A characterization of Pontryagin-van Kampen duality for locally convex spaces (Q1612197)

The Pontryagin-van Kampen theorem lies at the core of the duality theory of topological Abelian groups. It states that locally compact Abelian groups are reflexive, that is, for a locally compact Abelian group \(G\), the natural homomorphism from the group into its bidual is a topological isomorphism. In 1952, the first result about the validity of the the Pontryagin-van Kampen theorem in the class of topological vector spaces (understood as additive topological Abelian groups) appeared. M. Smith proved in [\textit{M. F. Smith}, Ann. Math. (2) 56, 248-253 (1952; Zbl 0804.46002)] that reflexive locally convex spaces and Banach spaces satisfy this theorem, in other words, they are reflexive groups. At that time the only known reflexive groups were locally compact Abelian groups, direct and inverse limits of sequences of locally compact Abelian groups and products and direct sums of reflexive groups. The first part of the paper is devoted to finding a new characterization of Pontryagin-van Kampen duality for locally convex spaces: A locally convex space is a reflexive group if and only if it has the convex compactness property (L2 property in the paper) and the \(\tau_c\)-compacts in the dual are equicontinuous. In order to clarify and place their work in context, the authors dedicate a second part referring and commenting, with very interesting examples, the main results in the literature being related with the validity of the Pontryagin-van Kampen theorem in topological vector spaces. The authors of the paper under review find some flaws in Lemma 2 of the paper by \textit{S. Kye} [Chin. J. Math. 12, 129-136 (1984; Zbl 0594.46001)], but still leave open the question about the validity of Kye's statement (actually this question has been answered negatively by \textit{S. Hernández} [Math. Z. 238, 493-503 (2001; Zbl 1014.22001)]). In fact, the purpose of finding a characterization of the locally convex spaces which are reflexive groups, making use only of properties of the topology of the underlying vector space, is not yet reached. It seems necessary to give a description, within the group, of the polars of compacts in the dual group similar to the one given for precompacts in Lemma 6. The reader interested in the subject can also find valuable information in the recent paper [\textit{S. S. Akbarov}, Izv. Math. 64, 653-694 (2000; Zbl 0996.46001)].