Surface area deviation between smooth convex bodies and polytopes (Q2040993)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Surface area deviation between smooth convex bodies and polytopes |
scientific article |
Statements
Surface area deviation between smooth convex bodies and polytopes (English)
0 references
15 July 2021
0 references
This paper is devoted to the study of the deviation between a convex body with twice differentiable boundary and a polytope with a given number of vertices, and no constraint on its position, by means of the surface area deviation. A convex body \(K\) in \(\mathbb{R}^n\), i.e., a compact convex subset of \(\mathbb{R}^n\) with non-empty interior, is of class \(\mathcal{C}^2_+\) if its boundary \(\partial K\) is a twice differentiable \((n-1)\)- dimensional submanifold of \(\mathbb{R}^n\) with strictly positive Gauss curvature \(\kappa(x)\) at every \(x\in\partial K\). The surface area deviation between two convex bodies \(K,L\subset \mathbb{R}^n\) is defined as \[ \Delta_s(K,L):=\mathcal{H}^{n-1}\left(\partial(K\cup L)\right)-\mathcal{H}^{n-1}\left(\partial(K\cap L)\right), \] where \(\mathcal{H}^{n-1}\) denotes the \((n-1)\)-dimensional Hausdorff measure. The main result in this paper establishes an upper bound for the surface area deviation between a convex body of class \(\mathcal{C}^2_+\) and a polytope with a sufficiently large given number of vertices in terms of the \(n\)-affine surface area of the convex body. For \(p\in [-\infty,\infty]\setminus\{-n\}\) the \(p\)-affine surface area of a convex body \(K\) having the origin in its interior is given by \[ \operatorname{as}_p(K):=\int_{\partial K}\frac{\kappa(x)^{\frac{p}{n+p}}}{\langle x, N(x)\rangle^{\frac{n(p-1)}{n+p}}}\mathcal{H}^{n-1}(dx), \] where \(N(x)\) denotes the unique outer normal vector at \(x\in\partial K\), and \(\langle\cdot,\cdot\rangle\) the Euclidean scalar product. The central result of the paper is the following. \textbf{Theorem.} Let \(K\) be a convex body in \(\mathbb{R}^n\), \(n\geq 2\), that is of class \(C^2_+\). Then, there exists a number \(N_K\in\mathbb{N}\) depending only on \(K\) such that for all \(N\geq N_K\) there exists a polytope \(P\subset \mathbb{R}^n\) with exactly \(N\) vertices such that \[ \Delta_s(K,P) \leq a \,n\,N^{-\frac{2}{n-1}}\,\operatorname{as}_n(K)^\frac{2}{n-1}\mathcal{H}^{n-1}(\partial K), \] where \(a\in(0,\infty)\) is an absolute constant. The proof is probabilistic, based on the construction of a random polytope, some precise control of the expected surface area of the convex hull of an appropriate number of random points on the boundary of a convex body, and certain convex and integral geometric tools, as a Blaschke-Petkantschin formula. In [\textit{S. D. Hoehner} et al., J. Theor. Probab. 31, No. 1, 244--267 (2018; Zbl 1390.52010)], the surface area deviation of the Euclidean unit ball \(\mathbb{B}^n\) and a polytope was examined, showing that for sufficiently large \(N\), there is a polytope \(P\) with \(N\) vertices so that \[ \Delta_s(\mathbb{B}^n,P) \leq a \,N^{-\frac{2}{n-1}}\,\mathcal{H}^{n-1}(\partial \mathbb{B}^n), \] with an absolute constant \(a>0\). Other deviations, as the volume deviation, and, more generally, the intrinsic volume deviations can be introduced, similarly to the surface area deviation. For the volume deviation \(\Delta_v\), a particular case of a result in [the first and the third author, Electron. J. Probab. 23, Paper No. 9, 21 p. (2018; Zbl 1395.52008)], provides the existence of an \(n\)-dimensional polytope \(P\) with a sufficiently large number of vertices \(N\), satisfying \[ \Delta_v(K,P)\leq a\, N^{-\frac{2}{n-1}}\,\operatorname{as}_1(K)^\frac{n+1}{n-1}. \] According to the authors, similar ideas as those used in the proof of the main result generalize to the intrinsic volume deviations.
0 references
approximation of convex bodies
0 references
\(p\)-affine surface areas
0 references
polytopes
0 references
probabilistic method
0 references
random polytopes
0 references
surface area deviation
0 references
0 references