Freed-Moore \(K\)-theory (Q6598521)

This paper is motivated by \textit{D. S. Freed} and \textit{G. W. Moore} [Ann. Henri Poincaré 14, No. 8, 1927--2023 (2013; Zbl 1286.81109)], which uses a version of twisted equivariant \(K\)-theory to study states of condensed quantum matter (like topological insulators, for example). The author works in the context of \emph{local quotient groupoids} \(\mathcal X\) in the sense of \textit{D. S. Freed} et al. [J. Topol. 4, No. 4, 737--798 (2011; Zbl 1241.19002)]. He then defines \(K\)-groups \(^\phi K^{(\tau,c)+n}(\mathcal X)\), where \(\phi\colon\,\mathcal X \to \mathrm{pt}//\mathbb Z_2\), \((\tau,c)\) is the data of a \(\phi\)-twist (the collection of which are classified by \(H^3(\mathcal X;\mathbb Z_\phi)\times H^1(\mathcal X;\mathbb Z_2)\)), and \(n\in \mathbb Z\). The very general (but a bit complicated) definition of twist data makes it possible to encompass many different versions of \(K\)-theory, including twisted equivariant \(K\)-theory and twisted \(KR\)-theory, in a single framework. The main content of the paper consists of two different approaches to constructing the Freed-Moore \(K\)-theory, one using Fredholm families and one using a Karoubi-like approach. These are used to prove Bott periodicity, using the algebraic periodicity of Clifford algebras, as well as the Thom isomorphism theorem. Then the author shows that the two theories are equivalent. There is one subtlety that comes into play. Namely, the isomorphism between the Fredholm-type theory and the Karoubi-type theory in general requires changing the twist in a way that doesn't change the \(K\)-groups.