English

On martingale approximations

Probability 2008-11-14 v2

Abstract

Consider additive functionals of a Markov chain WkW_k, with stationary (marginal) distribution and transition function denoted by π\pi and QQ, say Sn=g(W1)+...+g(Wn)S_n=g(W_1)+...+g(W_n), where gg is square integrable and has mean 0 with respect to π\pi. If SnS_n has the form Sn=Mn+RnS_n=M_n+R_n, where MnM_n is a square integrable martingale with stationary increments and E(Rn2)=o(n)E(R_n^2)=o(n), then gg is said to admit a martingale approximation. Necessary and sufficient conditions for such an approximation are developed. Two obvious necessary conditions are E[E(SnW1)2]=o(n)E[E(S_n|W_1)^2]=o(n) and limnE(Sn2)/n<\lim_{n\to \infty}E(S_n^2)/n<\infty. Assuming the first of these, let g+2=lim supnE(Sn2)/n\Vert g\Vert^2_+=\limsup_{n\to \infty}E(S_n^2)/n; then +\Vert\cdot\Vert_+ defines a pseudo norm on the subspace of L2(π)L^2(\pi) where it is finite. In one main result, a simple necessary and sufficient condition for a martingale approximation is developed in terms of +\Vert\cdot\Vert_+. Let QQ^* denote the adjoint operator to QQ, regarded as a linear operator from L2(π)L^2(\pi) into itself, and consider co-isometries (QQ=IQQ^*=I), an important special case that includes shift processes. In another main result a convenient orthonormal basis for L02(π)L_0^2(\pi) is identified along with a simple necessary and sufficient condition for the existence of a martingale approximation in terms of the coefficients of the expansion of gg with respect to this basis.

Keywords

Cite

@article{arxiv.0708.4183,
  title  = {On martingale approximations},
  author = {Ou Zhao and Michael Woodroofe},
  journal= {arXiv preprint arXiv:0708.4183},
  year   = {2008}
}

Comments

Published in at http://dx.doi.org/10.1214/07-AAP505 the Annals of Applied Probability (http://www.imstat.org/aap/) by the Institute of Mathematical Statistics (http://www.imstat.org)

R2 v1 2026-06-21T09:12:24.144Z