Strong bounds for weighted empirical distribution functions based on uniform spacings (Q1107206): Difference between revisions

Let \(U_ 1,U_ 2,..\). be a sequence of i.i.d. random variables with uniform distribution on (0,1). For \(n=2,3,...\), the transformed uniform spacings are defined as \({\mathbb{D}}_{i,n}=G(n D_{i,n})\), where the \(D_{i,n}'s\) are the differences of consecutive order statistics \(\{U_{i:n-1}\}\) of the first (n-1) rv's in the given sequence and where G is the distribution function of the exponential law with mean 1 \((U_{o:n-1}=0\), \(U_{n:n-1}=1).\) Let \(\hat F_ n\) be the empirical df based on the \({\mathbb{D}}_{i,n}\). The authors establish a characterization of the almost sure behavior of lim sup \(b_ n V_{n,v}\) and lim sup \(b_ n W_{n,v}\) for \(v\in [0,1/2]\) and a sequence of norming constants \(\{b_ n\}\), where \[ V_{n,v}=\sup_{0<t<1}n| \hat F_ n(t)-t| t^{1-v}\quad and\quad W_{n,v}=\sup_{0<t<1}n| \hat F_ n(t)-t| (1-t)^{1- v}. \]