This paper addresses the problem of modeling systems from measurements of their frequency response. For multiport devices, currently available techniques are expensive. We propose a new approach which is based on a system-theoretic tool, the Loewner matrix pencil constructed in the context of tangential interpolation. Several implementations are presented. They are fast, accurate, they build low order models and are especially designed for a large number of terminals. Moreover, they identify the underlying system, rather than merely fitting the measurements. The numerical results show that our algorithms yield smaller models in less time, when compared to vector fitting.

In this paper, the dominant spectral-zero method (dominant SZM) is presented, a new passivity-preserving model-reduction method for circuit simulation. Passivity is guaranteed via spectral-zero interpolation, and a dominance criterion is proposed for selecting spectral zeros. Dominant SZM is implemented as an iterative eigenvalue-approximation problem using the subspace-accelerated dominant-pole algorithm. Passive circuits are reduced automatically irrespective of how the original system equations are formulated (e.g., circuit models containing controlled sources or susceptance elements). Dominant SZM gives comparable and often more accurate reduced models than known techniques such as PRIMA, modal approximation, or positive real balanced truncation.

A rational Arnoldi method for passivity-preserving model-order reduction (MOR) with implicit multi-point moment matching for systems with frequency-dependent interconnects is described. The structure H(s) = sE - A - K√f, which arises from frequency-dependent effects in high speed interconnects, is preserved by the proposed MOR technique. Moment matching using congruence transforms and based on two types of moments that are derivatives of the transfer function w.r.t s and √f is described. Simulation results show that the proposed approach can significantly reduce the complexity of systems with frequency-dependent elements, while retaining high accuracy in comparison to the original system in both the time and frequency domains.