Iterative solution of large sparse systems of equations (Q289671): Difference between revisions

For the first German edition see [Iterative Lösung großer schwachbesetzter Gleichungssysteme. (Iterative solution of large sparse systems of equations). Stuttgart: B. G. Teubner (1991; Zbl 0729.65018)], for the first English edition see [New York, NY: Springer-Verlag (1994; Zbl 0789.65017)]. Jointly with its second English edition the present book celebrates its \(25^{\mathrm{th}}\) anniversary. Invariably the comment from my review of the first edition applies that ``the author presents the actual stand of the art in iteratively solving algebraic sytems in both treating the practical numerical importance and efficiency of the methods as well as their mathematical analysis in full and exhaustive generality''. One reason for the existence of such an excellent monography can be understood by a look to the 417 references at the end of the book among which fifty alone point to the author, the most recent ones from the year 2016. The titles of the 14 chapters of the book are listed under the keywords. There is a 82 pages long appendix with facts from linear algebra, hierarchical matrices and the Galerkin discretisation of elliptic PDEs, where the latter two are new in this edition. Necessarily, compared to the second edition, the book's page number has increased, by eighty, due to new material included. There are three new chapters. A new chapter in Part I (Linear iterations) is entitled ``Algebra of linear iterations''. Here new iterative schemes are built in a systematic way from simpler units exploiting the fact that the set of consistent linear iterations forms an algebra allowing for several operations (transposition, damping, addition, multiplication, left-, right- and two-sided transformations). Application of these principles leads e.g. to the accelerated SOR, SSOR, Kaczmarz and Cimmoni iteration. Part II is still devoted to ``Semi-iterations and Krylov methods''. In Part III (Special iterations) there are two new chapters. One is entitled ``\(\mathcal{H}\)-LU iteration'', using the author's hierarchical matrices (which are the topic of the new Appendix D). The second new chapter is about tensor-based iterative methods. Everybody who wants to solve algebraic equations numerically or is working in one way or the other, theoretically or practically, with iterative methods will find no way around paying its tribute, directly or indirectly, to this impressing book. Compared with the black-red covered first edition by Teubner in 1991 with its slightly smaller page layout and a size of ``only'' 382 pages, actually there is a considerable increase in content and corresponding size. But also the price has significantly increased, the original one was 42 DM (about 21 Euros) while the new edition costs 107 Euros.