Abstract

It has been widely assumed that, provided the hardware problems can be overcome, there are significant intrinsic speed advantages to optical approaches to certain computational problems. Methods for achieving high accuracy using optical analog components to execute digital algorithms⁽¹⁾ have further increased interest in optical computing. It is the purpose of this paper to examine some of the factors which determine the speed of integrated optical computational devices, and to present the results of an initial attempt to compare the performance of optical and electronic approaches to matrix-vector multiplication (MVM).

© 1985 Optical Society of America