On \(l^p\) norms of weighted mean matrices (Q2268155): Difference between revisions

Given a weighted mean matrix \(A=(a_{n,k})\) defined by \(a_{n,k}:={\lambda_k\over\Lambda_n}\) if \(1\leq k\leq n\) and \(a_{nk}:=0\) otherwise, where \(\lambda_1=1,\) \(\lambda_k\geq 0,\) and \(\Lambda_n:=\sum_{k=1}^n \lambda_k\), let \(\| A\| _{p,p}\), be the operator norm of the correspondent map \(A : \ell^p \rightarrow \ell^p\). The main result of this paper goes as follows. Let \(1<p<\infty\) be fixed and \(A\) be a weighted mean matrix. If for any integer \(n\geq 1\) there exists a positive constant \(0<L<p\) such that \[ {\Delta_{n+1}}{\lambda_{n+1}}^{-1}\leq {\Delta_{n}}{\lambda_{n}}^{-1}(1-(L{\lambda_{n}})(p{\Delta_{n}})^{-1})^{1-p}+ L p^{-1}, \] then \(\| A\| _{p,p}\leq p(p-L)^{-1}\). This result extends another theorem of the author [J.~Math.\ Anal.\ Appl.\ 332, No.\,2, 1477--1481 (2007; Zbl 1127.47031)]. As an application, the author proves a conjecture of \textit{G.\,Bennett} [Houston J.~Math.\ 32, No.\,3, 801--831 (2006; Zbl 1122.26017)].