A Zero-One Law for Virtual Markov Chains

TL;DR

This paper characterizes when the tail sigma-algebra of a virtual Markov chain is trivial, using a decomposition into independent staircase Markov chains and connecting to convex analysis.

Contribution

It provides an exact criterion for the triviality of the tail sigma-algebra in VMCs and introduces a decomposition theorem involving staircase Markov chains.

Findings

Characterization of the triviality of the tail sigma-algebra in VMCs.

Decomposition of VMCs into independent staircase Markov chains.

Connection between staircase Markov chains and convex analysis.

Abstract

A virtual Markov chain (VMC) is a sequence $\{X_N\}_{N=0}^{\infty}$ of Markov chains (MCs) coupled together on the same probability space such that $X_N$ has state space $\{0,1,\ldots, N\}$ and such that removing all instances of $N~+~1$ from the sample path of $X_{N+1}$ results in the sample path of $X_N$ almost surely. In this paper, we prove an exact characterization of the triviality of the $\sigma$-algebra $\bigcap_{N=0}^{\infty}\sigma(X_N,X_{N+1},\ldots)$. The main tool for doing this is a decomposition theorem that the $\sigma$-algebra generated by a VMC is equal to the $\sigma$-algebra generated by a certain countably infinite collection of independent constituent MCs. These constituents are so-called staircase MCs (SMCs), which are defined to be inhomoheneous Markov chains on the non-negative integers which transition only by holding or by jumping to a value equal to the current index. We also develop some general aspects of the theory of SMCs, including a connection with some classical but very much under-appreciated aspects of convex analysis.