On Nikodym boundedness property (Q380429)

Let \(\mathcal{A}\) be some algebra of subsets of a set \(\Omega\) and denote, as usual, by \(ba(\mathcal{A})\) and \(ca(\mathcal{A})\) the spaces of bounded additive and countably additive complex measures on \(\mathcal{A}\), respectively. Equipped with the variation as the norm, \(ca(\mathcal{A})\) becomes a closed subspace of the Banach space \(ba(\mathcal{A})\), and every \(A\in\mathcal{A}\) can be viewed as a dual element of norm one by \(A(\mu)=\mu(A)\). In fact, \(ba(\mathcal{A})\) is the dual of \(\ell_0^\infty(\Omega,\mathcal{A})\), the uniform closure of the linear span of the characteristic functions on \(\mathcal{A}\) by the action \(\mu(f)=\int_\Omega f\;d\mu\). Thus, if we associate \(A\in\mathcal{A}\) with \(\chi_A\in \ell_0^\infty(\Omega,\mathcal{A})\), we see that the functionals \(A\) on \(ba(\mathcal{A})\) described above are weak-star continuous. Now, for a moment, let \(X\) be an arbitrary Banach space. Let us say that a set \(M\subset X\) is a uniform boundedness deciding set (ubd-set) if, whenever a family \(\mathcal{F}\in X^\ast\) is pointwise bounded on \(A\), it is uniformly bounded. The Banach-Steinhaus theorem tells us that, \(X\) itself is a ubd-set and the beautiful Nikodým-Dieudonné theorem tells us that, if \(\mathcal{A}\) is a \(\sigma\)-algebra, then \(M=\{A\in\mathcal{A}\}\subset \ell_0^\infty(\Omega,\mathcal{A})\) is a ubd-set. There exists quite a big theory of ubd-sets, see the reviewer's survey paper [\textit{O. Nygaard}, Quaest. Math. 29, No. 1, 59--72 (2006; Zbl 1108.46006)]. Among other things, one can show that \(M\) being a ubd-set is equivalent to the following property: Whenever \(M\) is realized as a non-decreasing countable union \(M=\bigcup_{j=1}^\infty B_j\), then, for some \(n_0\in\mathbb{N}\), the absolutely convex hull of \(A_{n_0}\) contains some ball. It is therefore highly remarkable that the author in [Ann. Inst. Fourier 29, No. 3, 39--56 (1979; Zbl 0379.46004)] proved that, when \(\mathcal{A}\) is a \(\sigma\)-algebra, then the set \(M=\{A\in\mathcal{A}\}\subset \ell_0^\infty(\Omega,\mathcal{A})\) has the following property: Whenever \(M\) is realized as a non-decreasing countable union \(M=\bigcup_{j=1}^\infty B_j\), then, for some \(n_0\in\mathbb{N}\) \(A_{n_0}\), is a ubd-set! \(M\) is a ``super-ubd-set''. A subcollection \(\mathcal{B}\subset\mathcal{A}\) is said to have Nikodým's boundedness property if \(M=\{A\in\mathcal{B}\}\subset \ell_0^\infty(\Omega,\mathcal{A})\) is a ubd-set. The main new theorem in the paper under review (Theorem 2) is that the algebra (but not \(\sigma\)-algebra) \(\mathcal{A}\) of Jordan measurable subsets in \(\mathbb{R}^k\) is, in fact, a super-ubd-set. Before it was known that the algebra of Jordan measurable subsets in \([0,1]\) has Nikodým's boundedness property. The paper ends with an interesting question: Is it so that, whenever an algebra \(\mathcal{A}\) has Nikodým's boundedness property, then the set of characteristic functions \(\{\chi_A, A\in\mathcal{A}\}\) is automatically a super-ubd-set in \(\ell_0^\infty(\Omega,\mathcal{A})\)? Reviewer's remark: In fact, it is not known to the reviewer whether there exist ubd-sets which are not already super-ubd-sets in general.