On almost-everywhere variants of limit theorems (Q1385871): Difference between revisions

Let \((X_n)\) be a sequence of i.i.d. random variables belonging to the attraction domain of the stable law \(G_\alpha\), \(0< \alpha\leq 2\), and let \(S_n= b_n^{-1} (X_1+ \cdots +X_n) -a_n\) converge in distribution to \(G_\alpha\). Define \(Q_n(A) =(\log n)^{-1} \sum^n_{k=1} \mathbf{1}_A (S_k)/k\) the random probability distribution corresponding to \((S_n)\). It is proved that \(P(Q_n \to G_\alpha) =1\). Generalizations to nonidentically distributed and multivariate random variables as well as to the functional limit theorem are also given.