Strong approximation of empirical process with independent but non- identically distributed random variables

Let \(\{Y_ t,t=1,2,...\}\) be independent random variables with continuous distribution functions \(F_ i(y)\). For any y, denote \(s=\bar F_ t(y)=t^{-1}\sum^{t}_{i=1}F_ i(y)\). The empirical process is defined by \(t^{-}R(s,t)\) where \[ R(s,t)=t(t^{- 1}\sum^{t}_{i=1}I_{\bar F_ t(Y_ i)\leq s}- s)=\sum^{t}_{i=1}I_{\bar F_ t(Y_ i)\leq s}-ts= \] \[ \sum^{t}_{i=1}I_{\bar F_ t(Y_ i)\leq \bar F_ t(y)}-t\bar F_ t(y)=\sum^{t}_{i=1}I_{Y_{i\leq y}}-\sum^{t}_{i=1}F_ i(y). \] The purpose of this paper is to investigate the asymptotic properties of the empirical process R(s,t). We shall prove that for some integer sequence \(\{t_ k\}\), there is a \(\tilde K\)-process \(\tilde K(s,t)\) such that \[ \sup_{0\leq s\leq 1}| R(s,t_ k)-\tilde K(s,t_ k)| =O(t_ k^{1/2}(\log t_ k)^{-1/4}(\log \log t_ k)^{1/2})\quad a.s. \] where \(\tilde K(s,t)\) is a certain two-parameter Gaussian process.

Recommendations

Cites work

Cited in

(8)

This page was built for publication: Strong approximation of empirical process with independent but non- identically distributed random variables

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1092509)