The orthogonality of the eigenvector basis leads to a signi. Principles and analysis this book offers a detailed treatment of the mathematical theory of krylov subspace methods with focus on solving systems of linear. We refer the reader to 69,80 for a general introduction to krylov subspace methods and to 74 for a recent overview on krylov subspace methods. The krylov subspace k m generated by a and u is span u au a 2 u a m. Describes the principles and history behind the use of krylov subspace methods in science and engineering. This book offers a detailed treatment of the mathematical theory of krylov subspace methods with focus on solving systems of linear algebraic equations. Recycling krylov subspace methods for sequences of linear. Starting from the idea of projections, krylov subspace methods are characterised by their orthogonality and minimisation properties. When implementing a krylov subspace method, such as the one suggested by 1. The outcome of the analysis is very practical and indicates what can and cannot be expected from the use of krylov subspace methods, challenging some common assumptions and justifications of standard approaches. Geometric aspects in the theory of krylov subspace methods. Theory of inexact krylov subspace methods and applications.
Sadok dedicated to gerard meurant at the occasion of his 60th birthday. Anastasia filimon eth zurich krylov subspace iteration methods 290508 4 24. Krylov subspace methods for topology optimization on adaptive meshes. Foundations of functional analysis, including continuous spectrum, resolution of unity.
It is of dimension m if the vectors are linearly independent. However, the m matrices used in classical stationary. As i understand it, there are two major categories of iterative methods for solving linear systems of equations. The outcome of the analysis is very practical and indicates what can and cannot be expected from the use of krylov subspace methods challenging some common assumptions and justifications of standard approaches. In our approach krylov subspace methods are divided into three classes. We introduce a framework of krylov subspace methods that satisfy a galerkin condition. Krylov subspace methods are strongly related to polynomial spaces and their convergence analysis can often be naturally derived from approximation theory. Projections onto highly nonlinear krylov subspaces. Krylov subspace methods for the solution of large systems. Convergence analysis of krylov subspace methods tu berlin. Krylov subspace methods for functions of fractional di. An analysis of acceleration strategies including augmentation for minimal residual methods was given by saad 45 and for restarted methods by eiermann.
The presentation will show several examples of preconditioned conjugate gradient type methods and examples from air pollution problems will be used to make comparisons on the efficiency of such methods. What is the principle behind the convergence of krylov. Projections onto highly nonlinear krylov subspaces can be linked with. However, it is difficult to understand mathematical principles behind these methods. Topology optimization is a powerful tool for global and multiscale design of structures, microstructures, and materials. One idea is to use the quasiminimal residual qmr principle to obtain smoothed. Lanczos method krylov subspace methods suppose that a 2r n is large, sparse and symmetric, and assume that. Among such methods are different kinds of krylov subspace methods. Convergence analysis of krylov subspace methods liesen. Two broad choices for l m give rise to the bestknown techniques.
Krylov subspace methods from the analytic, application and. Forbetterbehavior construct an orthonormal basis for k ka,b starting with v. Msc 2000 15a06, 65f10, 41a10 one of the most powerful tools for solving large and sparse systems of linear algebraic equations is a class of iterative methods called krylov subspace methods. To get some understanding when and why things work, and when and why they do not. Key words krylov subspace methods, convergence analysis. In linear algebra, the orderr krylov subspace generated by an nbyn matrix a and a vector b of dimension n is the linear subspace spanned by the images of b under the first r powers of a starting from, that is. In this chapter we investigate krylov subspace methods which build up krylov sub. Projections onto highly nonlinear krylov subspaces can be linked with the underlying problem. Krylov subspace methods a more readable reference is the book by lloyd n. The mathematical theory of krylov subspace methods with a focus on solving systems of linear algebraic equations is given a detailed treatment in this principles based book. Thus, readers equipped with a basic knowledge of linear algebra should be able to understand these methods. In section 3 we present a convergence analysis for the sikm. In computational mathematics, an iterative method is a mathematical procedure that uses an initial guess to generate a sequence of improving approximate solutions for a class of problems, in which the nth approximation is derived from the previous ones.
The krylov subspace methods 8,9 have been developed and perfected, starting. It is necessary to use linear systems solvers that parallellize well. A specific implementation of an iterative method, including the termination criteria, is an algorithm of the iterative method. In section 4 we discuss some cases related to problems like 1. Krylovsubspace matrix solution methods jacob white thanks to deepak ramaswamy, michal rewienski, and karen veroy. Stationary methods jacobi, gaussseidel, sor, multigrid krylov subspace metho. Projections onto highly nonlinear krylov subspaces can be linked with the underlying problem of moments, and therefore krylov subspace methods can be viewed as matching moments model reduction. For these reasons we restrict our analysis to bcgstab and tfqmr. Principles and analysis this book offers a detailed treatment of the mathematical theory of krylov subspace methods. In the first part of the article, krylov methods are discussed in detail. Principles and analysis jorg liesen, zdenek strakos the mathematical theory of krylov subspace methods with a focus on solving systems of linear algebraic equations is given a detailed treatment in this principles based book.
Then we do an analysis that allows us to pick out a promising mdimensional subspace of k. The history of the employed techniques goes back to the mid 1990s when restarted and nested krylov subspace methods have been proposed, e. This volume describes the principles and history behind the use of krylov subspace methods in science and engineering. It includes the families of orthogonal residual or and minimal residual mr methods. We begin by generating a krylov subspace k ka,x of dimension k, where k is somewhat bigger than m, e. We pick mat least as big as mand preferably a bit bigger, e.
562 1435 1141 1407 1220 1427 453 719 453 1084 1361 902 902 258 522 429 323 721 1374 578 420 606 467 1450 784 395 92 763 285 1162 1154 117 1015 245