A discrete Blaschke theorem for convex polygons in 2-dimensional space forms (Q6568992)

A \(n\)-dimensional space form \(M^n_\lambda\) of curvature \(\lambda\) is a complete simply connected n-dimensional Riemannian manifold of constant sectional curvature \(\lambda\). For \(\lambda = 0\) this is the Euclidean space \(\mathbb{R}^n\), for \(\lambda> 0\) the space form is an \(n\)-dimensional sphere of radius \(\frac{1}{\sqrt{\lambda}} \subset \mathbb{R}^{n+1}\), and for \(\lambda < 0\) the space form can be visualized as the upper connected component of the Minkowski sphere of radius \(\frac{1}{\sqrt{|\lambda|}}\) in the Minkowski space \(\mathbb{R}_1^{n+1}\).\N\NWilhelm Blaschke proved in 1949: Let \(\Gamma \subset \mathbb{R}^{2}\) be closed convex regular curve in the Euclidean plane that bounds a compact convex region \(\Omega\). If the curvature \(\kappa\) of \(\Gamma\) is bounded from below by some constant \(\kappa_0 > 0\), then, for every point \(p \in \Gamma\), the circle tangent to \(\Gamma\) at \(p\) bounds a disk that contains \(\Omega\). The radius of the tangent circle is \(R = \frac{1}{\kappa_0}\) and the unit normal vector that points to its center of the circle points also to the interior of \(\Omega\).\N\NThe authors define and justify for a polygon \(P \subset M^2_\lambda\) the curvature \(\kappa_A\) of the polygon \(P\) at the vertex \(A\) using the angle \(\alpha_i\) at \(A\) together with the lengths \(\ell_1\), \(\ell_2\) of the sides of \(P\) that meet \(A\) by\N\[\N\kappa_A = \frac{\pi-\alpha}{ta_\lambda(\ell_1/2) + ta_\lambda(\ell_2/2)} \, \textrm{ where } ta_\lambda(x) := \left\{ \begin{array}{ll} \tan x, & \lambda > 0, \\\Nx, & \lambda = 0, \\\N\tanh x, & \lambda < 0. \end{array} \right.\N\]\NUsing this definition, the authors show, that if at all vertices of a convex polygon \(P \subset M^2_\lambda\) the curvature is bounded by \(\kappa_A \geq \kappa_0 > 0\) then the (general) circumradius \(R\) of \(P\) satisfies \(ta_\lambda(R) \leq \frac{\pi}{2 \kappa_0}\) and the equality holds if and only if the polygon is a doubly covered segment.\N\NFurthermore the authors find bounds (depending on \(\lambda\)) on the length of the edges \(\ell_i\) and the radius \(R\) with respect to \(\kappa_0\).