Taylor transformations into \(G_w\) (Q1769335)

Let \(r> 0\). The Taylor transformations \(T_r\) are given by the positive, regular matrix defined by \(a_{nk}= {k\choose n} r^{k-n}(1- r)^{n+1}\) (\(n,\,k= 0,1,2,\dots\)). Let \(c\) be the set of convergent sequences, and \(G_w= \{y: y_k= 0(r^k)\) for some \(r\in (0,w)\), \(0< w< 1\}\). The Taylor transformations for \(T_r\) which are c-c are well known in the literature (see e.g. \textit{G. H. Hardy} [Divergent series. 2nd (textually unaltered) ed. (Chelsea, New York, NY) (1991; Zbl 0897.01044)]). The author proves that \(T_r\) is a \(G_w\)-\(G_w\) matrix (i.e. \(T_r(u)\) is in \(G_w\) whenever \(u\) is in \(G_w\)), if and only if \(0< r\leq 1\). It is also proved among others that every \(G_w\)-\(G_w\) matrix \(T_r\) is \(G_w\)-translative.