English

Maximal equilateral sets

Metric Geometry 2013-09-17 v4 Combinatorics Functional Analysis

Abstract

A subset of a normed space XX is called equilateral if the distance between any two points is the same. Let m(X)m(X) be the smallest possible size of an equilateral subset of XX maximal with respect to inclusion. We first observe that Petty's construction of a dd-dimensional XX of any finite dimension d4d\geq 4 with m(X)=4m(X)=4 can be generalised to give m(X1R)=4m(X\oplus_1\mathbb{R})=4 for any XX of dimension at least 2 which has a smooth point on its unit sphere. By a construction involving Hadamard matrices we then show that for any set Γ\Gamma, m(p(Γ))m(\ell_p(\Gamma)) is finite and bounded above by a function of pp, for all 1p<21\leq p<2. Also, for all p[1,)p\in[1,\infty) and dNd\in\mathbb{N} there exists c=c(p,d)>1c=c(p,d)>1 such that m(X)d+1m(X)\leq d+1 for all dd-dimensional XX with Banach-Mazur distance less than cc from pd\ell_p^d. Using Brouwer's fixed-point theorem we show that m(X)d+1m(X)\leq d+1 for all dd-dimensional XX with Banach-Mazur distance less than 3/2 from d\ell_\infty^d. A graph-theoretical argument furthermore shows that m(d)=d+1m(\ell_\infty^d)=d+1. The above results lead us to conjecture that m(X)1+dimXm(X)\leq 1+\dim X for all finite-dimensional normed spaces XX.

Keywords

Cite

@article{arxiv.1109.5063,
  title  = {Maximal equilateral sets},
  author = {Konrad J. Swanepoel and Rafael Villa},
  journal= {arXiv preprint arXiv:1109.5063},
  year   = {2013}
}

Comments

15 pages. This version incorporates some suggestions from a referee obtained after galley proofs. Equations (18) and (23) are corrected

R2 v1 2026-06-21T19:09:19.331Z