Comparisons of stochastic matrices, with applications in information theory, statistics, economics, and population sciences (Q4212656)

The focus of this book is to compare the diversity or variation in a set of probability distributions. A discrete probability distribution is represented by probabilities \(p_1,\dots, p_n\) with \(p_i\geq 0\), \(\sum p_i= 1\). Consequently, a set of discrete probability distributions can be represented as the columns of a column-stochastic matrix. Therein lies the essence of the title. One of the goals of the book is to show comparisons in terms of different applications. From information theory a discrete memoryless channel relates to a stochastic matrix, whereas in statistics the outcome of an experiment relates to a stochastic matrix. In economics it is the distribution of assets over individuals, and in genetics it is allele frequencies of a single locus in different populations. A discussion of characteristics within each application is proved. The book is divided into two parts, the first dealing with discrete distributions is co-authored by the three authors. The second part deals with Markov kernels as extensions of stochastic matrices, and is authored by the third author. Although the various applications are presented, the book focuses more on an information-theoretic framework. The book is compact and is dense with information. There is considerable motivation in the introduction, but very little in the text. A reader would have appreciated a more extended exposition. As a result, the reader is required to spend considerable effort in reading this book. This is not a book for browsing, but does contain much material for the specialist.