Factorization homology of stratified spaces (Q508425)

This paper proposes to understand locality in topological quantum field theory by a fusion of the algebra of factorization homology and the geometry of stratifications. The authors first review the subject of factorization homology. From the introduction: ``Factorization homology is, heuristically, a procedure which takes an \(n\)-manifold \(M\) and an algebraic input \(A\), such as an \(\mathcal E_n\)-algebra, and produces an object \(\int_M A\). The manifold provides the gluing data, the algebra provides the gluing rules, and one can think of integrating the multiplication of the algebra over the gluing data of the manifold. Viewing this object as an invariant of the manifold, the procedure generalizes usual homology theories; viewing the object as an invariant of the algebra, it generalizes Hochschild homology when the manifold is the circle and offers a natural repository for traces or index-type invariants. Such a procedure was introduced for algebraic varieties by Beilinson and Drinfeld in their work on an algebro-geometric formalism for conformal field theory. Lurie defined a topological analogue of their construction known as factorization homology or topological chiral homology and this topological construction likewise generalizes the labeled configuration spaces of Salvatore and Segal. The main theorem of this area, nonabelian Poincaré duality, naturally generalizes the James construction and configuration space models of mapping spaces dating to the 1970s in work of McDuff. In mathematical approaches to topological field theory at least since Atiyah, it is common to organize the formalism around the functoriality of the state spaces in the theory. This choice leads to cobordisms and, proceeding deeper, higher categories of cobordisms after Baez and Dolan. In contrast, Costello and Gwilliam, following the factorization algebra structures, codify their theory around the structure of observables, or operators, rather than state spaces. In their work, the earlier renormalization machinery is married with the factorization point of view; the intuitive factorization homology procedure becomes a way of constructing a candidate object \(\int_M A\) of global observables on a space-time \(M\) from the algebra of observables \(A\) on Euclidean space. This candidate object \(\int_M A\) is intended to accurately capture the global observables on \(M\) if the quantum field theory is perturbative. They prove a quantization theorem, designed as a mathematical formulation of physicists' methods of perturbative renormalization, using the Batalin-Vilkovisky formalism applied to derived symplectic geometry. From this quantization theorem, they construct a number of interesting examples of such perturbative field theories. The structural favoring of observables over state spaces has geometric consequences, namely a favoring of open embeddings over cobordisms. Dual to restricting fields, observables can extend by zero. This is unlike state spaces, where there is no procedure for extension by zero, no naturality with respect to open embeddings, and no values for noncompact manifolds. The cobordism hypothesis with singularities, after Baez-Dolan and Lurie, gives a proposed classification for certain topological field theories in terms of their state spaces. With this in mind, one can ask whether there is a similar classification which applies to this class of topological field theories constructed in terms of the Batalin-Vilkovisky formalism, in terms of their observables. Our first main result, Theorem 2.43, can be viewed as just such a classification. In the present work, we lay the foundations for a general theory of factorization homology, following the outline of [\textit{D. Ayala} and \textit{J. Francis}, J. Topol. 8, No. 4, 1045--1084 (2015; Zbl 1350.55009)]. We do this for stratified spaces and, more generally, \(\mathcal B\)-manifolds, where \(\mathcal B\) is a collection of basic singularity types endowed with a tangential structure, applying the theory of stratified spaces and tangential structures developed in \textit{D. Ayala} et al. [Adv. Math. 307, 903--1028 (2017; Zbl 1367.57015)]. This extra level of generality is carried out for two reasons. First, the theory without stratifications is related to observables on space-time in perturbative field theory; adding nontrivial stratifications allows one to incorporate boundary conditions and defects in this theory, such as Wilson line operators in Chern-Simons. The second reason concerns the extension of our theory outside of the perturbative range in quantum field theories.''