Richard Fateman wrote: > I think the smart money on this computation is heading towards finding the > eigenvalues of the companion matrix. Is it? I thought Jenkins-Traub was still pretty much state of the art. Ray