A uniform CLT for uniformly bounded families of martingale differences (Q1823536)

Let P be a probability on the product (\(\Omega\),\({\mathcal A})=(S^{{\mathbb{Z}}},{\mathcal B}^{{\mathbb{Z}}})\), (S,\({\mathcal B})\) being a measurable space. Assume that the left shift T on \(\Omega\) is measure- preserving and ergodic. Let \(M_ i=\sigma (X_ j:\) \(j\leq i)\) and \(H_ i\) be the subspace of \(L^ 2(\Omega)\) of the \(M_ i\)-measurable functions. Let \({\mathcal F}\subset H_ 0\cap H^{\perp}_{-1}.\) Then for each \(f\in {\mathcal F}\), \(\{f(T^ i),M_ i\}\) is a stationary martingale difference sequence. For \({\mathcal F}\) uniformly bounded, one of the main results of the article gives sufficient conditions (including a metric entropy condition) for the validity of a uniform CLT for the processes \(\{n^{-1/2}\sum_{0\leq i\leq n}f(T^ i):\quad f\in {\mathcal F}\}.\) This is a version for the dependent case of part of a theorem of \textit{E. Giné} and \textit{J. Zinn} [Ann. Probab. 12, 929-989 (1984; Zbl 0553.60037), Theorem 3.1]. The proof includes a generalization of a martingale version of Prokhorov's ``arc sinh'' exponential inequality due to \textit{W. B. Johnson}, \textit{G. Schechtman} and \textit{J. Zinn} [ibid. 13, 234-253 (1985; Zbl 0564.60020)]. One of the applications presented, a uniform CLT for certain Markov chains under a condition in terms of metric entropy with bracketing, is a generalization of a theorem of \textit{R. M. Dudley} [École d'été de probabilités de Saint-Flour XII-1982, Lect. Notes Math. 1097, 1-142 (1984; Zbl 0554.60029), Theorem 6.2.1].