The question of equivalence for generalized Hausdorff matrices (Q864703): Difference between revisions

Let \(\alpha\) be a real number and \(\{\mu_n \}\) a real sequence. Let \(\Delta^n\) be the following operator: \(\Delta\mu_k=\mu_k -\mu_{k+1}\), \(\Delta^{n+1}\mu_k=\Delta(\Delta^n (\mu_k))\) A generalized Hausdorff matrix \(H^{(\alpha)}\) is an infinite matrix (as an operator) whose entries are defined by \(h^{(\alpha)}_{nk}=(^{n+\alpha}_{n-k})\Delta^{n-k}{\mu_n}\) if \(0\leq k \leq n\) and \(h^{(\alpha)}_{nk}\)={0} if \(k>n\). The set of these matrices forms a commutative, non-Archimidean domain. Two infinite matrices \(A\) and \(B\) are said to be equivalent on \(c\), the space of convergent sequences, if \(x=\{ x_k \}\) is a sequence such that the limit of \(A_n x=\sum_{k=0}^{\infty} a_{nk} x_k\) exists, then the limit of \(B_n x =\sum_{k=0}^{\infty} b_{nk} x_k\) exists and conversely. Two infinite matrices \(A\) and \(B\) are said to be equivalent on \(l^p\), \(1<p<\infty\) if for each sequence \(x=\{ x_k\}\) if \(\{A_n(x)\} \in l^p\) then \(\{B_n(x)\}\in l^p\) and vice versa. The authors consider the equivalence problem of infinite matrices for \(\alpha\geq{0}\) on \(c\) and obtain the following results. 1. There are cases in which \(H^{(\alpha)}\) and \(H^{(\beta)}\) are equivalent, but not equivalent to \(H\) (\(\alpha={0}\)). 2. There are cases in which \(H^{(\alpha)}\) and \(H^{(\beta)}\) are not equivalent to each other. 3. There are cases in which \(H^{(\alpha)}\), \(H^{(\beta)}\) and \(H\) are all equivalent. The authors also consider the equivalence problem in \({l^{p}}\) for \(1<{p}<\infty\). The paper ends with a conjecture and several results on Cesaro matrices.