Cannon-Thurston maps for CAT(0) groups with isolated flats (Q2170985): Difference between revisions

If $M$ is a closed hyperbolic 3-manifold fibering over the circle and if $F\subset M$ is a fiber of the fibration, Cannon and Thurston proved in the 80s that the natural map from the hyperbolic plane to hyperbolic 3-space lifting the inclusion of $F$ into $M$ can be extended to a continuous map from the circle (the boundary of the hyperbolic plane) to the $2$-sphere (the boundary of hyperbolic 3-space), which is equivariant with respect to the action of the fundamental group of $F$. Mj later extended this result by considering a hyperbolic group $G$ containing an infinite normal hyperbolic subgroup $H$ and by proving that the inclusion of $H$ into $G$ gives rise to an $H$-equivariant continuous surjective map from the Gromov boundary of $H$ to the Gromov boundary of $G$. In this work, the authors consider a group $G$ acting geometrically on a $\mathrm{CAT}(0)$ space with isolated flats and an infinite normal subgroup $H$. Assuming either that $H$ is hyperbolic or that it acts itself geometrically on a $\mathrm{CAT}(0)$ space with isolated flats (and is of infinite index in the latter case), the authors prove that there is no Cannon-Thurston map in this context. The boundaries under consideration here are the visual boundary of the model space on which $G$ acts and either the Gromov boundary or the visual boundary of the model space for $H$ (depending on the assumption made). We observe that the existence of Cannon-Thurston maps in a non-hyperbolic context depends on which type of boundary is considered. Such maps sometime exist in a relatively hyperbolic context, when one picks the Bowditch boundaries. We refer to the article for more detailed statements. The authors also prove some structural results for hyperbolic normal subgroups in groups acting geometrically on $\mathrm{CAT}(0)$ spaces with isolated flats (building, among other things, on work of Rips and Sela).