Generalizations of the Nikodym boundedness and Vitali--Hahn--Saks theorems (Q703659)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Generalizations of the Nikodym boundedness and Vitali--Hahn--Saks theorems
scientific article

    Statements

    Generalizations of the Nikodym boundedness and Vitali--Hahn--Saks theorems (English)
    0 references
    0 references
    0 references
    11 January 2005
    0 references
    Already in 1933, Nikodym showed that for a \(\sigma\)-algebra \(\Sigma\) of subsets of a set \(\Omega\) a setwise bounded family \((\mu_\alpha)_{\alpha\in A}\) of countably additive measures is uniformly bounded. That is, if \(\sup_{\alpha\in A}| \mu_\alpha (E)| <\infty\), then \(\sup_{\alpha\in A}| \mu_\alpha| (\Omega)<\infty\). In the very early fifties it was known to Dieudonné and Grothendieck that countable additivity is not needed and they also observed that the argument used in the scalar case goes through for Banach-space valued measures. The scalar case translates into a Uniform Boundedness Principle: Let the linear functionals \((\mu_\alpha)\) be pointwise bounded on the family \(\{\chi_E:E\in\Sigma\}\) in the Banach space \(B(\Sigma)\) of bounded measurable functions. Then \((\mu_\alpha)\) is already uniformly bounded. Subsets of Banach spaces that have this massiveness property that they decide uniform boundedness are more or less by definition exactly the sets which have barrelled span. It is well known that sets \(U\) with barrelled span are characterized by the following property: If \(Y\) is a Banach space and the range of the bounded operator \(T:Y\rightarrow X\) contains \(U\), then \(T\) is onto. A third characterization from Kadets and Fonf is the following: The set \(U\) can not be written as an increasing, countable union of non-norming sets. This clearly tells us that it is the `size' of the algebra or ring that counts. This `size' is usually called property (N). Various authors have found properties for algebras and rings in order to ensure it to have (N) but so far no internal characterization is found. The basic reference is the work of \textit{W. Schachermayer} [Diss. Math. 214 (1982; Zbl 0522.28007)]. The main tool in the paper under review is the ring property bounded subsequential completeness property (BSCP) which implies (N). It is inspired from a property defined by Haydon, the subsequential completeness property (SCP), which he showed is properly stronger than (N). The authors prove that many earlier properties are covered by BSCP. Among these are (SCP), density zero and uniform density zero. The Vitali-Hahn-Saks theorem states that if a sequence of finitely additive measures on a \(\sigma\)-algebra converges setwise, then the sequence is a uniformly strongly additive family of measures. When an algebra or a ring is such that this theorem holds it is said to have property (VHS). Diestel, Faires and Huff have shown that (VHS) is equivalent to (N) \& (G), where (G) denotes the so-called Grothendieck property. None of (N) or (G) implies (VHS), the last being a very `obscure' example by Talagrand. A property quite similar to (BSCP), the normed subsequential completeness property (NSCP), is introduced in the paper and proved to imply (VHS).
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    Nikodym property
    0 references
    Vitali-Hahn-Saks property
    0 references
    bounded subsequential completeness property
    0 references
    0 references