Boundary conditions for topological quantum field theories, anomalies and projective modular functors (Q2354000)

Let \(\mathrm{Bord}(n)\) denote the usual symmetric monoidal \((\infty,n)\)-category of \(n\)-dimensional (framed) bordisms, and denote by \(\mathrm{Cob}_k^\infty(n)\) its \((n-k)\)-fold delooping; this is the \((\infty,k)\)-category of \(n\)-dimensional bordisms extended down to codimension \(k\). The paper under review defines an \textit{\(n\)-dimensional topological quantum field theory (TQFT) extended down to codimension \(k\) with moduli level \(m\)} to be a symmetric monoidal functor \[ Z: \mathrm{Cob}_k^\infty(n) \to (m+k)\text{-}\mathrm{Vect}. \] Thus, for instance, taking \(k=1\) and \(m=0\) we recover Atiyah and Segal's original definition of TQFT, and taking \(k=n\) and \(m=0\) we recover the notion of a fully extended TQFT which is the subject of the cobordism hypothesis. The authors define an \textit{anomaly} to be an invertible TQFT of moduli level \(1\), and an \textit{anomalous TQFT} \(Z_W\) with anomaly \(W\) to be a natural transformation \(Z_W: 1 \to W\), where \(1\) denotes the trivial anomaly mapping all objects to the monoidal unit and all morphisms to identities. An easy way to produce an anomaly, or TQFTs of higher moduli level, is by restricting to \(\mathrm{Cob}_k^\infty(n) \hookrightarrow \mathrm{Cob}_{k+1}^\infty(n+1)\). Next, the authors discuss the notion of TQFTs with boundary conditions. It is possible to extend \(\mathrm{Cob}_k^\infty(n)\) to an \(\infty\)-category \(\mathrm{Cob}_k^{\partial, \infty}(n)\) of \textit{constrained bordisms}, whose boundaries are decomposed as a disjoint union of incoming, outgoing, and constrained components, and gluing is not allowed along the latter. A boundary condition for a TQFT \(Z: \mathrm{Cob}_k^\infty(n) \to (k+m)\text{-}\mathrm{Vect}\) is simply a choice of extension \(\tilde Z\) to \(\mathrm{Cob}_k^{\partial, \infty}(n)\). Finally, the authors study the relation between boundary conditions and anomalous TQFTs. Let \(Z\) be an invertible \((n+1)\)-dimensional TQFT of moduli level \(0\) and denote by \(W^Z\) the \(n\)-dimensional anomaly obtained by restriction. Also, denote by \([\underline 0, 1]\) the interval with outgoing boundary component \(\{1\}\) and constrained boundary component \(\{0\}\). Given any boundary condition \(\tilde Z\) for \(Z\), the assignment \(\Sigma \mapsto \tilde Z(\Sigma \times [\underline 0, 1])\) defines an \(n\)-dimensional anomalous TQFT with anomaly \(W^Z\) (this is a kind of dimensional reduction). In general, this procedure does not establish an equivalence between boundary conditions for an invertible TQFT and anomalous TQFTs with the corresponding anomaly. The authors show that if one starts with a fully extended TQFT \(Z\), then the above procedure in fact gives rise to an equivalence. The paper also contains a section discussing \(n\)-characters of \(\infty\)-groups (defined as \(\infty\)-functors \(\rho: BG \to B(\mathrm{Pic}((n-1)\text{-}\mathrm{Vect}))\)) and their homotopy fixed points, as well as the relationship between those and anomalous TQFTs, projective representations of mapping class groups, and projective modular functors. The following generalization of the cobordism hypothesis is conjectured: the \(\infty\)-category of \(n\)-dimensional fully extended TQFTs of moduli level \(m \geq 0\) is equivalent to the \((\infty, m)\)-category obtained from the subcategory of fully dualizable objects of \((m+n)\text{-}\mathrm{Vect}\) by discarding all non-invertible \(k\)-morphisms with \(k > m\).