Planar-CG Methods and Matrix Tridiagonalization in Large Scale Unconstrained Optimization

摘要: In this paper we aim at carrying out and describing some issues for real eigenvalue computation via iterative methods. More specifically work new techniques iteratively developing specific tridiagonalizations of a symmetric indefinite matrix A ∈ R n × , by means suitable Krylov subspace algorithms defined in [16], [26]. These schemes represent extensions the well known Conjugate Gradient (CG) method to case. We briefly recall these suggest comparison with [22], along discussion on practical application proposed results computation. Furthermore, focus motivating fruitful use ensuring convergence second order points, within an optimization framework.