Measure and integration theory. Transl. from the German by Robert B. Burckel (Q5936647)

It is a pleasure to see Heinz Bauer's famous monograph \textit{Maß- und Integrationstheorie} (1990; Zbl 0714.28001; 2nd ed. 1992; Zbl 0748.28001) published in English. The book consists of four chapters: I - Measure Theory. II - Integration Theory. III - Product Measures. IV - Measures on Topological Spaces. Chapter I contains a rather straight introduction to systems of sets, measures and their extension, Lebesgue measure, and measurable mappings. Chapter II develops the Lebesgue integral and proceeds with a comprehensive study of its properties, including convergence theorems, results on continuity and differentiability of integrals depending on a real parameter, and formulas for the change of variables, other topics include the Radon-Nikodým theorem, convergence in measure, and equi-integrability. Chapter III is devoted to finite products of measures, Fubini's theorem, and the convolution of finite Borel measures. These three chapters on abstract measure and integration theory with special consideration of the Lebesgue measure constitute a solid fundament for modern probability theory. Chapter IV studies Radon measures on topological spaces which are Polish or locally compact and hence serves for applications in probability theory as well. Measure and integration theory is essential not only in probability theory but also in functional analysis. From a functional analytic point of view, one may regret that in this book duality and order properties of \(L^p\)-spaces are neglected and that the discussion of the Hilbert space \(L^2\) does not include the projection theorem, which is of central importance in statistical prediction theory. I also find that too much prominence is attributed to the \(d\)-dimensional Lebesgue measure, which can more easily be obtained as a product measure, and that too little attention is given to signed measures, which arise quite naturally once the integral over a measurable set is introduced and which are essential in the proof of the Radon-Nikodým theorem. In the short section on signed measures that ``can be skipped over in a first reading'', I would prefer the Jordan decomposition to proceed the Hahn decomposition since the former can be extended to finite signed contents. However, the author has made his choice, and these local comments do not affect my global appreciation of the book. The translation by Robert B. Burckel is careful and close to the German original. He has replaced references to German textbooks by references to English textbooks, and he has also added several interesting exercises, comments, and references. The typographical appearance of the book is excellent.