A note on the rate of Poisson approximation of empirical processes (Q917151): Difference between revisions

Let \(X_ 1,X_ 2,..\). be a sequence of i.i.d. random vectors in \(R^ d\) with distributions F(x) \((x\in R^ d)\). Let \(F_ n(x)\) be the empirical distribution based on the sample \(X_ 1,X_ 2,...,X_ n\). The goal of the author is to approximate the empirical process \(\beta_ n(x)=\sqrt{n}(F_ n(x)-F(x))\) by a sequence of Poisson processes \(N_ n(x)\) \((x\in R^ d\), \(n=1,2,...).\) He proves a probability inequality for the sup-norm distance between \(\sqrt{n}\beta_ n(x)\) and \(N_ n(x)-nF(x)\) where \(N_ n(x)\) is a sequence of suitably constructed Poisson processes with parameter E \(N_ n(x)=nF(x)\).