Isometries of two dimensional Hilbert geometries (Q330033)

The \textit{Hilbert distance} between two points \(X\) and \(Y\) in a bounded convex domain \(\Omega\) of \(\mathbb R^n\) is defined as \[ d(X,Y):=\ln((X,Y;\bar X, \bar Y)):=\ln\left(\frac{|\bar Y-X|}{|\bar Y-Y|}:\frac{|\bar X-X|}{|\bar X-Y|}\right) \] where \(|U-V|\) denotes the usual Euclidean distance between two points \(U\) and \(V\) in \(\mathbb R^n\), and where \(X\) and \(Y\) are the points of intersection of \(\overleftrightarrow{XY}\) with the boundary of \(\Omega\). A full description of all isometries in Hilbert geometry is given by \textit{C. Walsh} in [``Gauge-reversing maps on cones, and Hilbert and Thompson isometries'', Preprint, \url{arXiv:1312.7871}]. In the paper under review, the authors give a short proof of the following result: Let \(\Omega_1\) and \(\Omega_2\) be two bounded convex domains in the plane \(\mathbb R^2\), and let \(d_1\) and \(d_2\) be the corresponding Hilbert metrics. If \(\Omega_1\) is not the interior of a triangle, then every isometry \(f:(\Omega_1,d_1)\to (\Omega_2,d_2)\) is the restriction of a projective transformation of \(\mathbb{RP}^2\). Even though the above result is a special case of the results in [loc. cit.], the proof given in the current paper uses methods that are different from those employed in [loc. cit.]. In particular, the proof uses the description of metric geodesics in Hilbert geometry, together with a result of \textit{W. Prenowitz} from line geometry [Trans. Am. Math. Soc. 38, 564--599 (1935; Zbl 0012.41301)].