Coping with numerical complexity in computational electromagnetics

The paper presents a selection of techniques that are currently being investigated at the Technical University of Gdansk and can be used to accelerate the computation of high and low frequency electromagnetic fields. The techniques discussed represent three different strategies leading to the reduction of solution times in frequency and time domain simulations. The first strategy uses new numerical techniques which offer near optimal numerical cost of solving large sparse linear systems of equations resulting from the discretization of partial differential equations. The second strategy involves taking advantage of the recent developments in computer hardware namely the SMP and scalable parallel architectures. Finally, the third one aims at finding a better field representation or problem parameterization and results in algorithms which are particularly useful in problems requiring repetitive calculations that are often encountered in computational electromagnetics.