Weak integral convergence theorems and operator measures (Q1061352)

In the first part of this paper (submitted in 1978 and revised in 1983), an integration theory is outlined for (certain measurable) functions with values in a locally convex (l.c.) space X with respect to a measure \(\mu\) with values in the continuous linear operators L(X,Y) from X into some (quasi-complete) l.c. space Y such that \(\mu\) (\(\cdot)x\) is countably additive (c.a.) for each \(x\in X\). (The authors use a generalization of the methods of Hahn, Vitali and F. Riesz.) For vector-valued functions (and even in the case of scalar measures), it is known that convergence in measure for the integrands and pointwise a.e. convergence are not related as they are in the scalar case, and the authors here note an example showing that, for functions on [0,1] with values in an uncountable product of real lines, convergence in (Lebesgue) measure no longer implies the a.e. convergence of a subsequence. In the second part of the article (which contains the main results), the authors are interested in the dominated and bounded convergence theorems for pointwise a.e. convergence. Following Goodrich's Riesz representation theory, they show that, for a compact Hausdorff space H and the representing measure \(\mu\) (defined on B(H)) of a (continuous linear) operator T from C(H,X) into Y, countable additivity of \(\mu\) (in the above sense) with respect to the natural topology on the bidual Y'' implies both types of convergence theorems (in some topology), and the converse results hold, too (with a certain change of topology). The authors also classify those operators T: C(H,X)\(\to Y\) (represented by the measures \(\mu)\) for which the convergence theorems hold (by some type of ''pointwise weak compactness'', resp., by the property that \(\mu\) (E)x\(\in Y\) for all x and E). The article closes with remarks on other types of integrals (e.g., using the technique of Bartle) and more general types of representation theorems.