Anosov subgroups: dynamical and geometric characterizations (Q1695699)

Among discrete subgroups of rank one Lie groups there is an important class, namely the geometrically finite subgroups. By definition they admit a fundamental domain with finitely many sides. The simplest ones are those without parabolics. These can be characterized in a number of non-trivially equivalent ways, for instance as convex cocompact, which means that they admit a closed convex invariant cocompact subset in the symmetric space of the ambient Lie group.``In higher rank'', the authors say, ``a satisfying and sufficiently broad definition of geometric finiteness \dots remains yet to be found''. Convex cocompactness is much too restrictive. Only few subgroups are convex cocompact, by work of \textit{B. Kleiner} and the second author [Invent. Math. 163, No. 3, 657--676 (2006; Zbl 1097.53034)] and \textit{J. F. Quint} [Geom. Dedicata 113, 1--19 (2005; Zbl 1077.22016)]. Undistortedness, on the other hand is too weak. The authors claim that unditorted subgroups may fail to be finitely presented. However, some of the other characterizations of geometric finiteness in rank one do admit useful modifications in higher rank. This leads to the class of Anosov subgroups, introduced by \textit{F. Labourie} [Invent. Math. 165, No. 1, 51--114 (2006; Zbl 1103.32007)] and further extended by \textit{O. Guichard} and \textit{A. Wienhard} [ibid. 190, No. 2, 357--438 (2012; Zbl 1270.20049)]. The goal of the paper under review is to extend the existing work in the rank one case to the class of Anosov groups in higher rank, to simplify the existing definitions of Anosov groups and to lay the foundations for ongoing and future work on this class of groups. The paper is long and hard to review. The following rough description of its contents is therefore necessarily sketchy. The groups that the authors consider are (mostly) \(\tau_{\mathrm{mod}}\)-regular, which states a certain contraction behavior on a suitable flag manifold \(\mathrm{Flag}_{\tau_{\mathrm{mod}}}\). The notions considered are (i) asymptotic embededness, (ii) expansivity, (iii) conicality and (iv) Morse property. The first one states that the orbit map extends to an equivariant homeomorphism of the Gromov compactification to the compactification obtained by adding \(\mathrm{Flag}_{\tau_{\mathrm{mod}}}\). The second states an expanding condition on the limit set \(\Lambda \subset\mathrm{Flag}_{\tau_{\mathrm{mod}}}\). These are dynamical conditions. The next one, (iii) is a geometric condition. Conicality is concerned with the asymptotic behavior of orbits. Finally, (iv) is concerned with coarse extrinsic geometry. According to the classical Morse Lemma, in rank one symmetric spaces quasigeodesic segments are uniformly Hausdorff close to geodesic segments with the same endpoints. This is no longer true in higher rank, since it fails already in the Euclidean plane. So the Morse property (iv) states a strengthening of undistortedness by imposing a Morse type property where geodesic segments are replaced by diamonds. The authors refer the reader to their survey [the authors, Transform. Groups 21, No. 4, 1105--1121 (2016; Zbl 1375.37131)] and to [the first author and the second author, ``Discrete isometry grups of symmetric spaces'', Preprint, \url{arXiv:1703.02160}] for background discussions and many examples.