A Baum-Connes conjecture for singular foliations (Q2286560)

A regular foliation decomposes a manifold into leaves in a locally trivial way. Various geometric situations produce foliations with some singularities. Typical examples are the foliation into orbits for a smooth Lie group action or the foliation into symplectic leaves of a Poisson manifold. The authors have associated \(C^*\)-algebras to singular foliations and extended index theory to this setting. This article studies the \(K\)-theory of these \(C^*\)-algebras. The \(C^*\)-algebra of a singular foliation is associated to a groupoid, called its holonomy groupoid. In the regular case, this is a Lie groupoid. The \(K\)theory of Lie groupoids is studied through the Baum-Connes assembly map. The main issue for a singular foliation is that the dimensions of the leaves may jump. There is, however, a filtration of the manifold by open subsets \(Y_i\) so that the restriction of the holonomy groupoid to each stratum \(Y_i\setminus Y_{i-1}\) is a Lie groupoid. This filtration induces a filtration on the \(C^*\)-algebra of the holonomy groupoid whose subquotients are \(C^*\)-algebras of Lie groupoids. The idea of the paper is to put the Baum-Connes assembly maps for these subquotients together to form an analogue of the Baum-Connes assembly map for the singular case. This requires some extra assumptions, however, in order to combine the topological \(K\)theory groups for the strata in a nice way. These regularity assumptions are satisfied in many cases. The authors discuss many examples in great detail to show how to completely compute the \(K\)theory groups. And they develop several techniques that are also useful for other purposes. The different strata are glued using a rather general telescope construction, which generalises the mapping telescope for an inductive system. In the simplest case of a single homomorphism -- then the authors speak of singularity height \(1\) -- the telescope is the mapping cone of a homomorphism. So the Baum-Connes map in this case becomes a relative Baum-Connes map for a homomorphism between two Lie groupoids. Such relative constructions have recently been studied for group \(C^*\)-algebras by several authors (see, for instance, [\textit{R. J. Deeley} and \textit{M. Goffeng}, J. Topol. 11, No. 4, 967--1001 (2018; Zbl 1415.58015)]). The authors also need to develop equivariant KK-theory in a way that covers the groupoids of singular foliations. They formulate the Kasparov technical theorem and similar basic results in such a way that they cover this new case, and also other equivariant KK-theories constructed previously. The arrow space of the holonomy groupoid of a singular foliation is not a manifold. There are, however, natural smooth structures on the objects and the source fibres, and the arrow space is a quotient of a smooth manifold by a certain equivalence relation. These properties are formalised in the definition of a longitudinally smooth groupoid. And equivariant KK-theory is defined in this generality. It seems possible to replace ``smooth'' by ``locally compact, Hausdorff'' everywhere in these definitions as far as the mere definition of equivariant KK-theory is concerned. This gives an equivariant KK-theory for actions of non-Hausdorff, locally compact groupoids, including a Kasparov product. Examples in [\textit{A. Buss} and \textit{R. Meyer}, Rocky Mt. J. Math. 47, No. 1, 53--159 (2017; Zbl 1404.46058)] show, however, that is is important in this generality to allow twisted actions or even more general actions because the stabilisation trick that allows to remove twists fails for non-Hausdorff groupoids. The construction of the Baum-Connes assembly map for longitudinally smooth groupoid needs another assumption besides smoothness. Namely, the authors assume in the body of the text that the universal proper actions of the Lie groupoids that occur can be realised by actions on smooth manifolds. They discuss in an appendix how this assumption may be weakened.