E§et des Erreurs díarrondi sur les systËmes linÈaires

Abstract

In this memory we have studied the numerical methods for the solving linear systems squares. These methods are direct and iterative (indirect). Direct methods generally used when n < 100 and they give an exact solution. And when n is large the solution of a system of order n by direct methods becomes quite complicated and often too much memory space so it is necessary to use the iterative methods because they are more efficient and more effective. The general principle of these last methods to solve the linear system is to generate a sequence of vectors which converges towards the exact solution.