Symmetrically separated sequences in the unit sphere of a Banach space (Q1788788): Difference between revisions

Since its appearance, Kottman's constant $K(\cdot)$ [\textit{C. A. Kottman}, Trans. Am. Math. Soc. 150, 565--576 (1970; Zbl 0208.37503)] -- the supremum of the separation that an infinite sequence of points of the unit ball of a Banach space may have -- has been a quite dear topic to Banach space geometry people and many papers containing an explicit calculation of Kottman's constant in particular spaces have been published. \textit{C. A. Kottman} [Stud. Math. 53, 15--27 (1975; Zbl 0266.46014)] stated a beautiful combinatorial result to show that every unit ball of an infinite-dimensional Banach space contains a sequence separated by more than one. This result was completed by the so-called Elton-Odell theorem [\textit{J. Elton} and \textit{E. Odell}, Colloq. Math. 44, 105--109 (1981; Zbl 0493.46014)], stating that $K(X)>1$ for every infinite-dimensional Banach space, which is an extraordinary and difficult result. On the other hand, the origins of the symmetric Kottman constant $K^s(\cdot)$ -- the supremum of the separation that an infinite symmetrically distributed sequence of points of the unit ball of a Banach space may have -- are obscured by clouds: it seems to have been introduced in [\textit{J. M. F. Castillo} and \textit{P. L. Papini}, Banach Cent. Publ. 92, 75--84 (2011; Zbl 1247.46008)], where the natural question ``does there exist an Elton-Odell theorem for $K^s(\cdot)$?'' was posed. \par This paper is devoted to that problem and provides a reasonably good partial answer. The authors prove Kottman's result for $K^s(\cdot)$ and obtain a good bunch of quantitative estimates and qualitative results that push things very close to an Elton-Odell theorem for $K^s(\cdot)$. For instance, Theorem 1.1 is really close: Let $X$ be a Banach space that contains a boundedly complete basic sequence. Then, for some $\varepsilon>0$, the unit sphere of $X$ contains a symmetrically $(1 + \varepsilon)$-separated sequence. \par This, plus the many other horizontal results obtained in the paper, leads to a thoughtful discussion in the final section about how close the results of the authors actually are to a positive solution to the symmetric Elton-Odell problem. For instance, it would be true if the following conjecture, of independent interest, were true: Every separable Banach space contains either an $\mathcal L_\infty$-space or admits a reflexive quotient. \par The techniques displayed in the paper also deserve to be mentioned. The authors prove Kottman's symmetric version not via a combinatorial improvement of Kottman's original argument but using a clever Riesz-like lemma approach. The paper is very well written and the authors provide the reader generously with hints and explanations; in fact, somebody else could have rendered a nice paper just from the comments spread throughout this one. \par Reviewer's remark. This is but an opinion, but it seems to me that there is still a remaining underwater 7/8 mass of the Kottman constant iceberg. A~paper one could invoke to support this assertion is [\textit{N. J. Kalton}, Isr. J. Math. 162, 275--315 (2007; Zbl 1148.46045)] who, as always, provides an unexpectedly deep use of Kottman's constant to estimate the Lipschitz extension constant (see also [\textit{J. S. de la Fuente}, Can. Math. Bull. 60, No. 4, 855--860 (2017; Zbl 1384.46016)]).