A note on the sum of distances under a diameter constraint (Q5946716): Difference between revisions

Consider the (euclidean) plane \(E\) with the usual distance function \(d(x,y)\). Let \(\{x_1,\dots,x_n\}\) be a set of \(n\) points in \(E\) with \(d(x_i,x_j) \leq 1\) for all choices of \(i,j\). An interesting question is: What is the maximum possible value \(\sigma_n\) of \((1/n^2) [\sum_{i,j} d(x_i,x_j)]\), where the sum ranges over all possible pairs \(i,j\). The author gives an upper bound for \(\sigma_n\). (Note that \(\sigma_3= 2/3.)\) Let \(X\) be a compact connected subset of \(E\) with diameter 1. A result of \textit{O. Gross} [Ann. Math. Stud. 52, 49-53 (1964; Zbl 0126.16401)] says there is a unique number \(r (X)\) with the property: For every positive integer \(n\) and for all \(x_1,x_2, \dots, x_n\in X\) there exists a \(y\in X\) such that \((1/n^2) [\sum_i d(x_i, y)] =r(X)\). [Gross's result is for compact connected metric spaces \((X,d)\).] The number \(r(X)\) is called the rendezvous number of \(X\). See \textit{J. Cleary}, \textit{S. A. Morris}, and \textit{D. Yost} [Am. Math. Mon. 93, 260-275 (1986; Zbl 0598.51014)] for a survey of this topic. If \(\gamma(E)= \sup( \sigma_n)\) and \(k_2=\sup (r(X))\), where \(X\) ranges over all compact connected subsets of \(E\), with diameter 1, the author proves: \(k_2\leq\gamma (E)\leq[687+ 1408(1-\sqrt 2)^{1/2}+ 95\sqrt 5]/2400 \approx 0.678442\), an improvement of the previously best known upper bound. See \textit{R. Wolf} [Isr. J. Math. 110, 125-151 (1999; Zbl 0934.46013)].