Title Details: | |
Numerical Analysis |
|
Authors: |
Plexousakis, Michalis Chatzipantelidis, Panagiotis |
Subject: | MATHEMATICS AND COMPUTER SCIENCE > MATHEMATICS > NUMERICAL ANALYSIS > NUMERICAL LINEAR ALGEBRA MATHEMATICS AND COMPUTER SCIENCE > MATHEMATICS > NUMERICAL ANALYSIS > MATHEMATICS AND COMPUTER SCIENCE > MATHEMATICS > ORDINARY DIFFERENTIAL EQUATIONS MATHEMATICS AND COMPUTER SCIENCE > MATHEMATICS > CALCULUS OF VARIATIONS AND OPTIMAL CONTROL; OPTIMIZATION > NUMERICAL METHODS MATHEMATICS AND COMPUTER SCIENCE > MATHEMATICS > LINEAR AND MULTILINEAR ALGEBRA; MATRIX THEORY > BASIC LINEAR ALGEBRA |
Keywords: |
Direct methods
Iterative methods SVD factorization QR decomposition Least squares Power method Francis QR step LU decomposition Cholesky decomposition Newton’s method Descent methods Steepest descent Conjugate gradient method Krylov subspace methods Single-step methods Runge-Kutta methods Multistep methods Methods for stiff problems |
Description: | |
Abstract: |
The book consists of two parts. In Part A we review some basic concepts from linear algebra, as these form the basis of what we will be looking at. We study the numerical solution of a linear system with a square matrix using direct methods, which compute the exact solution in a predetermined number of steps, known in advance. We also study iterative methods, where we construct a sequence of vectors that converge to the solution of the linear system. In addition, we consider linear systems with a non-square matrix, where we search for vectors that minimise an appropriate function, like e.g., a residual. We then address the problem of locating the eigenvalues of a matrix and consider methods for solving nonlinear equations and systems. Part B deals with the numerical solution of initial value problems (I.V.P.) for first-order systems of ordinary differential equations. The simplest numerical method for solving the initial value problem (I.V.P.) is Euler's method. We also consider other single-step methods, and more generally the Runge-Kutta family of methods. We also study the linear multistep methods. Their implementation is economical, and this is why they appeared and were applied before the advent of computers. We study these methods' stability, consistency, accuracy, and order of convergence. The implicit Runge-Kutta methods are more demanding to implement than both explicit Runge-Kutta methods and multistep methods. However, they have higher accuracy and excellent stability properties. We also consider stiff linear differential equations. Some numerical methods are unstable and require a small step size to approximate well the exact solution of these problems. This stability property associated with the approximate solution of a method for a stiff differential equation problem is called absolute stability.
|
Linguistic Editors: |
Aslani, Kyriaki Evangelia |
Technical Editors: |
Plexousakis, Michalis Chatzipantelidis, Panagiotis |
Graphic Editors: |
Plexousakis, Michalis Chatzipantelidis, Panagiotis |
Type: |
Postgraduate textbook |
Creation Date: | 04-12-2023 |
Item Details: | |
ISBN |
978-618-228-157-4 |
License: |
Attribution - NonCommercial - ShareAlike 4.0 International (CC BY-NC-SA 4.0) |
DOI | http://dx.doi.org/10.57713/kallipos-395 |
Handle | http://hdl.handle.net/11419/11684 |
Bibliographic Reference: | Plexousakis, M., & Chatzipantelidis, P. (2023). Numerical Analysis [Postgraduate textbook]. Kallipos, Open Academic Editions. https://dx.doi.org/10.57713/kallipos-395 |
Language: |
Greek |
Consists of: |
1. Basic Notions of Linear Algebra 2. Solving Linear Systems 3. Iterative Methods for Linear Systems 4. Least squares 5. Eigenvalue problems 6. Numerical solution of nonlinear equations 7. Numerical solution of nonlinear systems 8. Initial Value Problems 9. One Step Methods I: Euler Method 10. One Step Methods II: Runge-Kutta Methods 11. Multistep Methods 12. Stability of Stiff Differential Equations 13. Solutions of selected exercises |
Number of pages |
272 |
Publication Origin: |
Kallipos, Open Academic Editions |
You can also view | |
User comments | |
There are no published comments available! | |