have adapted ideas or technologies from other disciplines, developed them for their own needs, and returned a more powerful, practical product. One example is the effort in lattice gauge theory, illustrating the give and take with the computer industry and other branches of physics.

The most intensive computer jobs are in the domain of lattice gauge theory, part of the theory of elementary particles. As this work blossomed in the 1980s, it quickly became clear that the most powerful commercial supercomputers would be neither adequate nor cost-effective. A popular concept for reducing costs was to take commercial processors and connect them to each other. Such a computer is called "massively parallel" because very large numbers of processors compute simultaneously. With a massively parallel machine, one could, in principle, split up big problems and let each processor do a fraction of the job. The drawbacks are the difficulties of coordinating the split-up and of communicating the data among processors. Theoretical particle physicists decided to design and build parallel computers specifically for lattice gauge theory. They came up with elegant solutions to the coordination and communication problems, and the resulting machines are among the first practical examples of massively parallel computers. One of these consists of 8,000 50-MHz processors! Now, many computer vendors offer a parallel computing product.

The mathematical structure of lattice gauge theory has much in common with that of systems in condensed-matter theory, because both grapple with problems of large systems. After understanding the physical meaning of "renormalization" in elementary-particle theory, Ken Wilson sought a simpler problem on which to test his insights. He solved some outstanding problems of condensed matter physics (later winning a Nobel Prize for the accomplishment) and came back from the experience with a way to define the theory of quarks and gluons on a lattice. Techniques of the resulting "lattice QCD" have developed side-by-side with condensed-matter theory ever since. In particular, computer programs running on massively parallel machines offer the most reliable way to work out details of the attraction between quarks inside the proton. As a result, many nuclear physicists have started to study QCD to understand nuclei as (complicated) composites of quarks and gluons.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement