An adaptive method for solving nonsymmetric linear systems involving applications of SCPACK

AbstractConsider the following (large) system of linear equations: X = TX + c, where it is assumed that T is an n × n real matrix with 1 excluded from the spectrum of T, and that c is a real vector of dimension n. This paper uses results from function theory, approximation theory and conformal mapping theory, and constructs an adaptive numerical method to solve the system above with the aid of SCPACK. This adaptive method periodically estimates the eigenvalue set to obtain near-asymptotically optimal rates of convergence of the associated (k, l)-step semi-iterative methods