Best constants in Rosenthal-type inequalities and the Kruglov operator (Q1958465): Difference between revisions

Let \(X\) be a symmetric Banach space on \([0,1]\) and \(\{f_k\}_1^n\) be independent random variables (i.r.v.)\ in \(X\). The authors present sharp Rosenthal-type estimates of the quantity \(\left\|\sum_1^nf_k\right\|_X\) [\textit{H.\,P.\thinspace Rosenthal}, Isr.\ J.\ Math.\ 8, 273--303 (1970; Zbl 0207.42803)]. The method of estimation is based on study of the following Kruglov property. Let \((B_n)\) be a sequence of pairwise disjoint measurable subsets of \([0,1]\) with \(\lambda(B_n)=1/(en!)\). The Kruglov operator is defined by the formula \[ (Kf)(\omega_0,\omega_1,\dots)= \sum_{n=1}^{\infty}\sum_{k=1}^nf(\omega_k)\chi_{B_n}(\omega_0)\,, \] where \(f\in L_1[0,1]\), \((\omega_0,\omega_1,\dots)\in\prod_{n=0}^{\infty}([0,1],\lambda_n)\) and \(\lambda_n\) is the Lebesgue measure on \([0,1]\). For convenience, \(Kf\) denotes also another random variable defined on \([0,1]\) and having the same distribution as the previous one. The space \(X\) has the Kruglov property if \(K\) is bounded in \(X\). This property has been extensively studied by \textit{M.\,S.\thinspace Braverman} [``Independent random variables and rearrangement invariant spaces'' (London Math.\ Soc.\ Lecture Note Series.\ 194; Cambridge Univ.\ Press) (1994; Zbl 0817.46031)] and the authors [see, e.\,g., \textit{S.\,V.\thinspace Astashkin} and \textit{F.\,A.\thinspace Sukochev}, Isr.\ J.\ Math.\ 145, 125--156 (2005; Zbl 1084.46020)]. Let us present a result of the paper under review. For functions \(\{f_k\}_1^n\) with support in \([0,1]\), put \(F(t)=\sum_1^nf_k(t-k+1)\) and denote by \(F^*\) the decreasing rearrangement of \(|F|\). There exists an absolute constant \(\alpha\) such that if the Kruglov operator acts boundedly in \(X\), then for any finite sequence \(\{f_k\}_1^n\subset X\) of mean zero i.r.v.'s, \[ \left\|\sum_1^nf_k\right\|_X\leq \alpha\|K\|_X\left[\|F^*\chi_{[0,1]}\|_X+\left(\sum_1^nF^*(k)^2\right)^{1/2}\right]. \]