lecture6a
TRANSCRIPT
-
7/28/2019 lecture6a
1/4
Lecture 6 : Markov Chains
STAT 150 Spring 2006 Lecturer: Jim Pitman Scribe: Alex Michalka
Markov Chains
Discrete time
Discrete (finite or countable) state spaceS
Process {Xn}
Homogenous transition probabilities
matrix P = {P(i, j); i, j S}
P(i, j), the (i, j)th entry of the matrix P, represents the probability of moving to statej given that the chain is currently in state i.
Markov Property:
P(Xn+1 = in+1|Xn = in, . . . , X 0 = i0) = P(in, in+1)
This means that the states of Xn1 . . . X 0 dont matter. The transition probabilitiesonly depend on the current state of the process. So,
P(Xn+1 = in+1|Xn = in, . . . , X 0 = i0) = P(Xn+1 = in+1|Xn = in)
= P(in, in+1)
To calculate the probability of a path, multiply the desired transition probabilities:
P(X0 = i0, X1 = i1, X2 = i2, . . . , X n = in) = P(i0, i1) P(i1, i2) . . . P(in1, in)
Example: iid Sequence {Xn}, P(Xn = j) = p(j), and
jp(j) = 1.
P(i, j) = j.
Example: Random Walk. S = Z (integers), Xn = i0 + D1 + D2 + . . . + Dn, whereDi are iid, and P(Di = j) = p(j).
P(i, j) = p(j i).
Example: Same random walk, but stops at 0.
P(i, j) =
p(j i) if i = 0;1 if i = j = 0;0 if i = 0, j = 0.
6-1
-
7/28/2019 lecture6a
2/4
Lecture 6: Markov Chains 6-2
Example: Y0, Y1, Y2 . . . iid. P(Y = j) = p(j). Xn = max(Y0, . . . , Y n).
P(i, j) =
p(j) if i j;ki
p(k) if i = j;
0 if i > j.
Matrix Properties:P(i, ) = [P(i, j); j S], the ith row of the matrix P, is a row vector. Each row ofP is a probability distribution on the state-space S, representing the probabilities oftransitions out of state i. Each row should sum to 1. (
jS
P(i, j) = 1 i )
The column vectors [P(, j);j S] represent the probability of moving into state j.
Use the notation Pi = probability for a Markov Chain that started in state i (X0 = i).In our transition matrix notation, Pi(X1 = j) = P(i, j).
n-step transition probabilities:
Want to find Pi(Xn = j) for n = 1, 2,...This is the probability that the Markov Chain is in state j after n steps, given thatit started in state i. First, let n = 2.
Pi(X2 = k) = jS
Pi(Xi = j, X2 = k)
=jS
P(i, j)P(j, k).
This is simply matrix multiplication.
So, Pi(X2 = k) = P2(i, k)
We can generalize this fact for n-step probabilities, to get:
Pi(Xn = k) = Pn(i, k)
Where Pn(i, k) is the (i, k)th entry of Pn, the transition matrix multiplied by itself ntimes. This is a handy formula, but as n gets large, Pn gets increasingly difficult andtime-consuming to compute. This motivates theory for large values of n.
Suppose we have X0 , where is a probability distribution on S, so P(X0 = i) =
-
7/28/2019 lecture6a
3/4
Lecture 6: Markov Chains 6-3
(i) for i S. We can find nth step probabilities by conditioning on X0.
P(Xn = j) =iS
P(X0 = i) P(Xn = j|X0 = i)
=iS
(i)Pn(i, j)
= (Pn)j
= the jth entry of Pn.
Here, we are regarding the probability distribution on S as a vector indexed byi S.
So, we have X1 P,X2 P2, . . . , X n P
n.
Note: To compute the distribution of Xn for a particular , it is not necessary tofind Pn(i, j) for all i,j,n. In fact, there are very few examples where Pn(i, j) canbe computed explicitly. Often, P has certain special initial distributions so thatcomputing Pn is fairly simple.
Example: Death and Immigration Process
Xn = the number of individuals in the population at time (or generation) n.
S = {0, 1, 2, . . .}
Idea: between times n and n + 1, each of the Xn individuals dies with probabilityp, and the survivors contribute to generation Xn+1. Also, immigrants arrive eachgeneration, following a Poisson() distribution.
Theory: What is the transition matrix, P, for this process? To find it, we conditionon the number of survivors to obtain:
P(i, j) =ijk=0
i
j
(1 p)k(p)ik e
jk
(jk)!
Here, i j = min{i, j}. This formula cannot be simplified in any useful way, butwe can analyze the behavior for large n using knowledge of the Poisson/Binomialrelationship. We start by considering a special initial distribution for X0.
Let X0 Poisson(0). Then the survivors of X0 have a Poisson(0q) distribution,where q = 1p. Since the number of immigrants each generation follows a Poisson()distribution, independent of the number of people currently in the population, we haveX1 Poisson(0q+ ). We can repeat this logic to find:
X2 Poisson((0q+ )q+ ) = Poisson(0q2 + q+ ), and
-
7/28/2019 lecture6a
4/4
Lecture 6: Markov Chains 6-4
X3 Poisson((0q2 + q+ )q+ ) = Poisson(0q
3 + q2 + q+ ).
In general,
Xn Poisson(0qn +
n1k=0
qk).
In this formula, 0qn represents the survivors from the initial population, qn1 rep-
resents the survivors from the first immigration, and so on, until q represents thesurvivors from the previous immigration, and represents the immigrants in the cur-rent generation.
Now well look at what happens as n gets large.
As we let n : 0qn 0, and 0qn + n1k=0
qk 1q
= p
.
So, no matter what 0, if X0 has Poisson distribution with mean 0,
limn P(Xn = k) =ek
k!, where =
1q=
p.
It is easy enough to show that this is true no matter what the distribution of X0.The particular choice of initial distribution for X0 that is Poisson with mean 0 =
gives an invariant (also called stationary, or equilibrium, or steady-state) distributionof X0. With 0 = , we find that each Xn will follow a Poisson() distribution. This
initial distribution (i) = ei
i!is special because it has the property that
iS
(i)P(i, j) = (j) for all j S.
Or, in matrix form, P = . It can be shown that this is the unique stationaryprobability distribution for this Markov Chain.