A note on the Schur and Phillips lemmas (Q2676968)

Phillips's classical lemma refers to a sequence \((\mu_n)\) of finitely additive measures on the subsets of \(\mathbb{N}\) and states that if \(\mu_n(A)\to0\) for all \(A\in\{0;1\}^\mathbb{N}\), then \(\sum_k\mu_n(\{k\})\to0\). A reformulation reads: the canonical projection from \((\ell^\infty)^*=(\ell_1)^{**}\) onto \(\ell_1\) is sequentially weak\( ^*\)-norm continuous. In the special case when the \(\mu_n\) are \(\sigma\)-additive (equivalently, can be considered elements of \(\ell_1\)) Schur's lemma is obtained: if \(\mu_n(A)\to0\) for all \(A\in\{0;1\}^\mathbb{N}\), then \(\|\mu_n\|_{\ell_1}\to0\) (or, equivalently, weakly convergent sequences in \(\ell_1\) converge in norm). The starting point for the note under review is a recent result of \textit{D.~Sobota} [Proc. Am. Math. Soc. 146, No.~4, 1673--1680 (2018; Zbl 1441.46019)]: in Phillips's lemma it is enough to assume \(A\in C\) where \(C\subset\{0;1\}^\mathbb{N}\) is of second category in \(\{0;1\}^\mathbb{N}\) (where \(\{0;1\}^\mathbb{N}\) is considered with its product topology). First, the author gives a short proof of Sobota's analogous result for Schur's lemma. Then he defines ifp-sets: a set \(C\subset\{0;1\}^\mathbb{N}\) is an ifp-set (for `interpolation property for finite sets') if for any sequence \((F_n)\) of disjoint finite subsets \(F_n\) of \(\mathbb{N}\) there are an infinite set \(I\subset\mathbb{N}\) and \(A\in C\) such that \(\bigcup_{n\in I}A_n\subset A\) and such that \(A\) with the possible exception of finitely many elements is contained in \(\bigcup_{n\in N}A_n\). This notion is technical but natural in order to show that Phillips's lemma remains valid if \(C\) is such an ipf-set. A (possibly easy) question: are ifp-sets enough to detect sequential weak\(^*\)-convergence in \((\ell^\infty)^*\)?