scientific article; zbMATH DE number 1323364

Publication date 10 August 1999

Gaussian random field Lévy-Baxter theorem mean square convergence

The main result of this article is the following. Let for the mean square continuous Gaussian random field \(X(t), t\in [0,b]^{m}, b>1\), with zero mean, for some \(\gamma\in(0,2)\) the following conditions be fulfilled: 1) there exists a function \(f:Q=[0,1]^{m}\to R\) such that \(E[X(t+h)-X(t)]^{2}/ h^{\gamma}\to f(t)\) as \(h\to 0+\) uniformly on \(Q\); 2) there exist \(C>0\) and \(\delta>0\) such that \(\forall h\in (0,\delta), \forall \{t,s\}\subset Q, |t_{i}- s_{i}|>3h, 1\leq i\leq m\), the inequality \[ |E[(X(t+y)-X(t))(X(s+y)-X(s))]|\leq Ch^{2}/\|t-s\|^{(2-\gamma)} \] holds; 3) \(a_{n}\to\infty\) as \(n\to\infty\). Then \(a_{n}^{(\alpha\gamma/2-m)}S_{n}^{\alpha}\to\pi^{-1/2} 2^{\alpha/2}\Gamma((1+\alpha)/2) \int_{Q}f^{\alpha/2}(t)dt\) in the mean square as \(n\to\infty\). Here \[ \begin{aligned} S_{n}^{\alpha}&= \sum_{k\in D(n)}|X_{k}|^{\alpha},\quad n\geq 1,\quad \alpha\geq 2,\\ X_{k}&=X((k_{1}+1)/a_{n}, \dots,(k_{m}+1)/a_{n})-X(k_{1}/a_{n},\dots,k_{m}/a_{n}),\\ D(n)&=\{(k_{1},\ldots,k_{m}):0\leq k_{i}\leq a_{n}-1, 1\leq i\leq m\}.\end{aligned} \] If, in addition \(\sum_{n=1}^\infty((\ln a_{n})/a_{n}+a_{n}^{\gamma-2}<\infty\), then the convergence is almost sure. The example of Chentsov random field is considered.

Recommendations

Cited in

(18)

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 Q4255198)