This book was written for statisticians and econometricians whose interests lie in optimization paradigms. The book is not designed for recreational reading. Only the serious professional will appreciate the painstaking detail of the material. The authors clearly have not only academic credentials, but also the practical experience to put together a finely honed text to benefit academics and professionals. The book is divided into four parts. Part 1 deals with a broad background of optimization techniques used in various arenas. After this broad base of interest is established, Part 2 contains a deeper discussion of optimization heuristics.

Part 3 naturally continues with practical applications of optimization heuristics in statistics and econometrics. Part 4 is a summary of the main arguments proposed by the authors in the book. After an exhaustive introductory chapter, chapter 2 summarizes the importance of optimization paradigms in economics. Chapter 3 continues in the same mode with a discussion of optimization paradigms as used in statistics. While chapters 1 and 2 present some examples of optimization paradigms, chapter 4 details the specifics of the optimization paradigm with mathematical rigor.

Part 2 is exciting. After Part 1, which lays the groundwork, the second part of the text discusses optimization heuristics threshold accepting and all of its properties, performance and tuning. Chapter 5 again prepares the reader for threshold accepting by discussing other methods of classical and heuristic optimization methods. Classical methods of least squares, gradient, enumeration, grid search, and local search are presented. Under optimization heuristics, the reader is exposed to simulated annealing, generic algorithms, tabu search, great-deluge algorithm noising method, ruin and recreate, hybrid methods, and neural networks.

Chapter 6 introduces threshold accepting in rigorous detail. Chapter seven uses two different concepts to compare the performance of algorithms: a hypothetical case where the number of iterations is allowed to tend to infinity, and a second case where the ranking of different algorithms consists in comparing their performance for a specific problem and a given amount of computational resources. Chapter 8 discusses methods of tuning threshold accepting, while chapter 9 presents three practical steps in threshold accepting implementation: formulating the problem, analysis of the search space of potential solutions, and finally, the choice of a threshold sequence in which the parameters of the algorithm must be chosen. The third part of the text varies from the first half of the book by presenting a practical and successful implementation of threshold accepting. There is a chapter on each of experimental design, multivariate model identification, optimal aggregation, censored quantile-regression, and continuous global optimization. The variety of applications will appeal to almost every reader. Part 4 of the text, conclusion and outlook, is a review of the material covered as well as a value judgment for future research in threshold accepting. This book is well written, with thorough coverage. The serious statistician and the research economist will gain much from the material.