On numerical analysis of conjugate gradient method (Q1328993): Difference between revisions

The author spoils several pages by telling us what we should know since our first numerical analysis course: successive overrelaxation (SOR) methods, Banach's fixed point theorem and error analysis conclusions from it, the conjugate gradient method. Stopping criteria are developed using the well-known error estimates mentioned. More than half of the paper is devoted to numerical experiments. They compare line SOR with conjugate gradient methods with different preconditioning. The results show very different behaviour of the methods for standard and non-standard problems with discontinuous coefficients.