The normal distribution. Characterizations with applications (Q1892404)

This book consists of a Preface, Introduction, 8 chapters, and Solutions of selected problems. There are 153 references. Chapter 1, Probability tools. ``Most of the content of this section is fairly standard probability theory \dots The emphasis here is on analytic methods; in particular characteristic functions will be extremely used throughout''. Chapter 2, Normal distributions. ``In this chapter we use linear algebra and characteristic functions to analyse the multivariate normal random variables.'' Also, in section 2.5 the author presents two classical theorems (Cramér decomposition theorem and Marcinkiewicz theorem) giving criteria for normality. Chapter 3, Equidistributed linear forms. In Section 3.1 the classical Polya's characterization result is presented. It is used to define Gaussian distributions (in Polya's sense) on abstract linear spaces and a proof of the zero-one law for such distributions is given. In Section 3.3 a problem of characterization of the normal distribution by the equality of distributions of two general linear forms is considered. Then the author gives some analogous results for the exponential distribution. Chapter 4, Rotation invariant distributions. First the author defines and studies properties of spherically symmetric distributions (some properties of infinite spherically symmetric sequences are presented in Section 4.3). One characterization result is given here. Then a result due to Braverman is presented, characterizing the normal distribution by the rotation invariance of absolute moments. Chapter 5, Independent linear forms. ``In this chapter the property of interest is the independence of linear forms in independent random variables.'' First, the classical Bernstein's characterization theorem is proved. Then the author defines Gaussian distributions on groups (in the sense of Bernstein) and obtains the zero-one law for Gaussian distributions on vector spaces. In Section 5.4, following Fernique, the author gives yet another definition of Gaussian random variables and proves a zero-one law. In Section 5.3 a one-dimensional generalization of Berstein's theorem is presented, showing that the normal distribution can be characterized by independence of two linear forms of independent, identically distributed random variables. The result is slightly less general than that of Darmois and Skitovich. Chapter 6, Stability and weak stability. This gives ``\dots a short introduction to measures of dependence and stability issues. Theorem 6.2.2 establishes integrability under conditions of interest, e.g. in polynomial biorthogonality.'' Chapter 7, Conditional moments. ``In this chapter we use assumptions that mimic the behavior of conditional moments that would have followed from independence.'' Here the author also presents some results that characterize the normal distribution and Gaussian sequences by certain restrictions imposed on conditional moments. As an application, the central limit theorem for i.i.d. sums is proved, and a characterization of the normal distribution by the independence of empirical mean and empirical variance is given. Chapter 8, Gaussian processes. This is ``\dots a short introduction to continuous-parameter random fields, analysed through their conditional moments. We also present a self-contained analytic construction of the Wiener process.'' In Section 8.3, the author presents Lévy's characterization theorem. Some characterizations of processes without continuous trajectories are given. From the Preface: ``We have tried to make this book accessible to readers with different backgrounds. If possible, we give elementary proofs of important theorems, even if they are special cases of more advanced results\dots This book is primarily aimed at graduate students in mathematical statistics and probability theory who would like to expand their bag of tools, to understand the inner working of the normal distribution, and to explore the connections with other fields.'' Sections 4.2, 7.5 and 8.3 of this book present quite recent results. In general, the book neither describes the current state of the theory nor gives up-to-date references. In particular, in Chapter 6 the problem of stability of characterizations is outlined. Here the author's references leave out the (great number of) results after 1977. Another example: In Sections 3.2 and 5.2 different definitions of abstract Gaussian random variables are given. We read on p. 73: ``Clearly, any vector space is an Abelian group with vector addition as the group operation. In particular, we now have two possibly distinct notions of Gaussian vectors \dots\ introduced in Section 3.2 and \dots\ in this Section. In general, it seems to be not known when the two definitions coincide.'' But a complete solution to this very problem for Abelian groups can be found in the references of this book (items 49, 51 and 52).