scientific article; zbMATH DE number 3903652

From MaRDI portal
Publication:3681647





Let \(\{\xi_ i(t)\); \(i\geq 1\}\) be independent, identical, \(T_ 0\)- regular random fields such that \(\sup_{t\in T_ 0}| \xi_ i(t)| \leq L\) with probability 1. Let \(N_*=\max \{i\geq 1: L \max \{1, H_{\xi}(2^{-i})\}\leq n2^{-i}\}\) where \(H_{\xi}(\cdot)\) is the metric entropy. Let \(S_ k^{(n)}(t)=n^{- 1/2}\sum^{k}_{i=1}(\xi_ i(t)-E\xi_ i(t))\). The following theorem holds: For any \(y\geq 0\) and \(n\geq n_ 0=\min \{k: L \max \{1,H_{\xi}(2^{- 1})\}\leq k/2\}\) the probability \[ \Pr (\max_{k\leq n}\sup_{t\in T_ 0}| S_ k^{(n)}(t)| \geq CL^{1/2}\{\sum^{N_*+1}_{i=0}(H_{\xi}(2^{-i})2^{- i})^{1/2}+y+1+(L/n)^{1/2}N_*(y^ 2+N_*)\}) \] does not exceed C exp(-y\({}^ 2)\). Additionally, if \(\sup_{t\in T^ 0}| S_ k^{(n)}(t)|\) is measurable for all n and \(k\leq n\) the following holds: \[ \Pr (\max_{k\leq n} \sup_{t\in T_ 0}| S_ k^{(n)}(t)| \leq CL^{1/2}\{\sum^{N_*+1}_{i=0}(H_{\xi}(2^{-i})2^{- i})^{1/2}+y+1\})\leq 2 \exp \{-Cy^ 2\}. \]











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3681647)