On weak\(^*\) convergent sequences in duals of symmetric spaces of \(\tau\)-measurable operators (Q1686307)

The paper deals with property (K) defined (implicitly) in a paper of \textit{N. J. Kalton} and \textit{A. Pełczyński} [Math. Ann. 309, No. 1, 135--158 (1997; Zbl 0901.46008)]: A Banach space \(X\) has property (K) if every weak\(^*\) null sequence in the dual \(X^*\) admits convex blocks that converge uniformly to \(0\) on weakly null sequences of \(X\). If the uniform convergence is claimed only on weakly null sequences of the form \((T(f_n))\), where \(T:L^1([0,1])\rightarrow X\) is a bounded operator and \((f_n)\) a weakly null sequence in \(L^1([0,1])\), then one obtains the weaker property (k) as defined by \textit{T. Figiel} et al. [Isr. J. Math. 183, 199--231 (2011; Zbl 1235.46027)]. As an example, \(c_0\) fails property (K) while, by Mazur's theorem, Grothendieck spaces (for example, \(\ell^\infty\) or, more generally, von Neumann algebras) are easily seen to have property (K). But the paper under review focuses rather on preduals of von Neumann algebras and on weakly sequentially complete noncommutative symmetric spaces. In this regard, Figiel et al. have shown in [loc. cit.]\ that separable subspaces of von Neumann preduals, as well as weakly sequentially complete Banach lattices, have property (k). The first of the two main results states that the predual of a \(\sigma\)-finite von Neumann algebra has property (K) and hence also (k). The result is sharp in the sense that, by an example of Figiel et al. [loc. cit.], it may fail without \(\sigma\)-finiteness. The result answers in the positive a question of Figiel et al. [loc. cit.]. Some aspects of the latter paper are now more streamlined and put into a more systematic context; for example, a scale of properties (K\(_\mathcal{C}\)) is introduced where the \(\mathcal{C}\) are classes of bounded subsets of \(X\): \(X\) has property (K\(_\mathcal{C}\)) if every weak\(^*\) null sequence in \(X^*\) admits convex blocks that converge to \(0\) in the topology \(\mathcal{T}_\mathcal{C}\) of uniform convergence on the sets of \(\mathcal{C}\). Then (K) is recovered if \(\mathcal{T}_\mathcal{C}\) is the Mackey topology. Similarly, also property (k) corresponds to (K\(_\mathcal{C}\)) for a certain class \(\mathcal{C}\). After observing that property (K\(_\mathcal{C}\)) follows from the metrizability of \(\mathcal{T}_\mathcal{C}\) on (convex) bounded sets, the proof of the main result reduces to the metrizability of the Mackey topology in \(\sigma\)-finite von Neumann algebras. (Reviewer's side remark: The proof of this metrizability can be shortened even in the nonseparable case by using Theorem III.5.3 of \textit{M. Takesaki}'s book [Theory of operator algebras I. New York, Heidelberg, Berlin: Springer-Verlag (1979; Zbl 0436.46043)] together with the other theorems from the book used by the authors.) The second main result yields a noncommutative counterpart of the aforementioned result on lattices: If \(E\) is a weakly sequentially complete symmetric space of measurable operators affiliated with a semifinite and \(\sigma\)-finite von Neumann algebra, then \(E\) has property (k). The paper [\textit{A. Avilés} and \textit{J. Rodríguez}, Mediterr. J. Math. 13, No. 6, 4995--5007 (2016; Zbl 1367.46017)] deals with property (K), but is independent of the one under review (and vice versa); the reader may find it useful to combine the reading of both.