Passive reduced-order modeling of electromagnetic systems

Rapid electromagnetic field simulation is now being recognized as a critica l component of next-generation computer-aided design frameworks that will b e used for performance evaluation and design of the information processing and communication systems of the 21st century. Except for very simple syste ms, computer simulation of electromagnetic interactions is hindered by the very large number of degrees of freedom involved in the discrete model. One potentially useful approach to overcoming this computational bottleneck is model order reduction, where parts of the electromagnetic model are replac ed by models of substantially lower order, yet capable of capturing the ele ctromagnetic behavior of the original subsystems with sufficient engineerin g accuracy. Possible approaches to the development of such model order redu ction techniques are the subject of this paper. The potential applications and the benefits from such electromagnetic model order reduction are discus sed. The important issue of passivity of the generated reduced model is con sidered, and a set of constraints on the state representation of the discre te electromagnetic boundary-value problem are identified for the reduced or der model to be passive. The paper concludes with a presentation of several numerical examples from the application of model order reduction technique s to the analysis of electromagnetic waveguides and antennas. (C) 1999 Publ ished by Elsevier Science S.A. All rights reserved.