Author_Institution :
EMC Lab., Penn State Univ., University Park, PA, USA
Abstract :
Summary form only given. Commercial EM simulation softwares are invaluable tools without which the engineers, who are engaged in designing antennas, microwave circuits, RF packages, and systems utilizing these devices, simply cannot survive. This is because the number of real-world problems that we can solve by using analytical techniques is very limited indeed. To be sure, the commercial software world is very crowded, and the competition is stiff. However, most of these software modules are based upon algorithms that have their unique flavor, and this not only serves to give them a unique identity, but also makes them especially suited for a class of problems for which they offer a distinct advantage. As an example, a computer code based upon the Method of Moments (MoM) is best suited for perfectly conducting objects, for which it is typically more efficient in terms of CPU time and memory than its counterparts, which utilize Finite Methods such as the Finite Element Method (FEM) or the Finite Difference Time Domain (FDTD) technique. On the other hand, the Finite Methods can handle arbitrary geometries and material parameters with greater ease than can the MoM. Similarly, certain computer codes are strictly applicable in the frequency domain, whereas others are configured only for time domain analysis. Of course the use of Fourier transformation of the results, obtained either in the time or in the frequency domain, enables one to go back and forth from one to the other. Nonetheless, depending upon the nature of the problem, the frequency domain approach may be better suited than its time domain counterpart, or vice versa. In addition, it is not uncommon for these software vendors to offer multiple choices to the user, and let them decide which one of these is best suited for the problem at hand. It is sometimes interesting to carry out a study of the comparative performance of the various codes and benchmark the figures for their memory usage, as well as CPU time- - , to solve the same test problem. In this paper we will begin by presenting a Table that summarizes the comparative performances of a number of preeminent commercial EM solvers that are on the market. Such benchmarking can provide valuable information to the user, who is engaged in the process of making decisions regarding the choice of the software for the type of problem that he/she wants to solve. Next we will review some of the recent developments in the CEM world, such as the Fast Multipole Method (FMM), Discontinuous Galerkin (DG) method and the Dipole Moment (DM) method, and examine the issue of integrating these algorithms in the commercial codes. To develop a better understanding of this process and to appreciate why sometimes there is a relatively long time lag before such integrations are in place, we review briefly some of the intricacies of the commercial code development world, to get a feel of ´what makes it tick.´ Finally, we will propose a paradigm using which the universities, where much of the cutting edge research is carried out before it permeates into the commercial world, can accelerate the process of integration of the algorithms they develop. When implemented, such an integration would certainly be of great benefit to the user, who is often frustrated by the limitations of the commercial codes for certain types of problems that have been cropping up lately. Some examples of such problems are nanowires and nanotubes that are finding increasing number of applications, biosensors mounted on complex platforms, and the Square Kilometer Array, the chief protagonist of the 21st century antenna world. The design engineers typically find that the existing commercial codes have severe limitations when handling some of these problems, and although there is a considerable amount of activity in the academic world attempting to discover techniques for solving them, the road to transition them into the commercial world is not always paved very smoothly
