Abstract

Recently, there has been considerable interest in optical processors for performing linear algebra operations to exploit both the high degree of parallelism and speed associated with optical processing. It is important to understand, and wherever possible compensate for, the limitations imposed by sources of deterministic and random error. In this paper we consider the effects of system noise sources (e.g., light sources, modulators, and detectors) on the accuracies of basic linear algebra operations. The results of digital simulations performed to evaluate the accuracies of various optical linear algebra processors are also presented.

© 1986 Optical Society of America