No label defined (Q4515979)

The title of this book is far too wide and general since it deals almost exclusively (though in great depth) with the convergence properties of Newton-like iterative methods to solve nonlinear problems in abstract spaces. The purpose is indeed to create a general framework for the study of optimal iterative procedures for problems that are solved approximately. It is largely a one-man-book that summarizes the research of the author over many years. Basically, this is a book on theoretical numerical analysis where the typical implementation problems (roundoff error, efficient storage and handling) are not even mentioned. The author expects that not only mathematics students will read this book but readers may also come from mathematical economics, engineering, physics and applied science. However, the usual undergraduate courses in linear algebra and calculus are prerequisites to this book but not sufficient. The book assumes a working knowledge of Hilbert spaces, Banach spaces and event locally convex topological vector spaces. The pace is set in Chapter 1 which treats divided differences of nonlinear operators, Fréchet derivatives and the connections between them. In Chapter 2 Lipschitz conditions are discussed in spaces of continuous functions and in \(L_p\) and \(L_\infty\) spaces. In Chapters 3-8 convergence and optimal error estimates are proved for many variations of Newton-like methods. In Chapters 3-6 these are based on hypotheses of the Lipschitz-type on the first Fréchet derivative; in Chapters 7-8 hypotheses on the second Fréchet-derivative are used. Among special topics treated in the book we like to mention the mesh independence principle (\S 6.1) and convergence of Newton's method for polynomial equations (\S 7.5). Appendix A (Numerical algorithms) contains pseudocodes for several methods to find local zeros of nonlinear functions, quadrature methods and integration of initial value problems for systems of ordinary differential equations, solutions of linear systems, finite difference and finite element methods for boundary value problems. The book ends with an extensive list of references, a glossary of symbols and an index. Unfortunately the index is too short. For example, the authors claims to include applications from astrophysics, economics and mechanics but the index does not point to any of these applications (nor can they be found easily in the table of contents). To learn about the mesh independence principle the reader is referred to page 269 while the subject is already mentioned on page 247. The book is well-written and accessible to people with a good background in calculus and functional analysis. It can be recommended as a reference work for researchers with an interest in the solution of nonlinear problems. It would also be suitable for an advanced numerical analysis course.