3-dimensional defect TQFTs and their tricategories (Q2302239): Difference between revisions

An \(n\)\textit{-dimensional topological quantum field theory} (TQFT) á la \textit{M. Atiyah} [Publ. Math., Inst. Hautes Étud. Sci. 68, 175--186 (1988; Zbl 0692.53053); Turk. J. Math. 21, No. 1, 1--7 (1997; Zbl 0890.57019)] and \textit{G. Segal} [Philos. Trans. R. Soc. Lond., Ser. A, Math. Phys. Eng. Sci. 359, No. 1784, 1389--1398 (2001; Zbl 1041.81094)] is a symmetric monoidal functor from \(\mathrm{Bord}_{n}\) to vector spaces, providing topological invariants of manifolds in particular. TQFT with defects can provide finer invariants and a far and away richer theory. It is completely classical that \(2\)-dimensional \textit{closed} TQFTs are equivalent to commutative Frobenius algebras [\textit{J. Kock}, Frobenius algebras and 2D topological quantum field theories. Cambridge: Cambridge University Press (2004; Zbl 1046.57001)]. By allowing labelled boundaries, one ascends to \textit{open/closed} TQFTs \(\mathcal{Z}^{\mathrm{oc}}\)\ [\textit{C. I. Lazaroiu}, Nucl. Phys., B 603, No. 3, 497--530 (2001; Zbl 0983.81090)], which are known to be equivalent to a commutative Frobenius algebra (which \(\mathcal{Z}^{\mathrm{oc}}\)\ assigns to \(\boldsymbol{S}^{1}\)) together with a \textit{Calabi-Yau category} (whose \(\mathrm{Hom}\) are what \(\mathcal{Z}^{\mathrm{oc}}\)\ does to intervals with labelled endpoints) and certain relations between the two. As far as all known examples of open/closed TQFTs\ are concerned, it is only the Calabi-Yau category that matters, the Frobenius algebras being recovered as its Hochshild cohomology [\textit{K. Costello}, Adv. Math. 210, No. 1, 165--214 (2007; Zbl 1171.14038)]. This means that the transition from \(2\)-dimensional closed to open/closed TQFTs is from algebras to categories. This naturally leads us to consider \(2\)-dimensional defect TQFTs, of which closed and open/closed flavors are special cases. Every \(2\)-dimensional defect TQFT naturally gives rise to pivotal \(2\)-category [\textit{A. Davydov} et al., Proc. Symp. Pure Math. 83, 71--128 (2011; Zbl 1272.57023)], subsuming B-twisted sigma models [\textit{A. Căldăraru} and \textit{S. Willerton}, New York J. Math. 16, 61--98 (2010; Zbl 1214.14013)], symplectic manifolds and Lagrangian correspondences [\textit{K. Wehrheim}, Assoc. Women Math. Ser. 6, 3--90 (2016; Zbl 1396.57044)] and affine Landau-Ginzburg models [\textit{N. Carqueville} and \textit{D. Murfet}, Adv. Math. 289, 480--566 (2016; Zbl 1353.18004)]. The notion of defect TQFT generalizes to arbitrary dimension in principle. An \(n\)-dimensional defect TQFT should be a symmetric monoidal functor \[ \mathrm{Bord}_{n}^{\mathrm{def}}\rightarrow\mathrm{Vect}_{k} \] where \(\mathrm{Bord}_{n}^{\mathrm{def}}\)\ is some suitably augmented version of \(\mathrm{Bord}_{n}\)\ allowing decorated submanifolds of various codimensions. Such TQFTs are expected to be described by \(n\)-categories with additional structure [\textit{A. Kapustin}, in: Proceedings of the international congress of mathematicians (ICM 2010), Hyderabad, India, August 19--27, 2010. Vol. III: Invited lectures. Hackensack, NJ: World Scientific; New Delhi: Hindustan Book Agency. 2021--2043 (2011; Zbl 1233.57018)], but details have not been worked out for \(n>2\), nor even a precise definition of defect TQFT has appeared in the literature for \(n>2\). The principal objective in this paper is to remedy both issues for \(n=3\), paving the way to still higher dimensions. Inspired constantly by Lurie's splendid work on the \textit{cobordism hypothesis} [\textit{J. Lurie}, in: Current developments in mathematics, 2008. Somerville, MA: International Press. 129--280 (2009; Zbl 1180.81122)], higher categories are predominant in \textit{extended} TQFT, where decompositions along lower-dimensional subspaces promote \(\mathrm{Bord}_{n}\)\ itself to a higher category. This includes the classification of fully extended \(3\)-dimensional TQFTs valued in the \(\left(\infty,3\right)\)-category of monoidal categories by spherical fusion categories [\textit{C. L. Douglas} et al., ``Dualizable tensor categories'', Preprint, \url{arXiv:1312.7188}], the classification of \(3\)-\(2\)-\(1\) extended TQFTs valued in \(2\)-Vect by modular tensor categories [\textit{B. Barlett} et al., ``Modular categories as representations of the 3-dimensional bordism 2-category'', Preprint, \url{arXiv:1509.06811}], and the work on boundary conditions and surface defects [\textit{J. Fuchs} et al., Commun. Math. Phys. 321, No. 2, 543--575 (2013; Zbl 1269.81169); \textit{A. Kapustin} and \textit{N. Saulina}, Proc. Symp. Pure Math. 83, 175--198 (2011; Zbl 1248.81206)]. Although extended TQFTs assume higher categories and higher functors from the outset, this paper contends that a defect TQFT is an ordinary symmetric monoidal functor, while everything leading to higher-categorical structures is contained in an augmented bordism category. This paper consists of four sections. The authors claim that the most conceptual and economic way of systematically studying \(3\)-dimensional defect TQFT is through the notion of defect cobordism, which is introduced in \S 2.1. Roughly speaking, defect bordisms are bordisms coming with a stratification whose \(1\)-, \(2\)- and \(3\)-strata are decorated by a choice of label sets \(\mathbb{D}\). Defect morphisms form the morphisms in a symmetric monoidal category \(\mathrm{Bord}_{3}^{\mathrm{def}}\left( \mathbb{D}\right) \), whose objects are compatibly decorated stratified closed surfaces. \S 3 defines a \(3\)-dimensional defect TQFT to be a symmetric monoidal functor \[ \mathcal{Z}:\mathrm{Bord}_{3}^{\mathrm{def}}\left( \mathbb{D}\right)\rightarrow\mathrm{Vect}_{k} \] It is well known that every weak \(2\)-category is equivalent to a strict \(2\)-category, but the corresponding statement does not hold for the \(3\)-dimensional case [\textit{R. Gordon} et al., Coherence for tricategories. Providence, RI: American Mathematical Society (AMS) (1995; Zbl 0836.18001)]. As is reviewed in \S 3.1, the generically strictest version of a \(3\)-category is what is called a \textit{Gray category}. The following main result (Theorem 3.12 and Theorem 3.13) is established in \S 3.3 and \S 3.4. Theorem. Every \(3\)-dimensional defect TQFT \(\mathcal{Z}:\mathrm{Bord}_{3}^{\mathrm{def}}\left( \mathbb{D}\right) \rightarrow\mathrm{Vect}_{k}\)\ gives rise to a \(k\)-linear Gray category with duals \(\mathcal{T}_{\mathcal{Z}}\). The Gray category \(\mathcal{T}_{\mathcal{Z}}\)\ is an invariant associated to \(\mathcal{Z}\)\ measuring among other things how closed \(3\)-dimensional TQFTs associated to unstratified bordisms are glued together. Roughly speaking, the idea behind the construction of \(\mathcal{T}_{\mathcal{Z}}\)\ is to directly transport the decorated structure of bordisms in \(\mathrm{Bord}_{3}^{\mathrm{def}}\left( \mathbb{D}\right) \)\ into the graphical calculus for Gray categories with duals [\textit{J. W. Barrett} et al., ``Gray categories with duals and their diagrams'', Preprint, \url{arXiv:1211.0529}]. \S 4 gives three classes examples of \(3\)-dimensional defect TQFTs. In the first example the authors argue that even the trivial TQFT leads to an interesting tricategory, seeing how the trivial defect functor adds relations producing a \(3\)-groupoid from the free pre-\(2\)-category \(\mathcal{F}_{d}^{p}\mathbb{K}^{\mathbb{D}}\)\ associated to given defect data \(\mathbb{D}\) (\S 4.1). In the second example the authors consider the Reshetikhin-Turaev theory [\textit{N. Reshetikhin} and \textit{V. G. Turaev}, Invent. Math. 103, No. 3, 547--597 (1991; Zbl 0725.57007)] associated to a modular tensor category \(\mathcal{C}\), which can be regarded as a special class of defect TQFTs \(\mathcal{Z}^{\mathcal{C}}\)\ with only \(1\)-dimensional defects (\S 4.2). In the last subsection (\S 4.3) the authors exploit the work on homotopy quantum field theory [\textit{V. Turaev} and \textit{A. Virelizier}, Int. J. Math. 23, No. 9, 1250094, 28 p. (2012; Zbl 1254.57012)] to construct two classes of \(3\)-dimensional defect TQFTs for every \(G\)-graded spherical fusion category, obtaining interesting surface defects which allows of extending Turaev-Viro theories to defect TQFTs. The authors conclude \S 1 (introduction and summary) by providing five topics for future study. \begin{itemize} \item[1.] It would be interesting to compare the approach in this paper with the correspondence between TQFTs and higher categories in [\textit{S. Morrison} and \textit{K. Walker}, Geom. Topol. 16, No. 3, 1481--1607 (2012; Zbl 1280.57026); Proc. Natl. Acad. Sci. USA 108, No. 20, 8139--8145 (2011; Zbl 1256.18007)]. \item[2.] The approach in this paper enables one to extend Reshetikhin-Turaev theory to produce not only invariants for \(3\)-dimensional manifolds with embedded ribbons but also with embedded surfaces. \item[3.] It was argued in [\textit{A. Kapustin} et al., Nucl. Phys., B 816, No. 3, 295--355 (2009; Zbl 1194.81224); \textit{A. Kapustin} and \textit{L. Rozansky}, Commun. Number Theory Phys. 4, No. 3, 563--549 (2010; Zbl 1220.81169)] that Rozansky-Witten theory gives rise to a \(3\)-category \(\mathcal{T}^{\mathrm{RW}}\)\ bringing together algebraic and symplectic geometry. It would be interesting to comprehend \(\mathcal{T}^{\mathrm{RW}}\)\ as a Gray category with duals. \item[4.] There is a theory of orbifold completion of pivotal \(2\)-categories [\textit{N. Carqueville} and \textit{I. Runkel}, Quantum Topol. 7, No. 2, 203--279 (2016; Zbl 1360.18007)] generalizing certain group actions on \(2\)-dimensional TQFTs via the algebraic language of defects and leading to new equivalences of categories [\textit{N. Carqueville} et al., J. Pure Appl. Algebra 220, No. 2, 759--781 (2016; Zbl 1333.18004); \textit{N. Carqueville} and \textit{A. Quintero Velez}, ``Calabi-Yau completions and orbifold equivalences'', Preprint, \url{arXiv:1509.00880}]. The idea of orbifold completion applies to TQFTs of any dimension \(n\)\ and should be developed for \(n=3\), which would bring new TQFTs. \item[5.] \(3\)-dimensional TQFTs have applications in quantum computing, where orbifoldable group actions have received much attention. The approach in this paper would offer a robust and conceptual framework for generalizations of these constructions. \end{itemize}