A class of filtering superoptimal preconditioners for highly ill-conditioned linear systems (Q1860947)

The author presents a new scheme for obtaining highly stable preconditioners for very ill-conditioned linear systems and demonstrates it to systems with Toeplitz coefficient matrices. His aim is to develop regularizing preconditioners that inherit the regularizing behaviour of the superoptimal one. Review of some basic concepts and properties of the optimal and superoptimal preconditioners is presented. A circulant family of filtering superoptimal preconditioners for Toeplitz highly ill-conditioned systems is developed and their eigenvalues distribution is studied and analyzed. Results from some promising numerical tests are presented and discussed. The author conclude that the filtering preconditioners could be used to solve very ill-conditioned problems with large error on the data.