A rigidity property of some negatively curved solvable Lie groups (Q1760052)

The paper under review proves quasi-isometric rigidity for a class of negatively curved, simply connected, solvable Lie groups, namely for semidirect products \(G_A:=\mathbb R^n\rtimes_{e^{tA}}\mathbb R\) with diagonal matrices \(A\in GL(n,{\mathbb R})\), whose eigenvalues \(\alpha_1,\ldots,\alpha_r\) are assumed to be positive and of multiplicities \(n_1,\ldots,n_r\) strictly smaller than \(n\). It is proved that every quasi-isometry of \(G_A\) is an almost-isometry, i.e., the multiplicative factor in the definition of quasi-isometry can be taken to be \(1\). (Remarkably this is not true for \(n_1=n\). Indeed, for \(A=\alpha \mathbb I\) one has \(G_A=\mathbb H^n\), where quasi-isometries need not be almost-isometries.) The approach to quasi-isometries is (as usual) to study the induced maps on the ideal boundary \(\partial G_A\), which are quasi-symmetries with respect to the visual Bourdon metric. All vertical geodesics \(t\rightarrow (x,t)\in \mathbb R^n\rtimes_{e^{tA}}\mathbb R\) tend to the same \(\xi_0\in\partial G_A\) and the complement \(\partial G_A\setminus\left\{\xi_0\right\}\) can be identified with \(\mathbb R^n=\mathbb R^{n_1}\times\ldots\times\mathbb R^{n_r}\). The authors consider the parabolic visual metric \(D(x,y):=\max\left\{\mid x_1-y_1\mid,\mid x_2-y_2\mid^{\frac{\alpha_1}{\alpha_2}},\ldots,\mid x_r-y_r\mid^{\frac{\alpha_1}{\alpha_r}}\right\}\) on \(\partial G_A\setminus\left\{\xi_0\right\}=\mathbb R^n\), it is bilipschitz equivalent to the Hamenstädt metric. They reprove a result of Kleiner and Tyson that any quasi-symmetry of \((\mathbb R^n,D)\) preserves the horizontal foliation by the \(\mathbb R^{n_1}\times\left\{*\right\}\). The main technical result of the paper is then that every quasi-symmetry of \(\partial G_A\) preserves \(\xi_0\) and is bilipschitz on \((\partial G_A\setminus\left\{\xi_0\right\},D)\), i.e., for the parabolic visual metric. An immediate corollary is a result of Kleiner that \(G_A\) is not quasi-isometric to any finitely generated group. Indeed such a group would be Gromov hyperbolic and have a global fixed point for its induced action on its ideal boundary, easily yielding a contradiction. The quasi-isometric rigidity follows from the main technical result together with a (more generally proven) lemma which in particular implies that (under certain assumptions) a quasi-isometry of CAT(-1)-spaces is an almost-isometry if and only if the boundary map is bilipschitz with respect to the Hamenstädt metrics.