Steiner symmetrization using a finite set of directions (Q408120): Difference between revisions

Despite being one of the simplest old tools in convexity, the Steiner symmetrization does not fail to still surprise us. The Steiner symmetrization has been recently the object of several research papers, one of which is under review here. Essentially, one is interested in the convergence of a Steiner process \(S_{u_1} \cdots S_{u_i}K\), \(i \to \infty\), where \(K\) is a convex body (although one can ask \(K\) to be just a compact set in \(\mathbb{R}^n\)) and the \(u_i\)'s are directions of the unit sphere \(\mathbb{S}^{n-1}\). The problem is not trivial, for example in [\textit{G. Bianchi, D. A. Klain, E. Lutwak, D. Yang} and \textit{G. Zhang}, Adv. Appl. Math. 47, No. 4, 869--873 (2011; Zbl 1230.52007)] it is shown that a countable set of directions, even if dense in \(\mathbb{S}^{n-1}\), is not sufficient for the Steiner process to converge. The present paper considers a finite set of directions \(\{u_1, \dots, u_m\}\) in which the symmetrization is repeated, however, infinitely many times, whether via a periodic pattern, or by a random selection. The main result states that, in this case, the Steiner process always converges to a convex body which is symmetric under reflection in any of the directions \(u_i\) that appear infinitely often in the sequence. The author maintains a good control over the Steiner process through the \textit{layering function} of the convex body. The paper is nicely written and ends with several open problems.