Numerical linear algebra and its applications (Q2901229)

The book provides a concise introduction to numerical linear algebra with applications to solving initial value problems using boundary value methods and their preconditioning.NEWLINENEWLINEThe motivating introduction in Chapter 1 is followed by the overview of basic direct methods (LU factorisation with pivoting and Cholesky factorisation) for solving nonsingular systems of linear algebraic equations in Chapter 2.NEWLINENEWLINEChapter 3 is devoted to the elements of perturbation and error analysis of numerical algorithms in finite precision arithmetic. Particular attention is payed to the perturbation analysis of linear algebraic systems, the model of finite precision arithmetic, and the application to the backward error analysis of LU factorisation.NEWLINENEWLINEIn Chapter 4 the authors address direct solution methods for linear least squares problems based on QR factorisation.NEWLINENEWLINEIterative methods for solving linear systems are discussed in Chapters 5 and 6 devoted, respectively, to classical and Krylov subspace methods. Chapter 5 covers the basic iterative methods including the Jacobi, Gauss-Seidel, and SOR methods together with their convergence analysis. In Chapter 6 the authors consider the steepest descent method and two globally optimal Krylov subspace methods: namely, the conjugate gradient method for solving symmetric positive definite systems and the generalised minimum residual method for general ones. Preconditioning methods are discussed as well: the basic methods based on (block) diagonal splitting and incomplete factorisations including preconditioners for circulant matrices.NEWLINENEWLINEChapters 7 and 8 contain elements of solving nonsymmetric and symmetric eigenvalue problems. The authors describe some popular methods, in particular, the power and QR methods for general problems, and the Jacobi, bisection, and divide-and-conquer methods for symmetric ones.NEWLINENEWLINEChapter 9 is devoted to applications of methods of numerical linear algebra for solving initial value problems by boundary value methods. Particular attention is payed to the choice of the preconditioner for various types of ordinary differential equations.