A spiky ball

From MaRDI portal



Abstract: The Illumination Problem may be phrased as the problem of covering a convex body in Euclidean n-space by a minimum number of translates of its interior. By a probabilistic argument, we show that, arbitrarily close to the Euclidean ball, there is a centrally symmetric convex body of illumination number exponentially large in the dimension.


Let \(K \subseteq \mathbb{R}^n\) be a convex body with non-empty interior, and let \(i(K)\) be the smallest number of translates of the interior of \(K\) that cover \(K\) (or, equivalently, the smallest number of directions needed for illuminating the whole boundary of \(K\)). It is known that \(i(K) \geq n+1\), with equality e.g.\ if \(K=B^n\) is the Euclidean ball, and conjectured that \(i(K) \leq 2^n\), with equality e.g.\ if \(K=[-1,1]^n\) is a cube; see Chapter VI in [\textit{V. Boltyanski} et al., Excursions into combinatorial geometry. Berlin: Springer (1997; Zbl 0877.52001)].NEWLINENEWLINEThe paper under review shows by a probabilistic argument that there are bodies close to \(B^n\) whose covering numbers grow exponentially in the dimension: given \(1<D <1.116\) and \(n\) sufficiently large, there is a centrally symmetric convex body \(K \subseteq \mathbb{R}^n\) such that \(\frac{1}{D} B^n \subseteq K \subseteq B^n\) and \(i(K) \geq \frac{1}{20}D^n\).











This page was built for publication: A spiky ball

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2810744)