McAulay provides a first-rate comprehensive introduction to a relatively new area of computer science. He covers all of the theoretical foundations needed to understand how optical computing is possible, shows how these elements can be related in an optical computer, and applies these principles to the development of linear and nonlinear neural network applications.

One of the real strengths of this text is its methodical introduction of each topic in a way that makes the subject accessible to both engineers and computer scientists. The author covers the basic concepts needed for the computer scientist to understand optical switches and for the engineer to understand cellular automata.

The book is divided into three sections. The first section, which presents a background for optical computing, includes a justification of the need for optical computers, basic concepts in optics, Fourier transforms with lenses, and a discussion of devices used to interface optical devices with electronic devices. The second section presents the subsystems needed for optical computing. These subsystems include optical switches, optical memory, optical logic and optical logic circuits, and optical arithmetic. The third section presents models for optical computers. Optical sequential machines, such as RISC machines, are evaluated. These are then compared with optical dataflow machines. This section also discusses the relationship of elements of computer science theory, such as Turing machines, cellular automata, and Gödel numbers, to the construction of optical computers. The book concludes with a description of the application of optical computers to linear and nonlinear neural networks and optical auto-associative and self-organizing networks.

The text makes some complex concepts accessible to most students. Most of the concepts are clearly defined before they are related to optical computing. For example, McAulay explains RISC architecture, dataflow machines, neural networks, and modes of optical learning. Engineering elements such as crossbar switches are also carefully introduced. Whether you are a computer scientist or an engineer, this book will enhance your understanding of optical computing. Not only are the physical and the software elements clearly explained, but the relationship between them is made clear. For example, the discussion of optical logic is tied to a dual-rail system, which is then related to the discussion of cellular automata.

Chapters 1 through 3 lay the conceptual foundation for optical computing, including Fourier transforms and how these can be done with a lens, and the nature of holograms and how holograms can be used to do addition. Chapters 4 and 5 introduce the component computing devices that utilize these principles, such as acousto-optical devices. The technical information is mixed with practical advice and predictions about cost and potential availability of components. Sometimes, however, these observations get repetitive. For example, in the first 200 pages, the author mentions at least five times how laser diodes have become inexpensive because of their use in compact disks. He briefly mentions multivalued logic as a future development, but does not pursue this topic. Given that currently available multiplexors can parse at least 17 distinct values of light, the potential for replacing two-valued logic with multivalued logic should have been pursued.

Chapter 7 discusses different types of memory systems and how holographic memory is used and addressed in an optical computer. He also discusses optical associative memory and how it can be used to improve database accesses. Chapter 8 shows how logic is done on an optical computer by comparing conventional logic with highly parallel optical logic. This logic can, using optical shadow casting, compute all logic operations in a single step. Chapter 9’s discussion of optical circuit logic unfortunately does not mention the new optical chips that have been developed. Because of the optical properties of connections on a chip, higher chip densities and higher throughput can be achieved.

Part 3 presents and evaluates various computer models. An optical RISC architecture is rejected because it fails to exploit the parallelism available in optical computers. The author also shows how the inherent parallelism of Prolog can be achieved on an optical computer.

This book makes the recent strides in optical computing available to the undergraduate. It would be an excellent textbook. Exercises appear at the end of each chapter. These exercises are sometimes useful, but in several cases they are merely multiple-choice questions, and none of the exercises are worked for the student.

Several errors should have been caught. For example, the word “first” is repeated on page 177, “it” appears for “is” on page 174, and the introduction to exercise 3 on page 372 makes no sense.

This book has only a few deficiencies. They are outweighed by its quality and completeness, and by the breadth of understanding the book conveys. The style is clear and easily readable. This book is the best single resource currently available on this new and significant subject. I strongly recommend it to all college libraries and to anyone interested in optical computing.