Extremely low-precision integer cellular array algorithm for computational electromagnetics

Abstract

We present an algorithm for computational electromagnetics using extremely low-precision integer variables (only a single or a few bits) operating on a cellular array topology. We show, through the enforcement of field and energy conservation rules, that lossless linear wave behavior can be simulated enabling solution of the two-dimensional Maxwell's equations. Since the algorithm operates using only a few bits per cell and does not require floating-point operations, it is ideally suited for implementation on a fine grain computing engine.