Floating bodies and approximation of convex bodies by polytopes (Q2135725)

This is a survey on convex floating bodies, their variants such as surface bodies and weighted floating bodies, and their applications. Among the latter are representations of the affine surface area and \(L_p\) affine surface areas by limit relations for volumes, which allow to extend these functionals to convex bodies without smoothness assumptions. Several applications are possible to approximation of convex bodies by polytopes, provided that the degree of approximation is measured by the volume difference. Among the presented results, most of which are due to the author and Carsten Schütt, we mention the following one. Let \(K\subset{\mathbb R}^n\) be a convex body in which a ball rolls freely and which rolls freely in a ball. Let \({\mathbb P}_f\) be a probability measure on \(\partial K\) with a positive continuous density \(f\) with respect to \(\mu_{\partial K}\), the surface measure of \(K\). Let \({\mathbb E}(f,N)\) be the expected volume of the convex hull of \(n\) stochastically independent random points with distribution \({\mathbb P}_f\). Then \[ \lim_{N\to\infty} \frac{\mathrm{vol}_n(K)-{\mathbb E}(f,N)}{\left(\frac{1}{N}\right)^{\frac{2}{n-1}}}= c_n\int_{\partial K} \left(\frac{{\kappa}(x)}{f(x)^{2}}\right)^{\frac{1}{n-1}}d\mu_{\partial K}(x),\] where \(\kappa\) is the generalized Gauß-Kronecker curvature and \(c_n\) is an explicit constant. The integral becomes minimal if \(f\) is the density of the normalized affine surface area measure. For complementary reading about best and random approximation of convex bodies by polytopes, the author suggests the survey by \textit{J. Prochno} et al. [J. Complexity 71, Article ID 101652, 19 p. (2022; Zbl 1489.52009)].