Optical computing is an inherently multidisciplinary subject whose study routinely involves a spectrum of expertise that threads optical physics, materials science, optical engineering, electrical engineering, computer architecture, computer programming, and computer theory. Applying ideas from theoretical computer science, such as analysis of algorithms and computational complexity, enables us to place optical computing in a framework where we can try to answer a number of important questions. For example,which problems are optical computers suitable for solving? Also, how does the resource usage on optical computers compare with more standard (e.g. digital electronic) architectures? The physical principles behind some efficiencies in optical computing are outlined here.
1. Fan-in efficiency
Kirchoff’s Law is well understood in analog electronics as a natural and constant time means of summing the current at the intersection of an arbitrary number of wires. In optics, the same thing is possible by directing several light beams towards a point detector with a linear response to incident light. Such an optical arrangement sums n non-negative integers in O(1) addition steps. On a model of a sequential digital electronic computer this would require n − 1 addition operations and even many typical (bounded fan-in)parallel models, with n or more processors, take O(log n) time steps. Tasks that rely on scalar summation operations (such as matrix multiplication) would benefit greatly from an optical implementation of the scalar sum operation. Similarly, O(1) multiplication and O(1) convolution operations can be realised optically. Very recently, an optics-based digital signal processing platform has been marketed that claims digital processing speeds of tera (10^12) operations per second.
2. Efficiency in interconnection complexity
As optical pathways can cross in free space without measurable effect on the information in either channel, high interconnection densities are possible with optics. Architectures with highly parallel many-to-many interconnections between parallel surfaces have already been proposed for common tasks such as sorting. Currently, intrachip, inter-chip, and inter-board connections are being investigated for manufacturing feasibility.
3. Energy efficiency
Electrical wires suffer from induced noise and heat, which increases dramatically whenever wires are made thinner or placed closer together, or whenever the data throughput is increased. As a direct consequence of their resistance free pathways and noise-reduced environments, optical systems have the potential to generate less waste heat and so consume less energy per computation step than electronic systems. This has beend emonstrated experimentally with general purpose digital optical processors.