In this work, we present a dispersive finite-difference time domain (FDTD) algorithm using a 4-pole complex rational function (CRF). For the sake of a better curve fitting of the 4pole CRF dispersion model, we use a particle swarm optimization technique. We also discuss an efficient memory storage strategy by using a state-space approach. The numerical aspects of 4pole CRF-FDTD -the numerical accuracy and the numerical stability -are investigated in detail. Numerical examples are used to validate 4-pole CRF-FDTD and numerical stability issues are discussed in detail. We also discuss the computational accuracy and the computational efficiency of an arbitrary N -pole CRF-FDTD.
Index Terms-Dispersive media, FDTD methods, Numerical stability.0018-926X (c)