Martingale approximation and optimality of some conditions for the central limit theorem

TL;DR

This paper investigates the conditions under which the central limit theorem holds for stationary ergodic Markov chains, showing that certain weakened or non-normal conditions can lead to failure of the CLT.

Contribution

It extends existing results by demonstrating that the CLT may fail when the kernel is non-normal and certain convergence conditions are weakened or not met.

Findings

CLT can fail if the kernel is non-normal and conditions are weakened.

Normality of the kernel is crucial for the CLT under these conditions.

Weaker convergence conditions do not guarantee the CLT.

Abstract

Let $(X_i)$ be a stationary and ergodic Markov chain with kernel $Q$, $f$ an $L^2$ function on its state space. If $Q$ is a normal operator and $f = (I-Q)^{1/2}g$ (which is equivalent to the convergence of $\sum_{n=1}^\infty \frac{\sum_{k=0}^{n-1}Q^kf}{n^{3/2}}$ in $L^2$), we have the central limit theorem (cf\. \cite{D-L 1}, \cite{G-L 2}). Without assuming normality of $Q$, the CLT is implied by the convergence of $\sum_{n=1}^\infty \frac{\|\sum_{k=0}^{n-1}Q^kf\|_2}{n^{3/2}}$, in particular by $\|\sum_{k=0}^{n-1}Q^kf\|_2 = o(\sqrt n/\log^q n)$, $q>1$ by \cite{M-Wu} and \cite{Wu-Wo} respectively. We shall show that if $Q$ is not normal and $f\in (I-Q)^{1/2} L^2$, or if the conditions of Maxwell and Woodroofe or of Wu and Woodroofe are weakened to $\sum_{n=1}^\infty c_n\frac{\|\sum_{k=0}^{n-1}Q^kf\|_2}{n^{3/2}}<\infty$ for some sequence $c_n\searrow 0$, or by $\|\sum_{k=0}^{n-1}Q^kf\|_2 = O(\sqrt n/\log n)$, the CLT need not hold.