Refinements of Kolmogorov's law of the iterated logarithm (Q1199005): Difference between revisions

The author extends Kolmogorov's law of the iterated logarithm. The main result gives the following Theorem: Let \(X_ 1,X_ 2\dots\) be a sequence of independent r.v.'s with zero means and finite variances. For \(n\geq 1\), define \(S_ n=X_ 1+\dots+X_ n\), \(s_ n^ 2=ES_ n^ 2\) and \(t_ n^ 2=2\log\log s_ n^ 2\). If \(s_ n\to\infty\), \(t_ n X_ n/s_ n\to 0\) a.s. as \(n\to\infty\) and \(t_ n| X_ n|/s_ n\leq Y\) a.s. for every \(n\geq 1\) and some r.v. \(Y\) with \(EY^ 2<\infty\), then (*) \(\limsup_{n\to\infty} S_ n/s_ nt_ n=1\) a.s. In particular (*) holds if \(s_ n\to\infty\) and \(E(\sup_{n\geq 1}X_ n^ 2)<\infty\). The author also proves that the central limit theorem implies (*) whenever \(EX_ n^ 2=o(s_ n^ 2)\) and \(t_ nX_ n/s_ n\to 0\) a.s. as \(n\to\infty\), but the converse is not always true.