Area of ideal triangles and Gromov hyperbolicity in Hilbert geometry (Q1022577): Difference between revisions

The authors show, in the context of Hilbert geometry, the equivalence between the existence of an upper bound on the area of ideal triangles and the Gromov-hyperbolicity. A Hilbert geometry \((C,d_C)\) is a nonempty bounded open convex set \(C\) in \(\mathbb{R}^n\) endowed with a Hilbert distance \(d_C\). This metric space induces a Finsler metric \(F_C\) such that the Hilbert distance \(d_C\) coincides with the length distance of the Finslerian norm \(F_C\). This leads to the concrete construction of a Borel measure on \(C\), which is in fact the Hausdorff measure of the metric space \((C,d_C)\). Using this setting the authors are able to give an intrinsic condition equivalent to the Gromov-hyperbolicity in terms of the area of the ideal triangles of \((C,d_C)\). An ideal triangle \(T\subset C\) is the affine convex hull of three points \(a,b,c\in\partial C\) not on a line, and such that \(T\cap\partial C=a\cup b\cup c\), and its area is the area for the Hilbert measure. The main result of the paper is Theorem 1: A Hilbert geometry is hyperbolic in the sense of Gromov if and only if there is a bound on the area of ideal triangles. The paper is very interesting not only for the result itself (Theorem 1), but also for showing that Finsler metrics can appear in a very natural way in the geometry of metric spaces, and that they can be very useful tools. It is possible that a similar approach for other types of metric spaces could lead to interesting results.