scientific article; zbMATH DE number 4013655

From MaRDI portal
Publication:3025970





Let \(X_{1n},...,X_{nn}\) for each \(n\geq 1\) be i.i.d. r.v.'s with absolutely continuous distribution function \(F_ 0\) or \(F_ n\) concentrated on [0,1], let \[ \nu_{in}=\sum_{j\leq n}I[i/N<X_{jn}\leq (i+1)/N],\quad and \] \[ X_{k_ n}^{n,N}(t)=c_{n,N}^{- 1/2}\sum_{i<Nt}[g_ N(\nu_{in},i/N)-E^{k_ n}g_ N(\nu_{in},i/N)], \] where \(g_ N(x,t)\), \(x=0,1,...\), \(t\in [0,1]\) are measurable functions satisfying the condition \((a_ r)\quad g_ N(0,.)=...=g_ n(r-1,.)=0\), \(g_ N(r,.)\neq 0\), \(r\geq 2\), \(c_{n,N}=n^ r/N^{r-1}\to \infty\), \(n=o(N)\), and \(E^{k_ n}\) denotes the expectation with respect to \(F_{k_ n}\), \(k_ n=0,n.\) In the paper two theorems concerning the weak convergence of \(X_{k_ n}^{n,N}\) in D[0,1) to a Wiener process with time parameter and drift depending on \(g_ N\), \(F_ 0\) and \(F_ N\), are proved under suitable additional assumptions. The method of proof is based on the functional CLT for semimartingales given by \textit{R. Sh. Liptser} and \textit{A. N. Shiryaev}, ibid. 25, 683-703 (1980; Zbl 0454.60032). The presented results extend earlier ones obtained by the author for \(r=1\) [cf. ibid. 30, 584-588 (1985; Zbl 0578.60031)].











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 Q3025970)