scientific article; zbMATH DE number 2035082

From MaRDI portal
Publication:4445267





The book is a revised and updated version of its predecessor which appeared in 1992 with \textit{E. R. Hansen} as the only author. Therefore, most of the review in Zbl 0762.90069 is still up-to-date and will not be repeated here. The authors apply interval methods in order to verify solutions of systems of nonlinear equations and of global optimization problems with and without constraints. While in the first edition interval Newton methods were the most important part of the global optimization algorithms this place is now shared with consistency methods for increasing the speed of convergence of algorithms. Both methods can prove existence and uniqueness in an analytically rigorous way. Many older algorithms were now improved. A set theoretic foundation for computing with intervals guarantees that there are no longer undefined interval operations or functions in the algorithms which were developed by the authors and their collaborators. The concept is based on the set of extended real numbers. Numerical examples illustrate its efficiency. Chapter 1 introduces into the subject. In particular, it motivates the application of interval analysis and lists nowadays software packages which support the computation with intervals. Chapter 2 defines interval numbers and arithmetic including dependencies of interval operands and extended interval arithmetic. Functions of intervals continue these basics in Chapter 3. So-called closed interval systems form an essential and new chapter. They use an extended arithmetic in which division by zero is defined. Thus they enlarge the applicability of existing algorithms and admit generalizations. Chapter 5 deals with systems of linear equations -- also with intervals as input data. Inequalities are considered in Chapter 6. Taylor expansion and slope expansion in the one- and multidimensional case as well as automatic differentiation form the contents of the subsequent chapter. Quadratic equations and quadratic inequalities precede general nonlinear equations in one variable in Chapter 9. Consistencies for equations were studied in Chapter 10, systems of nonlinear equations in Chapter 11. Here Newton's method for interval vectors is discussed in detail. Various optimization problems such as unconstrained problems, constrained problems, inequality constrained optimization and equality constrained optimization are centered in the next four chapters and cover more than a quarter of the almost 500 pages. Box splitting, John conditions, line search and nondifferentiable functions are some of the keywords which occur. Inequality constraints and equality constraints are allowed simultaneously in Chapter 16. Perturbed problems and sensitivity analysis are considered in Chapter 17 followed by remarks on nondifferentiable functions and integer problems in the final chapter which acts as supplement. An extensive bibliography with 215 entries concludes a well written and interesting monography.




Cited in
(only showing first 100 items - show all)








This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4445267)