Two-person zero-sum games (Q382967)

This book is devoted to antagonistic games, they are also equivalently defined as two-person zero-sum games with exposition of various illustrative examples for their subclasses. The book contains the preface(s), nine chapters, three appendices, references, and an index. Chapter 1 gives a brief introduction to the decision making of one person based on utility and maxmin concepts. Chapter 2 describes maxmin and minmax strategies for two players. Chapter 3 presents some basic concepts of matrix games, including the saddle point theorem, dominance, solution methods for particular cases, relations to linear programming, and the fictitious play algorithm. Various classes of Markovian (multi-stage) games are given in Chapter 4. Chapter 5 describes games with infinite number of strategies, this topic is continued in Chapter 6 devoted to resource distribution (Blotto) games. Chapter 7 deals with some antagonistic problems that occur on networks; e.g. they are related to shortest path, maximal flow, and network detection problems under competition. Search games are described in Chapter 8. Some additional examples of antagonistic games, such as ratio and signaling ones are given in Chapter 9. Two appendices touch briefly linear programming and convexity concepts. The third appendix contains solutions to exercises. The book's presentation is given at the student level, contains many significant examples and exercises, is furnished by an electronic supplementary workbook, so that it can be used as a textbook. Nevertheless, the author gives only formulations for many basic results or refers to other books, proofs are given only for some selected results; say, for stochastic games. It would be useful for students to read first a book on game theory with a more strict exposition; e.g., [\textit{G. Owen}, Game theory. 3rd ed. San Diego: Academic Press (1995; Zbl 1284.91004)]. A minor comment: The vector \(s\) defining a tangent hyperplane to a convex function graph is not unique if the function is not differentiable. Hence it is a subgradient, rather than the gradient.