Law of iterated logarithm for random subsequences (Q1181104): Difference between revisions

Let \(\{X_ k\}\) be an independent non-negative random sequence with a common distribution which belongs to the domain of partial attraction of a semi-stable distribution \(G_ \alpha\) \((0<\alpha<1)\), \(\{n_ k\}\) be a sequence of natural numbers and \(\{t_ k\}\) be a random sequence of natural numbers. Under suitable assumptions on \(\{n_ k\}\) and \(\{t_ k\}\), the authors give laws of iterated logarithm of type \[ \limsup_ k (B^{-1}_{n_ k} S_{t_ k})^{1/\log \log n_ k}=e^{\varepsilon^*/\alpha},\;\hbox{a.s.}, \] where \(S_ n\) denotes \(\sum_{k=1}^ n X_ k\), \(\{B_{n_ k}\}\) and \(\varepsilon^*\) are decided by the distribution of \(X_ 1\) and \(\{n_ k\}\), respectively.