23 Long-Term Behavior of MC

1 Periodic & Aperiodic

We've introduced common concepts about Markov chain.
Consider a 2− state MC. The transition matrix is P=[1−aab1−b], where a,b∈(0,1]. Then Pn=1a+b[baba]+(1−a−b)na+b[a−a−bb].

Question: under what condition is limn→∞[Pn]ij=limn→∞P(Xn=j|X0=i)=πj,∀i,j∈S?

Period

The period d(i) of a state i∈S is defined as d(i)=gcd{n∈N|[Pn]ii>0}.

For example above, d(1)=d(2)=2.

Periodic / Aperiodic

A state i∈S is periodic / aperiodic, if d(i)>1 / d(i)=1.

Claim

If i↔j, then d(i)=d(j).

Suppose a class C of period d. Then there exists a partition C=B0∪⋯∪Bd−1.

Theorem

Let C be a class in a finite-state Markov Chain with period d. Then C can be partitioned into d subsets C=B0∪⋯∪Bd−1 s.t. all transitions from BKmodd go to BK+1modd.

Theorem

If a MC with finite state space S is irreducible and aperiodic, then

  1. (Regular or ergodic): ∃k∈N∗, s.t. [Pk]ij>0,∀i,j∈S.
  2. (Limiting distribution): limn→∞[Pn]ij=πj,∀i,j∈S where π=(πi)i∈S is the unique stationary distribution. (or in matrix notation limn→∞Pn=1π=[π⋮π].

2 Convergence Theorem

Another interpretation of the stationary distribution for an irreducible finite MC:

Lemma

Consider an irreducible finite MC with transition matrix P and the unique stationary distribution π. Then limn→∞I+P+⋯+Pnn+1=[π⋮π].

Hj(n)=1n+1∑k=0n1{Xk=j} is the fraction of time spent in state j during steps 0,1,⋯,n.

E[Hj(n)|X=i]=1n+1∑k=0nP(Xk=j|X0=i)⏟[Pk]ij.

By the lemma, this →πj as n→∞.

WLLN for Irreducible Finite MC

∀ε>0, limn→∞P(|Hj(n)−πj|>ε)=0, independent of the starting state X0=i.

A stronger result:

Theorem (Ergodic Theorem)

Let {Xn|n∈N0} be an irreducible, positive recurrent MC on state space S and stationary distribution π. Suppose g:S→R be a function s.t. ∑i∈Sg(i)πi<∞. Then for any initial distribution for X0, 1n∑k=1ng(Xk)→a.s.∑i∈Sg(i)πi,n→∞.

3 First Passage Time

First Passage time

For i∈S, Ti=min{n∈N|Xn=i}.

Fundamental Matrix of Irreducible MC

Z=(I−P+1π)−1.

  1. Z is well defined.
  2. If the MC is aperiodic, then Z=I+∑n=1∞(Pn−1π).
Mean First Pasage Time

For an irreducible finite Markov Chain with the fundamental matrix Z0=(zij) and the unique stationary distribution π, E[Tj|X0=i]=Zjj−Zijπj.

Suppose a finite MC is not irreducible (i.e., has more than one SCC)
P restricted to each terminal SCC is a valid transition matrix. Then there exists a stationary distribution for each terminal SCC.

4 Reversible MC

Pasted image 20241218225444.png|400

Theorem

Let {Xn|0≤n≤N} be an irreducible positive recurrent MC with transition matrix P=(pij) and unique stationary distribution π. Further, suppose Xn∼π,∀0≤N. Then the reversed process {Yn=XN−n,0≤n≤N} is a MC with transition matrix Q=(qij), where qij=πjpjiπi.

Reversibility

The chain {Xn} is said to be reversible if qij=pij, ∀i,j. That is, detailed balance πjpji=πipij,∀i,j.

Theorem

Let {Xn} be an irreducible Markov Chain with transition matrix P=(pij) and suppose ∃ a distribution ν=(νi) s.t. νipij=νjpji, ∀i,j. Then ν is a stationary distribution of the chain and {Xn} is reversible in equilibrium.