On martingale approximations (Q957521)

Let \((W_n)\) be a Markov chain (having values in a measurable space \({\mathcal W}\)) with stationary (marginal) distribution \(\pi\) and let \(g\in L^2(\pi)\) have mean \(0\) w.r.t. \(\pi\). If \(S_n:= g(W_1)+\cdots+ g(W_n)\) is of the form \(S_n= M_n+ R_n\), where \((M_n)\) is a square integrable martingale with stationary increments and \(E[R^2_n]= o(n)\), then \(g\) is said to admit a martingale approximation (MA) (see, e.g., \textit{M. I. Gordin} [Sov. Math., Dokl. 10, 1174--1176 (1969); translation from Dokl. Akad. Nauk SSSR 188, 739--741 (1969; Zbl 0212.50005)], \textit{W. B. Wu} and \textit{M. Woodroofe} [Ann. Probab. 32, No. 2, 1674--1690 (2004; Zbl 1057.60022)] and \textit{O. Zhao} and \textit{M. Woodroofe} [Ann. Probab. 36, No. 1, 127--142 (2008; Zbl 1130.60039)]). The authors obtain conditions which are necessary and sufficient for \(g\) to admit a MA. Applications include Bernoulli shifts, Lebesgue shifts, superlinear processes, the fractional Poisson equation, and the conditional central limit theorem for superlinear processes.