On Feller's criterion for the law of the iterated logarithm (Q1326621)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | On Feller's criterion for the law of the iterated logarithm |
scientific article |
Statements
On Feller's criterion for the law of the iterated logarithm (English)
0 references
24 October 1994
0 references
Let \(\{S_ n,n \geq 1\}\) be the partial sums of a sequence \(\{X_ n,n \geq 1\}\) of independent random variables (r.v.) with \(EX_ n = 0\), \(EX^ 2_ n < \infty\), \(n \geq 1\). Define \(s^ 2_ n = ES^ 2_ n\), \(a^ 2_ n = 2s^ 2_ n \log \log s^ 2_ n\) and assume \(s_ n \to \infty\). Using nonuniform bounds in the central limit theorem and a result of \textit{W. Feller} [Ann. Math. (2) 91, 402--418 (1970; Zbl 0252.60014)], the authors prove that the law of the iterated logarithm (LIL) holds (i.e., \(\lim \sup_{n \to \infty} S_ n/a_ n = 1\) a.s.) if \(\lim \sup_{n \to \infty} s_{n+1}/s_ n < \infty\) and \(\sum^ \infty_{n=1} E(X^ 2_ n g(X_ n))/(a^ 2_ n g(a_ n)) < \infty\) for some nonnegative, even function \(g\) such that both \(g\) and \(x/g(x)\) are nondecreasing on \((0, \infty)\). This theorem generalizes a result of \textit{R. Wittman} [Z. Wahrscheinlichkeitstheorie Verw. Geb. 68, 521--543 (1985; Zbl 0547.60036)], and is compared to LIL due to \textit{H. Teicher} [Ann. Probab. 2, 714--728 (1974; Zbl 0286.60013)] and the reviewer [Z. Wahrscheinlichkeitstheorie Verw. Geb. 65, 135-143 (1983; Zbl 0506.60023)]. A similar approach is used to establish conditions under which \(S_ n/b_ n \to 0\) a.s. for certain sequences \(\{b_ n\}\) with \(s_ n = o(b_ n)\). Finally, a theorem of \textit{V. A. Egorov} [Vestn. Leningr. Univ. 1971, No. 7, Mat. Mekh. Astron. No. 2, 22--28 (1971; Zbl 0244.60026)] is extended by showing that the LIL holds if \(\lim \inf_{n \to \infty} s^ 2_ n/n>0\), \(\lim \sup_{n \to \infty} s^ 2_ n/n < \infty\) and \(\sum^ n_{k=1} P[| X_ k |>x] \leq nP[X>x]\) for all large \(n\) and \(x\) and some r.v. \(X\) with \(E(X^ 2/ \log \log X) < \infty\).
0 references
sums of independent random variables
0 references
central limit theorem
0 references
law of the iterated logarithm
0 references