G-spaces and fixed point sets (Q1591213)

This paper addresses the question of finding conditions as to when a closed subset of a compact space is the fixed point set of an autohomeomorphism. Some precedents to the present paper can be found in \textit{H. Robbins} [Isr. J. Math. 5, 373-378 (1967; Zbl 0153.24702)] and \textit{H. Schirmer} [Isr. J. Math. 7, 46-50 (1969; Zbl 0172.48202)] where the same question is studied for the closed unit ball in the Euclidean \(n\)-space. More directly related precedents are to be found in \textit{J. R. Martin} and \textit{W. Weiss} [Trans. Am. Math. Soc. 284, 337-353 (1984; Zbl 0543.54037)] and \textit{A. Chigogidze, K. H. Hofmann} and \textit{J. R. Martin} [Trans. Am. Math. Soc. 349, 4537-4554 (1997; Zbl 0888.22001)] where metric spaces and products of compact metric groups are dealt with. A more comprehensive list of related references may be found in the paper under review. The authors place the problem in the setting of free \(G\)-spaces and improve many of the results found in the papers quoted above. They obtain a wide class of compact spaces \(M\) all of whose closed subsets are the fixed point set of an autohomeomorphism of \(M\). This class consists, namely, of those compact spaces on which a compact connected group \(G\) with \(w(M)\leq w(G)\) acts freely (here, \(w(X)\) stands for the topological weight of the topological space \(X\)). In the case of compact groups the description turns out to be complete: every nonempty closed subset of an infinite compact group is the fixed point set of an autohomeomorphism if and only if \(G\) is either totally disconnected or \(w(G)=w(G_0)\), where \(G_0\) denotes the connected component of the identity in \(G\). The great generality of the above summarized results imposes a wide range of technical tools on the proofs. These technicalities are precisely justified in an unusually illuminating introduction which includes the motivation and the description of the ideas to be found in the paper. The general approach to the problem consists in defining a ``sufficiently free'' flow (induced by the action of the group) on a compact space \(M\) on which a nonsingleton compact connected group \(G\) acts freely. This was also the technique used in \textit{Chigogidze} et al. [loc. cit.] to deal with the metric case. In this paper the authors overcome the absence of a metric by using suitably defined uniform flows on weakly complete vector lattices instead of classical flows on \(\mathbb{R}\). The proof is readily reduced to the case when the acting group \(G\) is Abelian and the Lie algebra associated to \(G\) is then proved to have a weakly complete vector lattice structure which is used to define a uniform flow on \(M\).