Abstract

Optical computing trails electronic computing significantly in mass producibility yet many optical computer schemes ambitiously assure what modern VLSI seldom assumes: 100% usefulness of each of many, say 512×512, individual components on the same chip. On the assumption that perfection is costly at best and impossible at worst, we seek ways to overcome or bypass faults in 2D optical component arrays. For definiteness, we will consider light addressed spatial light modulators (SLMs). All workers in the field of optical processors are familiar with almost-adequate SLMs. Can we collect the "good" parts together into a full array? Holography seems to be the obvious answer, but how must it work in detail.

© 1986 Optical Society of America