Dijkgraaf-Witten invariants of surfaces and projective representations of groups (Q2469971)

To any finite dimensional semi-simple algebra \(A\) over a field \(F\) there corresponds a state-sum invariant of oriented surfaces [\textit{M. Fukuma, S. Hosono} and \textit{H. Kawai}, Commun. Math. Phys. 161, No.~1, 157--175 (1994; Zbl 0797.57012)]. One common choice is to take \(A\) to be an algebra of square matrices over \(F\). This paper instead chooses \(A\) to be a twisted group algebra \(A^{(c)}\) of a finite group \(G\) associated to a normalized \(2\)-cocycle \(c\), defined to be the formal span of elements of \(G\) with multiplication law \(\cdot\) given by the formula \(g_1\cdot g_2 = c(g_1,g_2)g_1g_2\). There are two natural choices of bases for \(A^{(c)}\). First there is the basis of group elements, and second there is the basis of matrix elements in the projective representations of \(G\) associated with \(c\). With respect to the first basis, the author proves that the state-sum invariant obtained coincides with the Dijkgraaf--Witten invariant \(Z_{[c]}(M)\) of the oriented surface \(M\). Choosing the second basis yields a Verlinde-type formula for the Dijkgraaf-Witten invariant, proving that \(Z_{[c]}(M)\) is a positive integer if \(F\) has characteristic zero and \(M\neq S^2\), and is in \(\mathbb{Z}/p\mathbb{Z}\) if \(F\) has characteristic \(p>0\). It is striking how natural this result is made to look by the author's quantum-algebraic approach, since from the algebraic-topological viewpoint it seems highly non-trivial. The paper concludes by proving a version of the above results for non-orientable surfaces.