The law of large numbers for weakly correlated random elements in the spaces \(l_p, 1 \leqslant p<\infty\) (Q2674411)

From MaRDI portal
scientific article
Language Label Description Also known as
English
The law of large numbers for weakly correlated random elements in the spaces \(l_p, 1 \leqslant p<\infty\)
scientific article

    Statements

    The law of large numbers for weakly correlated random elements in the spaces \(l_p, 1 \leqslant p<\infty\) (English)
    0 references
    12 September 2022
    0 references
    \textit{A. Khintchine} obtained a result [C. R. Acad. Sci., Paris 186, 285--287 (1928; JFM 54.0544.03)] on the Law of Large Numbers (LLN) for weakly correlated random variables. In the paper under review this result is generalized for random elements with values in Banach spaces \(l_p, 1\le p<\infty\). In order to formulate the main result of the paper, let us introduce some notation. Let \(\xi_1,\xi_2,\ldots,\xi_n,\ldots\) be a sequence of weak second order centered random elements with values in the Banach space \(l_p, 1\le p<\infty\), and let the covariance operators \(R_n\equiv R_{\xi_n}:l_p^*\to l_p\) of \(\xi_n\) satisfy the condition \[ \sigma_n^s\equiv\sum\limits_{k=1}^\infty \langle e_k, R_n e_k\rangle^{s/2}<\infty, \qquad n=1,2,\ldots, \] where \(s=\min\{2,p\}\) and \(e_k = (0, \ldots,\frac{k}{1}, 0, \ldots), \, k = 1,2, \ldots \), is a sequence of unit vectors in the dual space \( l_p^*\). Let, besides there exists a nonnegative function \(g\), defined on the set of nonnegative integers such that for the correlation coefficients \(V_{mn}\) of \(\xi_m\) and \(\xi_n\) [the reviewer, Bull. Georgian Acad. Sci. 161, No. 3, 377--379 (2000; Zbl 1030.60005)] the following inequalities hold \[ \|V_{mn}\|\le g(|m-n|) \text{ for any }m, n=1,2,\ldots. \] Under these assumptions we have Theorem. If \[ \lim\limits_{n\to\infty} \frac{1}{n^2} \left(\sum\limits_{i=0}^{n-1} g(i)\right) \left(\sum\limits_{i=1}^n \sigma_i^s\right)^{2/s}=0, \] then \[ \lim\limits_{n\to\infty}\mathbb{E}\,\left\Vert\frac{1}{n} \sum\limits_{i=1}^n\xi_i \right\Vert_{l_p}^s=0, \] where \(\mathbb{E}\) is a symbol of the mathematical expectation. In particular, the sequence \(\xi_1,\xi_2,\ldots,\xi_n,\ldots\) satisfies the LLN. In the case of a Hilbert space the theorem implies the following result. Corollary. Assume that the random elements \(\xi_1,\xi_2,\ldots,\xi_n,\ldots\) with values in a separable Hilbert space additionally have a strong second order. If \[ \lim\limits_{n\to\infty} \frac{1}{n^2} \left(\sum\limits_{i=0}^{n-1} g(i)\right) \left(\sum\limits_{i=1}^n tr(R_i)\right)=0, \] then the sequence \(\xi_1,\xi_2,\ldots,\xi_n,\ldots\) satisfies the LLN, where \(tr(R_i)\) is the trace of the covariance operator \(R_i, i=1,2,\ldots\).
    0 references
    0 references
    0 references
    0 references
    0 references
    law of large numbers
    0 references
    \(l_p\) spaces
    0 references
    covariance operator
    0 references
    cross-covariance operator
    0 references
    correlation coefficient
    0 references
    Banach spaces
    0 references
    Hilbert space
    0 references
    0 references
    0 references