Strong invariance principle for dependent multi-indexed random variables (Q2455206): Difference between revisions

Let \(X= \{X_j\), \(j\in\mathbb{Z}^d,d\geq 1\}\) be a wide-sense stationary, centered random field with uniformly bounded \(p\)th moments, \(p> 2\), and \(S_N=\sum_{j\in (0,N]}X_j\), where \(N= (N_1,\dots, N_d)\in\mathbb{Z}^d\). It is assumed that \(X\) satisfies certain weak dependence condition introduced by \textit{A. Bulinski} and \textit{C. Suquet} [Stat. Probab. Lett. 54, No. 2, 215--226 (2001; Zbl 0989.60052)], called \((BL,\theta)\)-dependence and covering positively and negatively associated random fields. The following strong invariance principle is established. \(X\) can be redefined on some probability space in such a way that \(S_N- \sigma W_N= O((N_1N_2\cdots N_d)^{1/2-\varepsilon})\) almost surely as \(N\) tends tp infinity in a given sense, where \(W_t\) is a \(d\)-parametric Brownian motion. The sketched proof consists of three main steps: normal approximation for weakly dependent random fields, quantile transform technique of Csörgő and Révész developed by Balan and a new maximal inequality.