A supplement to precise asymptotics in the law of the iterated logarithm (Q1763402): Difference between revisions

Let \(\{X_i, i=1,2, \dots \}\) be a sequence of i.i.d random variables with \(EX_i=0, EX^2_i=1.\) Then the law of the iterated logarithm for their sums \(S_n\) holds and is related to the convergence or divergence of some series containing the probabilities \( P \{ | S_n| >\sqrt{(2+\varepsilon)n\log \log n}+a_n \} \). The problem is to find the normalizations as functions of \(\varepsilon\) such that non-trivial limits of the series yield as \(\varepsilon\rightarrow 0.\) The authors prove such statements under the conditions \(EX_i=0, EX^2_i=1\) only.