Weak\(^*\)-sequential properties of Johnson-Lindenstrauss spaces (Q1732307)

Recall that a compact space $K$ is Fréchet-Urysohn if every element contained in the closure of a set $F\subset K$ is the limit of a sequence in $F$; and that $K$ is sequential if sequentially closed subsets $F$ of $K$ are closed in $K$. \par There are variants of those properties for dual balls of Banach spaces. Namely, a Banach space $X$ has property $(\mathcal{E})$ if every element contained in the closure of a convex set $F\subset (B_{X^*},w^*)$ is the limit of a sequence in $F$; and it has $(\mathcal{E}')$ if sequentially closed convex subsets $F$ of $(B_{X^*},w^*)$ are closed in $(B_{X^*},w^*)$. \par Main result: assuming the continuum hypothesis (CH), there are two maximal almost disjoint families $\mathcal{F}_+$ and $\mathcal{F}_-$ of $\mathbb{N}$ for which the corresponding Johnson-Lindenstrauss spaces $JL_2(\mathcal{F}_+)$ and $JL_2(\mathcal{F}_-)$ enjoy and fail property $(\mathcal{E})$, respectively. \par This is related to a gap in a paper by \textit{A. Plichko} [Topology Appl. 190, 93--98 (2015; Zbl 1332.46024)] (where the author claims that $JL_2(\mathcal{F})$ has $(\mathcal{E})$ whenever $\mathcal{F}$ is maximal almost disjoint) and provides under (CH) first examples of Banach spaces which witness that none of the implications below may be reversed in general: \[ (B_{X^*},w^*)\text{ is Fréchet-Urysohn} \implies X\text{ has }(\mathcal{E}) \implies X\text{ has }(\mathcal{E}'). \] It remains open whether we may construct such families $\mathcal{F}_+$ and $\mathcal{F}_-$ in ZFC.