RUS  ENG
Full version
JOURNALS // Teoriya Veroyatnostei i ee Primeneniya // Archive

Teor. Veroyatnost. i Primenen., 1997 Volume 42, Issue 2, Pages 341–350 (Mi tvp1807)

This article is cited in 30 papers

Short Communications

On an exact constant for the Rosenthal inequality

R. Ibragimov, Sh. Sharahmetov

Tashkent State University

Abstract: Let $\xi_1,\dots,\xi_n$ be independent random variables having symmetric distribution with finite $p$th moment, $2<p<\infty$. It is shown that the precise constant $C^*_p$ in Rosenthal's inequality
$$ \biggl\|\sum_{i=1}^n\xi_i\biggr\|\le C_p\max\biggl(\biggl\|\sum_{i=1}^n\xi_i\biggr\|_2,\biggl(\sum_{i=1}^n\|\xi_i\|_p^p\biggr)^{1/p}\biggr) $$
has the form
\begin{align*} C_p^*&=\biggl(1+\frac{2^{p/1}}{\pi^{1/2}}\Gamma\biggl(\frac{p+1}2\biggr)\biggr)^{1/p}, \qquad 2<p<4, C_p^*&=\|\xi_1-\xi_2\|_p, \qquad p\ge4, \end{align*}
where $\Gamma(\alpha)=\int_0^\infty x^{\alpha-1}e^{-x} dx$, and $\xi_1$, $\xi_2$ are independent Poisson random variables with parameter 0.5. It is proved also that
$$ \lim_{p\to\infty}C_p^*\frac{\ln p}p=\frac1e. $$
.

Keywords: Rosenthal's inequality, random variables withsymmetric distribution, Poisson random variable, moment.

Received: 05.10.1995

DOI: 10.4213/tvp1807


 English version:
Theory of Probability and its Applications, 1998, 42:2, 294–302

Bibliographic databases:


© Steklov Math. Inst. of RAS, 2025