Remarks on anomalous symmetries of \(C^*\)-algebras (Q2244090)

A twisted action of a group~\(G\) on a unital \(C^*\)-algebra~\(A\) is given by automorphisms \(\alpha_g\) and unitaries \(u_{g,h}\) for \(g,h\in G\) such that \(u_{g,h} \alpha_g (\alpha_h(a)) u_{g,h}^* = \alpha_{g h}(a)\) for all \(g,h\in G\), \(a\in A\), and \(u\) satisfies the cocycle condition \(u_{gh,k}^* u_{g,hk} \alpha_g(u_{hk}) u_{g,h}^* = 1\) for all \(g,h,k\in G\); the latter product of unitaries is always central in~\(A\). More generally, one may ask the product above to be equal to a given \(3\)-cocycle \(\omega(g,h,k)\) on \(G\) with values in \(U(1)\) or the centre of~\(A\). This is called an anomalous action of \(G\) on \(A\). Anomalous actions appear naturally in the study of \(C^*\)-tensor categories and their actions on \(C^*\)-algebras. They may also be rewritten as actions of \(2\)-groups or crossed modules. This article studies the existence of such anomalous actions on some classes of \(C^*\)-algebras, namely, commutative \(C^*\)-algebras and their stabilisations, Roe \(C^*\)-algebras, and the quotients of Roe \(C^*\)-algebras by the ideal of compact operators. On the one hand, some \(C^*\)-algebras admit no anomalous actions at all. Namely, this is true for the Roe \(C^*\)-algebra of any discrete metric space of bounded geometry; for \(C(X)\) for a connected, locally path connected compact space~\(X\) provided \(H^1(X,\mathbb Z)\) vanishes; and it remains true for the \(C^*\)-stabilisation \(C(X) \otimes \mathbb K\) if, in addition, to the assumptions above, all complex line bundles over~\(X\) are trivial. On the other hand, for any \(n\ge 2\), group~\(G\) and 3-cocycle~\(\omega\), there is a connected closed \(n\)-manifold~\(M\) and an \(\omega\)-anomalous action on \(C(M)\otimes \mathbb K\). In addition, there is a discrete metric space~\(X\) with bounded geometry and property~A and an \(\omega\)-anomalous action on the quotient of the Roe \(C^*\)-algebra of~\(X\) by the ideal of compact operators. Another interesting result in the article says that if a \(C^*\)-algebra admits an anomalous action with a cocycle that is not a coboundary, then the induced action cannot fix any pure state. For instance, if the \(C^*\)-algebra admits a unique trace, then any automorphism must fix it, and so the \(C^*\)-algebra cannot admit any anomalous action. A key step to prove existence of anomalous actions is a construction of anomalous actions on certain twisted crossed products. This is reminiscent of a construction of crossed module actions on untwisted crossed products in Example~8 in [\textit{A. Buss} et al., Math. Ann. 352, No. 1, 73--97 (2012; Zbl 1242.46075)].