The almost sure central limit theorem (Q1567436)

Let \(X_1,X_2,\dots\) be a sequence of i.i.d. random variables with \(EX_j= 0\), \(\text{Var }X_j= 1\) and denote \(S_n= \sum^n_{j=1} X_j\). Then \[ \lim_{N\to\infty} {1\over \log N} \sum^n_{j=1} {1\over j}1_{\{s_j/\sqrt j\in A\}}\to \int_A \varphi(x) dx,\quad\text{almost surely}, \] where \(\varphi(x)\) is the standard normal density. This is the statement of the almost sure (or pathwise) central limit theorem (ASCLT), proved independently by \textit{G. A. Brosamler} [Math. Proc. Camb. Philos. Soc. 104, No. 3, 561-574 (1988; Zbl 0668.60029)], \textit{P. Schatte} [Math. Nachr. 137, 249-256 (1988; Zbl 0661.60031)] and \textit{A. Fisher} [Adv. Math. 63, 213-246 (1987; Zbl 0627.60034)]. \textit{I. Berkes} and \textit{H. Dehling} [Ann. Probab. 21, No. 3, 1640-1670 (1993; Zbl 0785.60014)] established analogous results for sequences in the domain of attraction of a stable law. In the past decade, the ASCLT has been extended and strengthened into various directions. The present paper provides a survey of recent developments and of the techniques involved in the proofs. Topics included are extensions to independent, not identically distributed sequences, dependent random variables and speed of convergence in the ASCLT.