Rates of uniform convergence for the empirical characteristic function

From MaRDI portal





Let \(X_ 1,X_ 2,..\). be a sequence of i.i.d. d-dimensional random vectors with common distribution function F(x), \(x\in R^ d\), and characteristic function C(t), \(t\in R^ d\). The nth empirical characteristic function of the sequence is \[ C_ n(t)=n^{- 1}\sum^{n}_{j=1}\exp (i<t,X_ j>) \] and for any extended number \(0<T\leq \infty\) consider the random variable \[ \Delta_ n(t)=\sup_{| t| <T}| C_ n(t)-C(t)|. \] An earlier result of the author and \textit{V. Totik} [ibid. 45, 141-149 (1983; Zbl 0518.60039)] shows that \(\Delta_ n(\exp (n/G_ n))\to 0\) almost surely for any sequences \(G_ n\) such that \(G_ n\to \infty\). In the present note the rate of this convergence is discussed.











This page was built for publication: Rates of uniform convergence for the empirical characteristic function

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3723410)