Ideal convergence generated by double summability methods (Q253740)

An admissible ideal \(I\) in \(\mathbb N\times \mathbb N\) is a collection of subsets of \(\mathbb N\times \mathbb N\) that does not contain \(\mathbb N\times \mathbb N\), contains the empty set, is closed under finite unions, and \(B\in{I}\) and \(A\subset{B}\) implies \(A\in{I}\). Similar to the technique used to generate densities via ordinary summability methods, given a nonnegative RH-regular method \(T\), one can define a density \( \mu_{T}\) by setting NEWLINE\[NEWLINE\mu_{T}(A)=\lim_{m,n} \sum \left\{t_{m,n,j,k}: (j,k)\in{A}\right\}NEWLINE\]NEWLINE for \(A\subset \mathbb N\times \mathbb N\) provided the limit exists. The author proves that if \(I\) is an ideal generated by a regular double summability matrix method \(T\), that is the product of two nonnegative regular matrix methods for single sequences, then \(I\)-statistical convergence and convergence in \(I\)-density are equivalent. He points out that the densities used to generate statistical convergence, lacunary statistical convergence, and general de la Vallée-Poussin statistical convergence are generated by these types of double summability methods, and if a matrix \(T\) generates a density with the additive property, then \(T\)-statistical convergence, convergence in \(T\)-density and strong \(T\)-summability are equivalent for bounded sequences. A counterexample is also given to show that not every regular double summability matrix generates a density with additive property for null sets.