Physics Today. October 1987. Vol.
40, no. 10. Pages 25-72. American
Institute of Physics, New York.
The special articles in the October 1987 issue of Physics Today explore not only the use of computers by scientists, but also the discipline of computational science—a mode of operation complementary to, and distinguishable from, the familiar methods of theoretical and experimental science. The introduction and four review articles show clearly that computer simulations can discover unexpected behavior in dynamical systems. As Martin Karplus points out in his article, for example, progress in molecular dynamics is hastened when researchers are able to make comparisons between simulation and experiment.
"Exploiting Highly Concurrent Computers for Physics” by Ken C. Bowler et al is a marvelous article that discusses the use of parallel processing on a lattice to computer disparate quantities such as the hadronic masses from the standard model of the strong interaction to the structure of neural networks. Most computational physicists have yet to take advantage of the power of parallel processing because of the difficulty of embedding conventional serial programs into a parallel environment.
The ability of researchers to assimilate voluminous amounts of data is addressed in “The Numerical Laboratory.” Karl-Heinz A. Winkler et al explain how the results of “numerical experiments” are displayed graphically with color movies that make use of the viewer’s synthetic abilities residing in the right side of the brain. Indeed, computational resources equal to or even beyond those applied to the solution of complicated systems are required to convey the results to the interested viewer.
Experimental theoretical and computational physicists often choose to concentrate on a problem where length and time scales are controllable. Physical oceanographers and climatologists, however, are not permitted this luxury when attempting to understand global circulation of the oceans and the planetary air mass. The article by William Holland et al on computer simulations in physical oceanography illustrates the problems inherent in systems with many length and time scales. Even with the resources of today, initial value problems in these fields remain intractable. Except for occasionally stating results without identifying where they came from, the article does a nice job of presenting a compelling argument for gaining environmental benefits by better comprehending the results of our interaction with the ocean environment.
Supercomputers or near-super-computers are the tools common to the review articles, leading me to suggest that the somewhat more limiting title of super-computational simulation physics should have been applied to this special issue. Even super-computational physics, however, is too broad a topic to be described adequately in four articles. The editors of Physics Today are to be commended for allowing us to sample some of the arenas in which computer simulation of non-linear processes can complement theoretical and experimental work.
Peterkin is staff scientist at Mission Research Corporation, 1720
Randolph Road, SE., Albuquerque, NM 87106.
- (The Scientist, Vol:2, #2, p.20, January 25, 1988)
- (Copyright © The Scientist, Inc.)
- (Copyright © The Scientist, Inc.)
- WE WELCOME YOUR OPINION. IF YOU WOULD LIKE TO COMMENT ON THIS STORY, PLEASE WRITE TO US AT EITHER ONE OF THE FOLLOWING ADDRESSES:
- The Scientist, 3600 Market Street, Suite 450, Philadelphia, PA 19104, U.S.A.