Weak convergence of infinite order U-processes (Q1181095)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Weak convergence of infinite order U-processes
scientific article

    Statements

    Weak convergence of infinite order U-processes (English)
    0 references
    27 June 1992
    0 references
    \(X_ 1,X_ 2,\dots\) are IID with distribution function \(F\). \(\theta=\theta(F)\) is a parametric function which equals \(E_ F(h(X_ 1,X_ 2,\dots))\) for some function \(h: R^ \infty\to R\). Define \[ \tilde U_ n(t)=(1/[nt]!)\sum_{P^{[nt]}} h(X_{i_ 1},\dots,X_{i_ m},X_{n+1},\dots) \] where \(P^ a=\{1,\dots,a\}^ a\), \(m=m(n)=\alpha n^ r\) for some \(\alpha\) in \((0,1]\) and \(r\) in \((0,1]\), \(\tilde U_ n(t)=0\) if \(t<\alpha\). Let \(h^ n: R^ m\to R\) be functions such that \(E[h(X_ 1,\dots,X_ n,\dots)-h^ n(X_ 1,\dots,X_ m)]^ 2\) approaches zero as \(n\) increases. Define \[ U_ n(t)=(1/[nt]!)\sum_{P^{[nt]}} h^ n(X_{i_ 1},\dots,X_{i_ m}). \] \(U_ n(t)=0\) if \(t<\alpha\). Define \[ C^ b_ a = \{(i_ 1,\dots,i_ b)\in\{1,\dots,a\}^ b\mid 1\leq i_ 1 < i_ 2 < \dots < i_ b \leq a\} \] and \[ \tilde X_ n(t)={[nt]\choose m}^{- 1}\sum_{C^{[nt]}_ m} [h(X_{i_ 1},\dots,X_{i_ m}, X_{n+1},\dots)-h_{m-1}(X_{i_ 1},\dots,X_{i_{m- 1}},X_{n+1},\dots)]. \] The limit distributions of \(n^{c/2}(\tilde U_ n(t)-\theta)\), \(n^{c/2}(\tilde X_ n(t)-\theta)\), and \(n^{c/2}(\tilde U_ n(t)-\tilde X_ n(t))\) are given in terms of multiple Wiener integrals, where \(c\) is a suitably chosen positive integer.
    0 references
    0 references
    infinite order \(U\)-processes
    0 references
    limit distributions
    0 references
    multiple Wiener integrals
    0 references

    Identifiers