Rate of convergence to a stable law in the space \(R^ k\) (Q1116166): Difference between revisions

Let \(X_ i\), \(i\geq 1\), be i.i.d. \(R^ k\)-valued random variables and let \(Y_{\alpha}\) be stable random vectors in \(R^ k\) with exponent \(\alpha\), \(0<\alpha \leq 2\). Denote \[ S_ n=n^{- 1/\alpha}\sum^{n}_{i=1}X_ i,\quad \Delta_ n(M)=\sup_{A\in M}| P\{S_ n\in A\}-P\{Y_{\alpha}\in A\}|, \] where \(M\) is some class of Borel subsets in \(R^ k\) (the main two examples considered in the paper are the class of all convex sets and the class of all balls in \(R^ k).\) A precise estimate of \(\Delta_ n(M)\) is given, which refines and generalizes previous results obtained by the reviewer [Litov. Mat. Sb. 15, No. 1, 207--228 (1975; Zbl 0324.60022)] and \textit{I. Yu. Mikhailova} [Mosc. Univ. Comput. Math. Cybern. 1983, No. 3, 78--80 (1983); translation from Vestn. Mosk. Univ., Ser. XV 1983, No. 3, 60--62 (1983; Zbl 0518.60027)]. In the proof, ideas from the paper by \textit{V. Bentkus} [Lith. Math. J. 26, 110--114 (1986); translation from Litov. Mat. Sb. 26, No. 2, 205--210 (1986; Zbl 0612.60022)] are used.