Free actions on surfaces that do not extend to arbitrary actions on 3-manifolds (Q2116736)

Working with orientable manifolds and orientation-preserving maps, the basic question considered in the present paper is the following. (When) does a \textit{free action} of a finite group \(G\) on a closed surface \(T\) extend to a not necessarily free action on a compact 3-manifold \(N\) with boundary \(\partial N = T\) (as considered recently and answered positively for some classes of finite groups by \textit{J. E. Domínguez} and \textit{C. Segovia} [Topology Appl. 305, Article ID 107898, 14 p. (2022; Zbl 1478.57027)]? By classical bordism theory, the obstruction to an extension which is free also on the 3-manifold \(N\) is an element of the 2-dimensional bordism group \(\Omega_2(BG) \cong H_2(G) \cong H_2(BG)\) (isomorphic to the Schur multiplier \(H_2(G)\) of \(G\)), so in general a free action does not extend freely. In the present paper, the author gives the first examples of free actions on a surface which do not extend to a compact 3-manifold, i.e. neither freely nor non-freely. ``We initially found them using a systematic search with GAP. We later realized that work of \textit{P. Moravec} [Am. J. Math. 134, No. 6, 1679--1704 (2012; Zbl 1346.20072)] and others [\textit{A. Hoshi} et al., Asian J. Math. 17, No. 4, 689--714 (2013; Zbl 1291.13012)] on the calculation of Bogomolov multipliers allows for a computer-free proof (given in main body of this note) that infinitely many counterexamples exist. Intriguingly, any group with such a non-extending action is also a counterexample to the Noether problem over the complex numbers \(\mathbb C\).'' The Bogomolov multiplier of a finite group is isomorphic to the quotient of the Schur multiplier by the subgroup generated by toral classes (induced by maps from the torus into \(BG\)); its elements occur as obstructions to extend a free action on a surface to a \textit{non-singular action} on a 3-manifold, i.e. such that the union of the the fixed point sets of nontrivial elements consists of disjoint circles (maybe empty), avoiding branch points. Then if \(G\) has odd order (which avoids such branch points since the local stabilizer of a branch point in \(SO(3)\) is a spherical group of even order) and the Schur multiplier is not generated by toral classes, \(G\) has a free surface action which does not extend. It is known that free actions for certain classes of finite groups such as abelian groups, dihedral groups, alternating and symmetric groups do extend (in fact for dihedral grous all actions extend but in general this is not true for non-free actions: already for cyclic and abelian groups one needs some conditions). In the concluding remarks of the paper the author indicates some interesting further directions and projects (for example, he asks whether there exists an action which extends only singularly).