Computational Physics

Computational physics Kenneth Wilson — comparing today's particle colliders with the microscopes of Galileo's time. (Photo ICTP) Computers have for many years played a vital role in the acquisition and treatment of experimental data, but they have more recently taken up a much more extended role in physics research. The nu merical and algebraic calculations now performed on modern com puters make it possible to explore consequences of basic theories in a way which goes beyond the lim its of both analytic insight and ex perimental investigation. This was brought out clearly at the Conference on Perspectives in Computational Physics, held at the International Centre for Theoretical Physics, Trieste, Italy, from 29- 31 October. It was directed by Fred James (CERN), Alvise Nobile (Trieste), and Claudio Rebbi (Brook- haven and Boston). The birth of computational phy sics can be traced back to the late 1960s with the first journals, con today will be like today's particle areas of computational physics ferences and schools on the sub colliders compared with the micro reviewed the state of the art from ject. Although endrmous progress scopes of Galileo's time. all points of view: physics algo has been made since then and For Wilson, perhaps the greatest rithms, software techniques, and whole new fields such as lattice algorithmic challenge facing con hardware developments, as well gauge calculations have started temporary computational physics as their interrelationships. up, it is clear that computational is demonstrated by the problem Supercomputer architecture was physics is still in its infancy. In fact of electronic structure. Here is an of course a topic of interest to all Ken Wilson (Cornell) in his invited area where the basic theory, quan participants, and was covered in talk compared the current situation tum electrodynamics, is known to talks by several physicists and in computational physics with that be valid to an extremely high accu representatives of computer man of experimental physics at the time racy, sufficient to predict the phy ufacturers. The physics areas cov of Galileo when important discov sical, chemical, and biological pro ered in greatest detail, in addition eries were made using rudimentary perties of all atomic and molecular to several aspects of electronic microscopes, telescopes and lean states. However current calculation structure, were lattice gauge the ing towers. Four hundred years techniques are barely powerful ory, stellar dynamics, and many- later, experimental physics has enough to compute gross proper nucleon systems. developed techniques capable of ties of systems involving a few One of the highlights of the con penetrating many orders of mag hundred electrons, with computing ference was the real-time demon nitude deeper into matter and into time increasing typically as the stration of the possibilities of cel the universe. square of the number of electrons, lular automata by Tom Toffoli By analogy we expect that four and interesting chemistry and biol (MIT). In these discrete systems, hundred years from now it will be ogy taking place in the region up each successive state is derived possible to perform calculations to millions of electrons at least. from the previous state by a rela many of orders of magnitude more But the conference was not only tively simple rule, which may be complex, and future computing devoted to dreaming about the deterministic or partly random. By engines compared with those of future: world experts in many varying the replication rule, Tom 4 CERN Courier, January/February 1987 At the Conference on Perspectives in Computational Physics held at the International Centre for Theoretical Physics, Trieste, Italy, in October. Left to right: G. Parisi, A. Nobile, F. James, G. Jacucci, C. Rebbi, T. Toffoli and K. Wilson. (Photo G. Montenero) was able to model many mathe matical and physical phenomena from shock waves to fractal growth and stellar dynamics. Using his own special hardware board under the control of an Olivetti M24 personal computer he was able to calculate successive states of the automata with sufficient speed that large-screen colour projection gave an uncanny feeling of observing the evolution of com plex continuous systems obeying known 'physical' laws. This pro vides considerable insight into important phenomena like order and disorder, phase transitions, stability and reversibility in physical systems. The conference took place im mediately after the School on Ad vanced Techniques in Computing in Physics, held also at |CTP and with the same organizers. The three-week school offered in-depth courses on pure computing topics (programming languages, operating systems, networking, etc.), numer ical and symbolic analysis tech niques, and physics applications, to 150 students selected from over 400 applicants, and coming mostly from the developing countries. These 'students' turned out to be highly qualified computational scientists, which made for an unex pectedly lively and stimulating school. One of the lecturers even remarked that it was his most res installation of an entire computer about 70 more people (nearly all ponsive audience, despite having centre, with everything from a false from Western Europe and North given similar talks as seminars in floor to a team of systems ana America). some of the world's most presti lysts, all of which was beyond the One evening was devoted to an gious laboratories. One explanation possibilities of many countries. open discussion of the future of was the school's highly selective Nowadays the same computer Computational Physics. The pres acceptance procedure, but it is power is available just by plugging ence of large delegations from also a clear sign that competence in a PC. The effects of this quiet developing countries made it a in computing is increasing fast in revolution have been spectacular. natural forum for computational many less developed countries. Most of the 150 school partici physics research in the poorer re Only a few years ago, access to pants stayed on for the confer gions of the globe. One western a reasonable computer meant the ence, where they were joined by participant said he had never be- CERN Courier, January/February 1987 5 Wandel &Goltermann Electronic Measurement Technology The wide variety of different telecom mo dules such as CODECs and line cards requires a wide range of telecommuni cations measurements. The AMS-964 can make these measurements at a rate AMS-964 of up to 4 per second. It can also be adapted to new test objects rapidly, mak ing it ideal for laboratory use or for testing for versatile testing small series. Results are reproducible to ±0.02 dB, and a range of standard of CODECS, COFIs, interfaces ensure optimum matching of the AMS-964 to the test item, at an economic price. The many advantages line cards etc. - make the AMS-964 worth getting to know! Send for free info. fast accurate Wandel &Goltermann, Abt. PMW, Postfach 45, D-7412 Eningen Fed. Rep, of Germany, Tel. +(49) 7121-8911, Tlx. 729833 I would like and economical • a free copy of the AMS-964 colour brochure • a visit from a sales engineer Name ... Company Street Town , Tel. ,. E6373K 6 CERN Courier, January/February 1987 fore been at a conference where tion of the Schrôdinger equation), Computational Physics library is so many developing countries had turbulence, function minimization enormous. And there is network been represented, and it clearly (for protein folding, spin glasses, ing, which both solves and intro came as a surprise to him and oth etc.), quantum field theory, and duces many problems, but does ers that it was not only possible stellar evolution. The difference in not obviate the need for centres to carry on computational physics the time-, length-, and energy- of excellence. research in such places, but that scales is impressive. The economic aspects of the it was being done actively. The Communications issues are also research cannot be neglected. The obvious conclusion is that this very important. The usual language future of Computational Physics enormous source of intellectual of Computational Physics, Fortran, will depend on how we interact potential cannot be neglected. has long been recognized as inad with the larger scientific computing Ken Wilson's summary talk cov equate in many respects, especially market, which has a huge industrial ered the major outstanding issues as a vehicle for explaining what a base representing about $ 10 bil for computational science as he program is expected to do. Yet lion per year, and is truly inter saw them and as they were none of the many other languages national. With the technology ad brought out at the conference. His has had widespread acceptance. vancing rapidly on many fronts, first point was: what is quality There are other important as the prospects for Computational research? Or, equivalently, what pects of the publication issue: Physics are apparently limited only research will still be respected four Where should new papers be pub by our own skill and imagination. centuries from now? lished, where and how should pro The great algorithmic challenges grams be published? The list of By Fred James remain : electronic structure (solu journals needed for a complete At the 'Jackfest' marking the History of the weak 65th birthday of Jack Stein- berger (see July/August 1986 issue, page 29), T. D. interactions by T. D. Lee Lee gave an account of the history of the weak interac tions. Lee was a graduate In 1898 Rutherford discovered people on my track. I have to pub student with Steinberger un that the so-called Becquerel ray lish my present work as rapidly as der Enrico Fermi in Chicago actually consisted of two distinct possible in order to keep in the from 1946, and went on to types of radiation : one that is read race.

Load more