A reflection on the implicitly restarted Arnoldi method for computing eigenvalues near a vertical line (Q413531): Difference between revisions

The authors analyze a method for computing eigenvalues of a large sparse generalized eigenvalue problem \(Ax=\lambda Bx\), closest to the imaginary axis. Their method is based on inverse subspace iteration on the Lyapunov problem. They also show the link between inverse subspace iteration and implicitly restarted Arnoldi process through the extra projection step which corresponds to the decrease of the Krylov subspace dimension by executing \(QR\)-steps.