Molino theory for matchbox manifolds (Q2397832)

A foliation is Riemannian when its holonomy pseudogroup consists of local isometries for some Riemannian metric on local transversals (``transverse rigidity''). These foliations were described by P.~Molino in terms of the more special class of transversely parallelizable (TP) foliations, where the holonomy transformations preserve a parallelism on local transversals. Many properties of Riemannian foliations were derived from this description. The submitted paper gives a version of the Molino's description for the case of an equicontinuous matchbox manifold \(\mathfrak M\). It is based on a previous result by A. Clark and S.~Hurder, stating that any such \(\mathfrak M\) is a weak solenoid \({\mathcal S}_{\mathcal P}\), defined as an inverse limit of a sequence of non-trivial covering maps between compact manifolds, \({\mathcal P}=\{\,p_{\ell+1}:M_{\ell+1}\to M_\ell\mid\ell\ge0\,\}\). Moreover \({\mathcal S}_{\mathcal P}\) is homogeneous if and only if \({\mathcal P}\) can be chosen to consist of regular coverings. The fundamental groups of the manifolds in \(\mathcal P\) define a chain of subgroups of finite index, \({\mathcal G}=\{G_0\supset G_1\supset\cdots\}\). Then the authors consider an associated chain of subgroups, \(C_0\supset C_1\supset\cdots\), where every \(C_\ell\) is the maximal normal subgroup of \(G_0\) contained in \(G_\ell\) (the core subgroup). This corresponds to a sequence of regular covering maps between compact manifolds, \(\widehat{\mathcal P}=\{\,\hat p_{\ell+1}:\widehat M_{\ell+1}\to\widehat M_\ell\mid\ell\ge0\,\}\), defining a homogeneous solenoid \(\widehat{\mathfrak M}={\mathcal S}_{\widehat{\mathcal P}}\). The canonical projections \(\widehat M_\ell\to M_\ell\) induce a foliated map \(\hat q:\widehat{\mathfrak M}\to\mathfrak M\). On the other hand, the quotient groups \(D_\ell=C_0/C_\ell\) form an induced sequence of surjective homomorphisms, \(\cdots\to D_2\to D_1\to D_0=\{e\}\), whose inverse limit \(\mathcal D\), the discriminant group, was previously studied by the authors. By identifying every quotient group \(C_0/C_\ell\) with the group of deck transformations of the regular covering \(\widehat M_\ell\to\widehat M_0=M_0\), they get an induced action of \(\mathcal D\) on \(\widehat{\mathfrak M}\) so that \(\widehat{\mathfrak M}/\mathcal D\equiv\mathfrak M\) via \(\hat q\). This defines the Molino sequence \({\mathcal D}\to\widehat{\mathfrak M}\overset{\hat q}{\to}{\mathfrak M}\). This sequence is unique if \({\mathfrak M}\) satisfies a certain strong quasianalyticity condition (SQA). Then the authors are led to studying the stability of \(\mathcal D\) and \(\widehat{\mathfrak M}\), introducing the concept of tame matchbox manifolds, and proving the following additional results. \(\mathfrak M\) is tame if \(\mathcal D\) is finite. \(\mathfrak M\) is SQA if the fundamental groups of all leaves are finitely generated. The closure of the holonomy pseudogroup is SQA if moreover \(\mathcal D\) is finite. The Molino sequence is non-trivial if the holonomy group of some leaf is non-trivial. The Schori solenoid is not SQA. Finally, the authors prove nice theorems about the realization of Cantor groups as discriminant groups of equicontinuous matchbox manifolds, using deep results of Lubotzky.