On a one-sided James' theorem (Q504869)

The following ``one-sided James' theorem'' has been proved by \textit{B. Cascales} et al. [J. Math. Anal. Appl. 445, No. 2, 1267--1283 (2017; Zbl 1362.46005)] under the additional hypothesis that the unit ball of \(X^*\) is weak\(^*\) convex compact and is now shown in complete generality, and with different methods, in the paper under review. Let \(X\) be a real Banach space, \(X^*\) its dual, and \(A,B\) be two non-empty bounded closed convex subsets of \(X\) with strictly positive distance. If every \(x^*\in X^*\) such that \(\sup\langle x^*(B)\rangle<\inf\langle x^*(A)\rangle\) attains its infimum on \(A\) and its supremum on \(B\), then both \(A\) and \(B\) are weakly compact. The proof is short, essentially because it uses James' theorem as an ingredient (while the proof of Cascales et al.\ [loc.\,cit.]\ admits (under an additional hypothesis) James' theorem as a corollary). More precisely, the proof uses a result of \textit{J. Saint Raymond} [Mediterr. J. Math. 10, No. 2, 927--940 (2013; Zbl 1273.46057)] of which the author has given a short proof based on James' theorem in [Proc. Am. Math. Soc. 145, No. 8, 3377--3379 (2017; Zbl 1382.46012)]. (In the latter article, the author uses the Fenchel conjugate in a similar way as in the paper under review.) The author remarks that the theorem remains valid (even with \(A\) and \(B\) not necessarily disjoint) if the inequality in the hypothesis is reversed, in which case, by a nice elementary trick, the statement becomes practically equivalent to James' theorem.