The geodesic flow of a nonpositively curved graph manifold. (Q1865768)

This paper concerns with stability phenomena of geodesic flows. In this context, there are known results on the stability under metric changes when the manifold is negatively curved, for example: homotopy equivalences of negatively curved manifolds produce homeomorphisms between the 1-dimensional foliations defined by their geodesic flows [\textit{M. Gromov}, Enseign. Math., II. Sér. 46, 391--402 (2000; Zbl 1002.53028)]. Furthermore, when two compact negatively curved manifolds \(M_1\), \(M_2\) have the same fundamental group \(G\), then \(G\) acts isometrically on the universal coverings \(M_1'\), \(M_2'\) and the induced actions of \(G\) on the boundaries \(\partial_\infty M_1'\), \(\partial_\infty M_2'\) are \(G\)-equivariantly homeomorphic. In this paper, the authors relax the assumption on the curvature, considering the case of nonpositive curvature. There is a clear link to the following question posed by \textit{M. Gromov} [Geometric Group Theory, Vol. 2, Sussex, (1991; Zbl 0841.20039)]. Do two cocompact actions of a deck group \(G\) on Hadamard manifolds \(X_1\), \(X_2\) induce \(G\)-equivariantly homeomorphic boundary actions of \(G\) on \(\partial_\infty X_i\)? The question has negative answers due to \textit{S. V. Buyalo} [St. Petersbg. Math. J. 10, No.2, 293--313 (1999; Zbl 0918.53018)] and to the authors [Topology 39, 549--556 (2000; Zbl 0959.53014)]. In the paper under review, the authors consider locally compact Hadamard spaces \(X\), or \(\text{CAT}(0)\)-spaces, i.e. complete, simply connected length spaces with nonpositive curvature in the sense of Alexandrov. The boundary \(\partial_\infty X\) is the collection of equivalent classes of geodesic rays equipped with the ``cone'' topology. The group acting on \(X\) belongs to a special class of groups, called admissible, which generalize fundamental groups of 3-dimensional graph manifolds. They firstly consider the case of nonpositively curved 3-dimensional graph manifolds. They develop also new tools for studying geodesic flows of nonpositively curved spaces and a kind of ``coding'' for geodesic rays in a Hadamard space \(X\) which allows to understand the boundary action of an admissible group \(G\) on \(X\) and the Tits metric on \(\partial_\infty X\). Then, in the general case, the main result is the following. Assume that an admissible group \(G\) acts discretely and cocompactly on a Hadamard space \(X\). The authors associate geometric data with any vertex group \(G_v\), i.e. a class function \(MLS_v: G_v\to\mathbb{R}_+\) and a homomorphism \(\tau_v: G_v\to \mathbb{R}\). If the action is isometric, the geometric data is determined, up to scale factors, by the topological conjugacy class of the boundary action of \(G\) on \(\partial_\infty X\) and vice versa. Given another such action of \(G\) on \(X\), then the actions have the same (up to scale) geometric data if and only if \(G\)-equivariant quasi isometries \(f: X\to X'\) canonically extend to the compactifications \(X\cup \partial_\infty X\), \(X'\cup \partial_\infty X'\). The equality, up to scale, of the geometric data is also equivalent to the existence of a unique \(G\)-equivariant homeomorphism between the two boundary actions and to the following property: for any \(G\)-equivariant quasi isometry \(f\) there exists a function \(\vartheta: \mathbb{R}_+\to \mathbb{R}_+\), with \(\lim_{r\to \infty} \vartheta(r)= 0\), such that for every unit speed geodesic ray \(\gamma\) in \(X\) there is a geodesic ray \(\gamma'\) in \(X'\) with \(d(f(\gamma(t)),\gamma'(t))< (1+ t)\vartheta(t)\). Finally, they outline some open problems.