On integration with respect to LCTVS-valued finitely additive measures (Q1891449)

Let \(X\) be a real locally convex Hausdorff space. Let \(m\) be a strongly bounded \(X\)-valued finitely additive measure defined on a \(\sigma\)-algebra \(\Sigma\) of subsets of a set \(\Omega\) \((\neq \emptyset)\). For a \(\Sigma\)-measurable real function \(f\) on \(\Omega\), the authoress defines with respect to \(m\) the concepts of \(m\)-integral, weak \((\widehat{\;})\)-integral, strong \(m\)-integral, strong \((\widehat{\;})\)-integral and strong (*)-integral and makes a comparative study of these integrals. When \(X\) is a real Banach space, strong \(m\)-integral and strong \((\widehat{\;})\)-integral are the same as the \(m\)-integral and \((\widehat{\;})\)-integral introduced in the earlier work of \textit{J. K. Brooks} and the authoress [Atti. Semin. Mat. Fis. Univ. Modena 40, No. 1, 285-302 (1992; Zbl 0797.28007)]. Remarks of the reviewer: The proof of Theorem 2 holds only when the strong topology of \(X^*\) is metrisable. The definition of the strong \(m\)-integral of \(f\) should be strengthened by requiring the existence of a finitely additive positive control measure \(\lambda_p\) equivalent to \(|m |_p\) for each continuous seminorm \(p\) on \(X\). In the light of the observation made on p. 207 of \textit{R. Sikorski}: ``Boolean algebras'' (1964; Zbl 0123.01303) about the pointwise convergence of the functions \(\overline f\) associated with \(\Sigma\)-measurable functions \(f\) on \(\Omega\), the validity of the arguments involving the use of \(\int \overline fd \overline m\) in the Stone representation space of \(\Sigma\) and other properties of \(\overline f\) not mentioned in Sikorski's book (the authoress tacitly assumes them, without proving either in the present paper or in the earlier said joint paper) is suspected by the reviewer. Consequently, Theorem 3 (and hence 4(i) and 4(ii) of Theorem 4 and Theorem 5) and Theorems 12 and 13 are questionable. Moreover, for the validity of the second part of Theorem 11 the author should include the hypothesis that \(\varphi\) is Pettis integrable. The last section is devoted to the study of Burkill-Cesari integrals with respect to \(m\) (the details of which were not checked by the reviewer).