On the exactness of the Wu-Woodroofe approximation (Q2389226)

Let \((X_i)\) denote a stationary process adapted to a filtration \(({\mathcal F}_i)\) with \(E(X_i)=0\) and \(E(X_i^2)<\infty\). For every positive integer \(n\), set \(S_n=\sum_{i=0}^{n-1}X_i\) and \(\sigma_n^2=\|S_n\|_2^2\). It is shown that there exists such a process with the following additional properties: \(\|E(S_n|{\mathcal F}_0)\|_2=O(\sqrt{n/\log n})\) and \(\sigma_n^2/n\) converges to a strictly positive, finite constant as \(n\to\infty\), but the central limit theorem does not hold. This example sheds some light on sufficient conditions for the central limit theorem of partial sums of stationary processes which are obtained through martingale approximations by \textit{W. B. Wu} and \textit{M. Woodroofe} [Ann. Probab 32, No. 2, 1674--1690 (2004; Zbl 1057.60022)] and \textit{M. Maxwell} and \textit{M. Woodroofe} [Ann. Probab. 28, No. 2, 713--724 (2000; Zbl 1044.60014)].