Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Markov processes - III
• review of steady-state behavior
• probability of blocked phone calls
• calculating absorption probabilities
• calculating expected time to absorption
1
'V i
)(0=' • - - ',;. ;'-,
1.,,__ 0 _.. ; " \..
review of steady state behavior
• Markov chain with a single class of recurrent states. aperiodic; and some transient states; then.
lim " i)'(n) = lim P ( X " = j I Xo = i) = " J' n-+oo 11-+00
(P(l(...:.j .; ~ r1J ... ,. I ... /"1fj" 4-. 1P~O-)1....
• can be found as the unique solution to the balance equations
j = 1, .. . ,
• together with L " j = 1 J
2
• •
on the use of steady state probabilities, example III"O.S~1I'"\.J'O.'l. ~ 1
, o.qqB!).,8'"
iTi -to TTl. =d Od~ I M
assume process
____
starts in state 1
P (Xl = 1 ilnd XIOO = 1 1 Xo = 1) =
IrCx,=.,/'ICo,.l).'p(X,oo='/ )(,:~~) t. f" J' r" Cq'l) 'R:i f" ~ 1Tj OS~~Co
P (XIOO = 1 and XIOI = 2 1 Xo = 1) = .N ) jP(X.o--I/ '1.0.')'" W(X.o. ~2...J XOOO= ~ ~I
f,,(14)O) )I f't. W1f,"'l'l... ~ 2....,.00$ P (XlOO = 1 and X f OO = I J XO = 1) = :}~./ )
,i>OC.,._. Xo•• ). 0'( X?Q)~II X,eo ..'... ~, t. = (,,{.oo) .... r.,(IOO) ~-,y;"',,. :'TI;'-: (~)
, O.S
.' F'2./r ________3>
'" 5""" ",:,.5 ..e ~ct.(, ',,1 ....0 ~..., 10 5 cl."> J
t) O/d..t" .:> ~ .... ..,..~..
:£> PI' -Wc.o/F ...
(c:. \ O<"C < I
_<::0.3 _ C ::: o.~'}
3
••
••
design of _a~hone system (8 r '"",)
• calls originate as a Poisson process. rate oX • each call duration is exponential (parameter f.J.) • need to decide on how many lines, B~
t , ,• tIme
• for time slots of small duration at! ".l I I
l..,.S P(a new call arrives) :::::: Ad if you have i active calls. then P{a departure) ::::::: i J.L§
v;lI. j~ _ t,.~
( ,,,,,I<) --
" .n
" ." "
••
_ _ .1111 -< 7 ~1111
•• • • • ~ II , . M II ~
4
design of a phone system, a discrete time approximation ,
time
• approximation: discrete time slots of (small) duration 8
!P(i. •...,"'")~U ; TI(1."'II"'~ll":b.sy)~<I"S;
o
1 ). ;;::.]0 c.cJ/, I..:... ~ #=->3
J. ""',....It.t 'fo calls• balance e uations
A7fi_ l = iJ1-7ri
11" "
'13", '0 ? "! Ai
- 11"01 -i"1H
~---..,
1
110 , 1r. -) -,rEo - --~- -
• P(arriving customer finds busy system) is@
5
-
calculating absorption probabilities
• absorbing state: recurrent state k with Pkk = b..:
• what is the probability @ that the chain eventually settles inA'given it started in i ?
S
1
i = 4, ai = I i =5 , ai = O
3 OJ I
otherwise, a i = ! ., a.: ,BAd" elL'" tons06
1 r-__..:-.<lI::,..:l ::;.1>/t.8-08 ..., • unique solution from i = L Pij aj
.'Ib. .... .l . : I ().s =- -:L. / • -d '-- 6
expected time to absorption
• find expected number of transitions I-'i until reaching 4, given that the initial state is i
" "
" "
J.Li=o for i = 4o.tfor all others, j..1.'i = ?
, Ih:.IIO/B'0.4
_ /-<'1.:: qt;~ :. ~
~1= 1Iy&
• unique solution from i = 1 + L PijI.Lj .
J
7
mean first passage and recurrence times
• chain with one recurrent class; fix a recurrent state s
"
8
••
gambler's example
• a gambler starts with i dollars; each time. she bets $1 in a fair game, until she either has 0 or n dollars. • what is the probability ai that she ends up with having n '&lIars? , $.v ••r..,. t
• = n, Ui =
~~ •••
>'-,~~ a...:
, I• expected wealth at the end!
'-' • how long does the gambler expect to stay in the game! a.;. =- % .""A-llcO
/.Li = expected number of plays. starting from i
for i = 0 , n: J-li = 0 \ <.: c .. ~ :. i + I ,/".1+1. -+ - in general
J-Li = 1 + L Pijl.Lj J
• in case of unfavorable odds? f ;: O·s 1._1....• r:: ,-t ~...: =
f I - r"" 9
MIT OpenCourseWarehttps://ocw.mit.edu
Resource: Introduction to ProbabilityJohn Tsitsiklis and Patrick Jaillet
The following may not correspond to a p articular course on MIT OpenCourseWare, but has been provided by the author as an individual learning resource.
For information about citing these materials or our Terms of Use, visit: https://ocw.mit.edu/terms.