Finite symmetries of \(S^4\) (Q254055)

The basic question considered here is whether a finite group \(G\) which acts effectively and locally linearly on \(S^4\) must be isomorphic to a subgroup of \(O(5)\). The first main result completes earlier work of \textit{M. Mecchia} and \textit{B. Zimmermann} [Topology Appl. 158, No. 6, 741--747 (2011; Zbl 1234.57021)] by showing that if \(G\) acts locally linearly on a \(\mathbb{Z}\)-homology 4-sphere \(\Sigma\) and the action is orientation preserving then \(G\) embeds in \(SO(5)\). The argument uses the Atiyah-Bott theorem relating the tangential actions at distinct fixed points, together with a number of lemmas of a more elementary nature. Each of the restrictions on the action is necessary, even if \(\Sigma=S^4\). In the second part of the paper 4-dimensional TOP surgery over generalized quaternionic groups \(Q(8p,q)\) is used to construct counterexamples, for certain pairs of primes \(p,q\). If \(p,q\) are distinct primes then \(Q(8p,q)\) does not embed in \(O(5)\). It remains unknown whether there is a finite group \(G\) which acts \textit{smoothly} on \(S^4\) but which does not embed in \(O(5)\). The surgery arguments are elaborated to show that for certain pairs \(p,q\) there are two actions of \(Q(8p,q)\) on \(S^4\), one stably smoothable and another not stably smoothable.