Regular matrices of unbounded linear operators

TL;DR

This paper extends classical Silverman--Toeplitz theorems to matrices of linear operators acting on Banach spaces, characterizing regularity through ideal convergence and establishing connections with multidimensional and scalar cases.

Contribution

It introduces a multidimensional Silverman--Toeplitz type theorem for matrices of operators using ideal convergence, generalizing classical results and characterizing various matrix classes.

Findings

Characterization of regular matrices of operators via ideal convergence

Extension of Silverman--Toeplitz theorems to multidimensional and operator matrices

Generalization of Hahn--Schur theorem for operator matrices

Abstract

Let $X,Y$ be Banach spaces, and fix a linear operator $T \in \mathcal{L}(X,Y)$, and ideals $\mathcal{I}, \mathcal{J}$ on $\omega$. We obtain Silverman--Toeplitz type theorems on matrices $A=(A_{n,k}: n,k \in \omega)$ of linear operators in $\mathcal{L}(X,Y)$, so that $$ \mathcal{J}\text{-}\lim Ax=T(\hspace{.2mm}\mathcal{I}\text{-}\lim x) $$ for every $X$-valued sequence $x=(x_0,x_1,\ldots)$ which is $\mathcal{I}$-convergent [and bounded]. This allows us to establish the relationship between the classical Silverman--Toeplitz characterization of regular matrices and its multidimensional analogue for double sequences, its variant for matrices of linear operators, and the recent version (for the scalar case) in the context of ideal convergence. As byproducts, we obtain characterizations of several matrix classes and a generalization of the classical Hahn--Schur theorem. In the proofs we will use an ideal version of the Banach--Steinhaus theorem which has been recently obtained by De Bondt and Vernaeve in [J.~Math.~Anal.~Appl.~\textbf{495} (2021)].