A comparison and evaluation of the PML and COM mesh truncation techniques for FDTD simulation

Open region FDTD simulation requires terminating the computational domain with an operator or with an absorbing material that simulates the effect of an unbounded medium. Two techniques for mesh termination have emerged that succeed in bringing the artificial reflections generated from the terminal boundaries of FDTD space to unprecedented low levels. These two techniques are the perfectly matched layer (PML) introduced by Berenger (1994) and the complementary operators method (COM) introduced by Ramahi (1995). No direct comparison between the two techniques has yet been published. In this paper, an evaluation and comparison of these two techniques is presented that is based on a numerical experiment which has been discussed in the literature (Ramahi). Here, the primary factors in the comparison are: accuracy, memory overhead and operations count. Although this example gives numerical data, it is still intended to give primarily a qualitative perspective on the relative performance of the two techniques. The example demonstrates that COM provides comparable accuracy to that obtained by an 18-layer PML, while requiring less memory and fewer operations.