Bounded domains of generalized Riesz methods with the Hahn property (Q2443769): Difference between revisions

The author answers some questions posed in [\textit{J. Boos} and \textit{T. Leiger}, Math. Slovaca 63, No. 3, 573--586 (2013; Zbl 1349.46002)]. For a sequence space \(E\), let \(\chi\left( E\right) \) denote the linear hull of the sequences of 0's and 1's contained in \(E\). If \(\chi\left( E\right) \subset F\Rightarrow E\subset F,\) whenever \(F\) is an arbitrary FK-space, a separable FK-space, or a matrix domain \(c_{B}\), then \(E\) is said to have the Hahn property (HP), the separable Hahn property (SHP), or the matrix Hahn property (MHP), respectively. Clearly, (HP) \(\Rightarrow\) (SHP) \(\Rightarrow\) (MHP), none of the converse implications holds in general. If \(E\) has the MHP, then \(E\subset\ell^{\infty}\) and \(\chi\left( E\right) ^{\beta}=E^{\beta}\), on the other hand, a solid sequence space \(E\) with \(\chi\left( E\right) ^{\beta }=E^{\beta}=\ell^{1}\) has the SHP see [\textit{G. Bennett} et al., Stud. Math. 149, No. 1, 75--99 (2002; Zbl 0995.46010); addendum ibid. 171, No. 3, 305--309 (2005)]. In the paper under review, the author gives an example which shows that one cannot replace SHP with HP in the last result, and proves that the relation \(\chi\left( \ell^{\infty}\cap c_{R_{p}}\right) ^{\beta}=\ell^{1}\) implies the HP of \(\ell^{\infty}\cap c_{R_{p}}\), where \(R_{p}=\left( a_{nk}\right) \) is a regular Riesz matrix defined by \(a_{nk}:=p_{k}/P_{n}\) if \(k\leq n\), and \(a_{nk}:=0\) otherwise (here, \(p_{1}>0\), \(p_{k}\geq0\) and \(P_{n}:=p_{1}+\dots+p_{n}\) \(\left( k,n\in\mathbb{N}\right) \)). For a fixed index sequence \(\left( m_{n}\right) \), a generalized Riesz method \(R_{p,m}=\left( a_{nk}\right) \) in the title is defined by \(a_{nk}:=p_{k}/P_{m_{n}}\) if \(k\leq m_{n}\), and \(a_{nk}:=0\) otherwise \(\left( n,k\in\mathbb{N}\right) \). The author's main result states that, if \(\left( p_{k}\right) \notin\ell^{1}\), then the following two conditions are equivalent: \(\left( \text{a}\right) \) the matrix \(R_{p,m}\) has spreading rows, i.e., \(\lim_{n}\frac{1}{P_{m_{n}}}\sup_{1\leq k\leq m_{n}}p_{k}=0,\) and \(\left( \text{b}\right) \) the density \(d_{R_{p,m}}\) defined on the power set \(\mathcal{P}\left( \mathbb{N}\right) \) by \(d_{R_{p,m}}\left( N\right) :=\limsup_{n}\frac{1}{P_{m_{n}}}\sum_{k\in N}p_{k}\) is strongly nonatomic, i.e., for every \(\varepsilon>0\) there exists a finite partition \(N_{1} ,\dots,N_{r}\) of \(\mathbb{N}\) such that \(d_{R_{p,m}}\left( N_{\nu}\right) \leq\varepsilon\) \(\left( 1\leq\nu\leq r\right) \).