Potential theory and geometry of the farthest distance function (Q380030): Difference between revisions

The farthest distance function \(d_{E}\) of a compact set \(E\subset \mathbb{R}^{N}\) is simply the distance to the farthest point of \(E\). In two dimensions \(\log d_{E}\) is the logarithmic potential of a probability measure \(\sigma_{E}\), and this fact has been useful in the study of supremum norms of polynomials. \textit{R. S. Laugesen} and \textit{I. E. Pritsker} [Can. Math. Bull. 46, No. 3, 373--387 (2003; Zbl 1052.31002)] conjectured that \(\sigma _{E}(E)\leq 1/2\) when \(E\) contains more than one point, and this was subsequently proved by the reviewer and \textit{I. Netuka} [J. Anal. Math. 101, 163--177 (2007; Zbl 1134.31001)], along with a description of the extremal cases. In higher dimensions \(d_{E}^{2-N}\) is the Newtonian potential of a probability measure. In this case Laugesen and Pritsker have conjectured that at most \(2^{1-N}\) of the measure can live on \(E\), and that this bound is attained only when \(E\) is a ball. This conjecture remains open. In the present paper it is shown that, when \(N=2\) and \(E\) is a convex polygon, \(\sigma _{E}\) can be studied in terms of angles subtended by certain sets at the vertices. This geometric approach is then used to investigate the higher dimensional conjecture in the particular case of convex polytopes \(E\), and the author provides the first substantial numerical evidence in its favour.