Boundaries for Banach spaces determine weak compactness (Q607717): Difference between revisions

R.\,C.\thinspace James's weak compactness theorem is one of the most powerful and difficult results in functional analysis. It says that a bounded subset \(A\) of a real Banach space \(X\) is weakly compact if (and only if) each continuous linear functional on \(X\) attains its supremum over \(A\), that is, for each \(x^*\in X^*\) there exists some \(a_0\in A\) such that \(x^*(a_0)= \sup_{a\in A} x^*(a)\). In particular, \(X\) is reflexive if (and only if) each \(x^*\in X^*\) attains its norm. The proof of this result is notoriously complicated, especially in the nonseparable case. Inspired by James's theorem, \textit{G.~Godefroy} [``Boundaries of a convex set and interpolation sets'', Math.\ Ann.\ 277, 173--184 (1987; Zbl 0597.46015)] introduced the notion of a boundary of a Banach space \(X\): \(B\subset B_{X^*}\) is called a (James) boundary if, for each \(x\in X\), there is some \(x_0^*\in B\) such that \(x_0^*(x)=\|x\|=\sup_{x^*\in B_{X^*}} x^*(x)\). In the sequel, \(B\) will stand for a boundary. Godefroy asked whether a bounded subset \(A\subset X\) is weakly compact provided that it is compact with respect to the topology \(\sigma(X,B)\) of pointwise convergence on \(B\); this has become known as the boundary problem. A~number of authors have given positive answers under various additional assumptions, e.g., for convex sets \(A\) or for certain boundaries like the set of extreme points of the dual unit ball. In this paper, the author gives a positive solution to the boundary problem in complete generality. In particular, he obtains a new proof of James's weak compactness theorem that he shows to be a consequence of his theorem by means of Grothendieck's double limit criterion. The author's argument is combinatorial in nature and has several highly nontrivial ingredients: the study of various moduli for the diameter of the set of weak cluster points of a sequence, a quantitative version of Rosenthal's \(\ell_1\)-theorem in the spirit of E.~Behrends's version, an application of the Simons lemma, and a variant of the Hagler-Johnson construction. The latter is needed in order to rule out that a \(\sigma(X,B)\)-compact set contains an \(\ell_1\)-basis. Indeed, the main theorem of the paper is somewhat more general than the solution of the boundary problem; it turns out that a bounded \(\sigma(X,B)\)-countably compact set is actually weakly compact. However, the author points out that it remains open whether bounded sets are \(\sigma(X,B)\)-angelic, although he had claimed that in a preliminary version of the paper.