scientific article; zbMATH DE number 3842965
Let \((X_ n)_{n\in {\mathbb{N}}}\) be a sequence of i.i.d. r.v.'s, let \(f(x_ 1,...,x_ m)\) be a symmetric function of m arguments. For \(n\in {\mathbb{N}}\), \(n\geq m\), let \(C(n)\subset \{(i_ 1,...,i_ m):1\leq i_ 1<...<i_ m\leq n\}\) with the property that each \(i\in \{1,...,n\}\) is present in the same number of elements of C(n). The statistic \(\Sigma_{C(n)}f(X_{i_ 1},...,X_{i_ m})\) is called a reduced U- statistic. Asymptotic normality of such statistics has been proved by \textit{G. Blom} [Biometrika 63, 573-580 (1976; Zbl 0352.62034)] and \textit{B. M. Brown} and \textit{D. G. Kildea} [Ann. Stat. 6, 828-835 (1978; Zbl 0408.62024)]. In the present paper a standardized partial sum process is considered. Under suitable assumptions the weak convergence to a Gaussian process is established, and it is shown by examples that the covariance structure of the limit process crucially depends upon the sequence of index sets \(C_ n\).
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 Q3313099)