Linear topologies on \(\mathbb Z\) are not Mackey topologies (Q456821): Difference between revisions

The Mackey topology is an important topic in the theory of topological vector spaces. The analogue of the Mackey topology for locally quasi-convex (LQC) abelian groups was introduced by \textit{M. J. Chasco}, \textit{E. Martín-Peinador} and \textit{V. Tarieladze} [Stud. Math. 132, No. 3, 257--284 (1999; Zbl 0930.46006)], where they proved in particular that the topology of a complete metrizable LQC abelian group is the Mackey topology. Recall that a LCQ topological group \((G,\tau)\) is called Mackey or LCQ-Mackey if it has the following property: if \(\nu\) is a group LCQ topology in \(G\) such that \((G,\nu)\) has the same character group as \((G,\tau)\), then \(\nu\leq\tau\). It is not known whether every locally quasi-convex non-complete metrizable group admits the Mackey topology. Moreover, even the following natural question posed by D. Dikranjan was open. Are the \(p\)-adic topologies on \(\mathbb{Z}\) Mackey topologies? The authors prove that, for every linear non-discrete topology \(\tau\) on the integers, there exists a strictly finer metrizable LCQ group topology \(\nu\) on \(\mathbb{Z}\) such that \((\mathbb{Z}, \tau)\) and \((\mathbb{Z}, \nu)\) have the same dual group. As a corollary of this result, they give a negative answer to Dikranjan's question.