A stratified homotopy hypothesis (Q1737976)

Grothendieck's homotopy hypothesis states, informally, that a space is the same thing as an \(\infty\)-groupoid. Given a space, the objects of the associated \(\infty\)-groupoid are its points; morphisms are paths; higher homotopies are continuous maps of higher simplices into the space. More precisely, Grothendieck hypothesized that any appropriate model of higher categories should yield an equivalence between the homotopy theory of topological spaces, and the homotopy theory of higher groupoids (higher categories whose morphisms are all invertible). Using standard homotopy theory and approximating reasonable spaces using manifolds, one can reformulate Grothendieck's homotopy hypothesis using manifolds (rather than spaces). A key feature of the hypothesis is that the space of maps between manifolds (or spaces) is homotopy equivalent to the space of functors between the associated \(\infty\)-groupoids. The hypothesis is, typically, easily verified in models of \((\infty,1)\)-categories. In the present review, we will have in mind the model of \(\infty\)-categories (otherwise known as quasi-categories or weak Kan complexes). The present work generalizes the homotopy hypothesis to (i) characterize \(\infty\)-categories (rather than groupoids), and (ii) incorporate stratified spaces, which, roughly speaking, are generalizations of manifolds to well-behaved spaces with a notion of smooth structure that allows one to decompose the space into strata consisting of smooth manifolds. (The formulation utilized in the present work was developed in [\textit{D. Ayala} et al., Adv. Math. 307, 903--1028 (2017; Zbl 1367.57015)].) The key first step is to show that a well-known construction (the exit path category of a stratified space) is a fully faithful functor from the \(\infty\)-category of stratified spaces to the \(\infty\)-category of \(\infty\)-categories. As a result, any \(\infty\)-category \(\mathcal{C}\) is a presheaf on the category of stratified spaces (sending a stratified space to the space of functors from its exit path category to \(\mathcal{C}\)). It is at this point that one could hope for a ``geometric'' characterization of all \(\infty\)-categories, by understanding how \(\infty\)-categories play with stratified spaces -- and the authors achieve this in their first main theorem of the work: An \(\infty\)-category is the same thing as a certain class of sheaves on the site of stratified spaces, which the authors call striation sheaves. (More precisely, there is an equivalence of \(\infty\)-categories between the \(\infty\)-category of striation sheaves and the \(\infty\)-category of \(\infty\)-categories.) It has long been observed that sheaves on stratified spaces can encode intricate homotopical data, and in ways that are not easily obtained by methods that evade sheaf theory. The authors illustrate this principle here by exhibiting an example of a striation sheaf that is not easily articulable using other models of \((\infty,1)\)-categories: \(\mathcal{B}\mathsf{un}\). This is a classifying \(\infty\)-category for constructible bundles, in that a functor \(\mathsf{Exit}(K) \to \mathcal{B}\mathsf{un}\) encodes the data of a constructible bundle on a stratified space \(K\). The second main result of the present work justifies the term ``representing \(\infty\)-category;'' indeed, \(\mathcal{B}\mathsf{un}\) is most naturally constructed as a fibration over (i.e., a presheaf on) the \(\infty\)-category of stratified spaces, and the second main result is that this presheaf is actually a striation sheaf. The authors view a chief application of this result to be the construction of an \(\infty\)-category of vari-framed compact manifolds -- informally, a vari-framed compact manifold can encode the shape of a diagram in an \((\infty,n)\)-category, and allows one to ``integrate'' another \((\infty,n)\)-category \(\mathcal{C}\) over the vari-framed manifold to obtain the space of all diagrams of that shape in \(\mathcal{C}\). In other words, vari-framed manifolds are a natural domain for factorization homology of higher categories. Ayala and Francis propose to give an alternate proof of the cobordism hypothesis using this framework; this is summarized briefly in the introduction of the work. Of interest is the characterization of transversality sheaves in Section 5. Transversality sheaves are practical ways of producing striation sheaves (and hence \(\infty\)-categories, by the main theorem), and seem applicable in any situation where one wants an \(\infty\)-category of manifold-theoretic objects that are parametrizable using the language of constructible bundles. Indeed, the proof that \(\mathcal{B}\mathsf{un}\) is an \(\infty\)-category involves showing that the associated fibration defines a transversality sheaf (Theorem 6.3.3). Moreover, many \(\infty\)-categories of manifolds and their embeddings can fit into such a framework, as the mapping cylinder construction (and its iterates) produces constructible bundles whose fibers are the domain and codomain of embeddings. This is a highly useful construction, as many differential-geometric constructions are easier to articulate using bundles, rather than by trying to construct smooth families of structures that depend smoothly on infinite-dimensional embedding spaces. (This is, philosophically, the same reason that fibrations of \(\infty\)-categories are easier to produce than functors of \(\infty\)-categories; though the two are formally equivalent via the Grothendieck, or un/straightening, construction.)