Large-scale matrix computations in control (Q1294573)

This paper gives a brief overview of some of the recent Krylov subspace techniques for large-scale matrix computations and their application to the solutions of large-scale control problems. After an introduction to Arnoldi and Lanczos methods, the author considers the application of an Arnoldi method to the rank-one Lyapunov equation and to the Sylvester-observer equation (single-output case). Other applications are a projection algorithm for the partial pole-assignment problem and model reduction via the Lanczos method.