Comparing the sparse iterative method (SIM) with the banded Jacobi and conjugate gradient techniques

The sparse iterative method (SIM) has been developed by Fourie and Nitch (see IEEE AP-S International Symposium, 1994). This paper demonstrates that the SIM is more general, and executes faster than either the banded Jacobi or conjugate gradient techniques. First, the banded Jacobi solution method was implemented in an object oriented C++ program based on the well known NEC2 program and results were compared to the SIM method for a grid example obtained from the GEMACS documentation of Kadlec and Coffey (1983). Secondly, the SIM was compared to the conjugated gradient method by simulating a generic missile case for which previous results were obtained by Davidson (see PHD thesis, University of Stellenbosch, 1991). The SIM is shown to be computationally faster than both of these methods. The improved speed of the SIM, combined with the very general nature of this solution technique, makes it an attractive choice for the solution of MoM problems.<>