Cellular Machines - Theory and Practice

Abstract

The idea of cellular automata or iterative arrays of processors is far from new and owes much to the pioneering work of John von Neumann¹ in association with others, such as Stanislaw Ulam² who saw, in the example of the human brain, an almost unbelievably powerful computing structure which owed its strength to a regular organisation of fundamentally simple computational units (neurons). However, it could be said, with the benefit of hindsight, that von Neumann’s contribution was impeded by a technologically premature realisation that large arrays of computing elements would be subject to serious failure unless some self-repair mechanism could be built into them. He devoted the last years of his life, until 1957, studying the complexity requirements necessary to enable array elements to reproduce themselves, this being an effective way of achieving self-repair. In fact, much can be usefully done with cellular arrays before the need to incorporate self-repair is an important factor and it is to be regretted that the brilliant von Neumann did not apply his mind to the more practical development of his subject.