Passive Reduced-Order Modeling via Krylov-Subspace Methods

In recent years, Krylov-subspace methods have become popular tools for computing reduced-order models of large-scale time-invariant linear dynamical systems. These techniques produce in a Pade sense optimal reduced-order models. However, when applied to passive systems, the Pade models do not preserve passivity in general. In this paper, we describe a Krylov-subspace technique that generates passive reduced- order models. Results of numerical experiments with a passive system arising in VLSI circuit simulation are reported.