Extended Krylov subspace for parameter dependent systems

The Extended Krylov subspace has recently received considerable attention as a powerful tool for matrix function evaluations and other problems involving large matrices. In this paper we show that this space has a great potential within projection-type methods for effectively solving several other important large-scale algebraic problems: we focus on the solution of shifted systems and of more general parameter-dependent matrix equations, and on the approximation of the transfer function by projection. Numerical experiments stemming from real applications show the effectiveness of the approach.