0
1
Example 1: Box function
17
bxUaxUxf )(
x
xf
0,0 a b
1
Dirac’s delta function
18
xxUdxd
aaxUxUaF
axUxUFxfaxUxUFxf
aFa
aFa
aFa
)(
lim
limlim)(
01
0
01
01
0
0
0
00
Dirac’s delta function
19
axaxUdxd
afdxaxxf
dxax
axax
1
for 0
Example-2: Stair case function
20
bxaxaxxdxdf
bxUaxUaxUxUxf
xf
x 0,0
a b
Commonly encountered random variables
• Models for rare events• Models for sums• Models for products• Models for extremes
– Highest– Lowest
• Models for waiting times
21
Limit theorems
2222
.1)1()1()()(
:Check).1()1()()1()1()(
110
Let FS
failure. and success :outcomes only two has experiment random The
--X
X
X
dxxpxpdxxp
xpxpxpxUpxpUxP
-p)P(XP(F)p )P(XP(S)
variable random Bernoulli
)(xpX
Remarks:• p is the parameter of the Bernoulli random variable.•Discrete random variable•Finite sample space•Basic building block
2323
Repeated Bernoulli trials: .The random experiment here consists of repeated Bernoulli trials.Assumptions(a) The random experiment consists of independent trials.(b) Each t
N
N
Binomial random variable
rial results in only two outcomes (success/failure)(c) P(success) remains constant during all trials.Define =number of successes in trials; 0,1,2,3, , .X N X N
(1 ) ; 0,1,2, ,N k N kkP X k C p p k N
0
Binomial theorem:
!( )! !
nn n r n r
rr
Nk
p q C p q
NCN k k
Notes
2424
Consider a sequence of trials resulting in successes.Occurrence of successes implies the occurrence of ( - ) failures.
Probability of occurrence of one such sequence= (1 ) .
Number of such pos
k N k
N kk N k
p p
0
0
sible sequences= .These sequences are mutually exclusive.
(1 )
(1 )
(1 ) 1
(By virtue of binomial theorem. )Hence the name binomial random variable
Nk
N k N kk
mN k N k
kk
NN k N k
kk
C
P X k C p p
P X m C p p
P X N C p p
.
25
A binomial random variable is denoted by ( , ). and are the paramters of this random variable.
B N pN p
Remarks
finite is space Sample variablerandom Discrete
26
xpX
xPX
x
x
10,0.5B
2727
Random experiments: as in binomial random variable.=number of trials for the first success; 1,2, , .first success in N-th trial (success on the -th trial failures on the
N NP P N
Geometric random variable
1
1
1 1
-1 -1
1 1
2 3
first ( -1) trials)
(1 ) ; 1,2, , .
(1 ) (this must be =1).
: let p=0.6
(1- ) 0.4 0.6
0.6 1+ 0.4 + 0.4 0.4
The expression inside the bracket is a g
n
n
n n
n n
n n
N
P N n p p n
P N n p p
Ex
p p
=
eometric progression.Hence the name geometric random variable.
systems gengineerin of timeslife modelingin Useful
space sample infiniteCountably variablerandom Discrete
28
xPX
xpX
Geometric random variable with p=0.4
x
29
( 1)( 1)
Define Number of trials to the -th success.It can be shown that
(1 ) ; , 1, 2, ,
k
w w k kk k
W k
P W w C p p w k k k
Pascal or negative binomial distribution
3030
(a) We are looking for occurrence of an isolated phenomenonin a time/space continuum. (b) We cannot put an upper bound on the number of occurrences.(c) A
Models for rare events : Poisson random variable
ctual number of occurrences is relatively small.
: goals in football match (time continuum), defect in a yarn(1- d space continuum), typos in a manuscript (2 - d continuum), defect in a solid (3 - d
Examples
0 0
continuum).
exp ; 0,1,2,!
exp exp exp exp 1.! !
k
k k
k k
aP X k a kk
a aP X a a a ak k
Stress at a point exceeding elastic limit during the life time of a structure.
Check
31
x
x
xPX
xpX
Poisson random variable with a=5
•Discrete RV•Countably infinite
sample space•Useful in wide variety of contexts