Equivariant modular categories via Dijkgraaf-Witten theory (Q1951485)

The authors present an equivariant generalization of Dijkgraaf-Witten theory as a 3-extended TQFT. At the same time they resolve an associated algebraic Tannaka-Krein reconstruction problem. After an overview of the main results, the paper starts with a review of Dijkgraaf-Witten (DW) theory as a 3-dimensional topological quantum field theory (TQFT). (Throughout, the authors limit their attention to DW theory with trivial action in the path integral.) This review has the purpose to present DW theory in a way amenable to the generalization the authors subsequently perform and might be of independent interest to the reader. The point of view taken is throughout a category theoretic one and the theory is descried as a 3-extended TQFT. That is, on the geometric side the ingredients are (collared) oriented compact manifolds of dimension 3, 2 and 1. Algebraically, this setting is captured in terms of a symmetric monoidal 2-category of cobordisms. A 3-extended TQFT is then a 2-functor from this category to a 2-category of vector spaces. This functor is described explicitly for DW theory, with particular attention paid to the category associated to a circle. As recalled in the paper, the latter can be naturally identified as the category \(C(G)\) of representations of the Drinfeld double \(D(G)\) of the finite group \(G\) that underlies the DW theory. The authors proceed to present their main construction, an equivariant 3-extended TQFT generalizing DW theory. Given a finite group \(G\) as the gauge group of DW theory, the additional data of the equivariant theory consists of another finite group \(J\) with a weak action on \(G\). The concept of weak action used amounts to that of an extension \(G\to H\to J\) of \(J\) by \(G\) together with a unital section \(J\to H\). The geometric data of the equivariant theory is richer than that of the original DW theory. The manifolds of dimension 3, 2, and 1 are equipped in addition with principal \(J\)-bundles. While the basic objects to construct the original DW theory are \(G\)-bundles over the manifolds, in the equivariant theory these a twisted \(G\)-bundles. The ''twist data'' is precisely given by the principal \(J\)-bundles. Of main interest is again the category associated to the circle. The twist datum on the circle is parametrized by an element of the group \(J\), yielding one category \(C(G)_j\) for every \(j\in J\). The relevant object for describing all twisted sectors coherently, including their fusion etc.\ is then the direct sum category \(C^J(G)=\oplus_{j\in J} C(G)_j\). In the subsequent more algebraic part of their paper the authors show that, similar to \(C(G)\) in the ordinary DW case, the category \(C^J(G)\) also arises as a category of representations. To this end the authors develop the notion of a \(J\)-ribbon Hopf algebra and that of a \(J\)-modular category arising as the category of representations of the former. They then introduce the notion of a \(J\)-Drinfeld double \(D^J(G)\) given a finite group \(J\) acting weakly on the finite group \(G\). This is a \(J\)-ribbon Hopf algebra and its category of representation can be identified precisely with the category \(C^J(G)\). In this way the authors are moreover able to show that \(C^J(G)\) has further desired properties such as a suitable modularity property.