No label defined (Q4953560)

In this paper some comparisons between Bochner and monotone integrals for scalar measurable functions with respect to Hilbert-space-valued finitely additive measures are given. Let \(\Omega \) be an arbitrary nonempty set, \(\Sigma \) a \(\sigma \)-algebra of subsets of \(\Omega \), \(H\) a separable Hilbert space; \(m\:\Sigma \to H\) a bounded finitely additive measure. A measurable function \(f\:\Omega \to \mathbb R\) is \(m\)-integrable if there exist a control measure \(\lambda \: \Sigma \to [0,+\infty)\) and a defining sequence of simple functions \((f_n)_n\). The symbol \(L^1(m)\) denotes the set all \(m\)-integrable functions. If \(m\) is \(s\)-bounded and \(f\:\Omega \to [0,+\infty)\) is an \(m\)-measurable function, let \(\widehat \varphi \), \(\overline \varphi \:[0,+\infty)\to [0,+\infty)\) be defined as follows: \[ \widehat \varphi (t)=\|m\|(\{\omega \in \Omega \:f(\omega)>t\}),\quad \overline \varphi (t)=\|m\|^2(\{\omega \in \Omega \:f(\omega)>t\}). \] \(f\) is \((\sphat )\)-integrable if \(\widehat \varphi \) is Lebesgue integrable. If \(f\:\Omega \to \mathbb R\), then \(f\) is said to be \((\sphat )\)-integrable if \(f^+, f^-\) are \((\sphat )\)-integrable. \(\widehat {L^1} (m)\) denotes the set of all \((\sphat )\)-integrable functions. Let \(m\) be an orthogonally scattered finitely additive measure, namely a finitely additive measure such that \(\langle m(A),m(B)\rangle=0\) if \(A\cap B=\emptyset\), and let \(f\:\Omega \to [0,+\infty)\) be an \(m\)-measurable function: \(f\) is \((\sptilde)\)-integrable if \(\widehat \varphi \) is Lebesgue integrable. If \(f\) takes values in \(\mathbb R\), then its integrability is given again in terms of \(f^+\) and \(f^-\), and \(\widehat {L^1} (m)\) is the set of all \((\sptilde)\)-integrable functions. It is well known (see Theorem 3.9. of \textit{J. K. Books} and \textit{A. Martellotti} [Atti Sem. Mat. Fis. Univ. Modena 40, No.~1, 285-302 (1992; Zbl 0797.28007)]) that \(\widehat {L^1} (m)\subset L^1(m)\) (even if \(m\) takes values in a separable Banach space), but the converse inclusion is not true in general (see ibidem, Example 3.10). In the paper some sufficient conditions are given, under which \(\widehat {L^1} (m)= L^1(m)\). Let \(\widetilde H\) be a separable Hilbert space, in general, different from \(H\), and let \(\widetilde m\:\Sigma \to \widetilde H\) be an orthogonally scattered dilation of \(m\), as defined by \textit{S. D. Chatterji} [Lect. Notes Math. 945, 269-281 (1982; Zbl 0489.28007)]. If \(m\) is an orthogonally scattered dilation of \(m\), the authors prove that \(L^1(\widetilde m)\subset L^1(m)\), and the converse inclusion does not hold in general. Finally, conditions ensuring that the inclusions \(L^1(\widetilde m)\subset\widetilde{L^1} (\widetilde m)\) and \(\widehat {L^1}(m)\subset \widetilde {L^1}(\widetilde m)\) and related results are given.