NEW DISTRIBUTED IMPLEMENTATION OF THE FDTD METHOD

Citation
N. Takada et al., NEW DISTRIBUTED IMPLEMENTATION OF THE FDTD METHOD, Electronics & communications in Japan. Part 2, Electronics, 80(5), 1997, pp. 8-16
Citations number
16
ISSN journal
8756663X
Volume
80
Issue
5
Year of publication
1997
Pages
8 - 16
Database
ISI
SICI code
8756-663X(1997)80:5<8:NDIOTF>2.0.ZU;2-X
Abstract
The FDTD method is a numerical technique that can be used to solve ele ctromagnetic boundary value problems in the time domain. This method i s excellent in computational efficiency, practicality, and simplicity. In recent years, we have frequently used this method in order to anal yze antennas, electromagnetic scattering problems, and so on. However, the electromagnetic boundary value problems which we can analyze by t his method are limited because computer memory is limited. In order to overcome this important problem, Rodohan and colleagues recently deve loped the distributed FDTD method. In this method, the necessary compu ter memory is provided by several computers and we can analyze problem s which need a large computer memory space. In this paper, we first pr opose a new distributed FDTD method in which the number of communicati ons and the communication costs except for absorbing boundary conditio ns are half those of the Rodohan method and the absorbing boundary con ditions are highly accurate. Second, when we calculated the scattering of a plane wave by a square conducting cylinder with our method and t he conventional FDTD method, the results were in agreement. We confirm ed the high speed of calculations with our method, Finally, we were ab le to show the validity and the effect of our method. (C) 1997 Scripta Technica, Inc.