Education

B.S., Cornell, 1970
Ph.D., Physics, University of California, 1975

Research Interests

Theoretical plasma physics; computational physics.

Current Projects

Applied Physics Simulation: Developing a Predictive Capability

One of the great ironies of modern applied physics is our continuing inability, given the prodigious computer power now so commonplace, to solve seemingly simple physics problems of practical significance. Why, for example, can we not compute the airflow around a golf ball sufficiently accurately to predict its drag? Or calculate the heat transfer from a coolant flowing through a typical heat exchanger? We have known the equations governing these phenomena for almost 200 years. The computing power that was termed "supercomputer" in 1990 now sits on the desktop of virtually every modern researcher. Why are these types of problems so difficult to simulate? Is a predictive simulation method possible in the near future? Our applied physics simulation project, with support from the nuclear weapons program at Los Alamos National Laboratory, is addressing these questions.

Two practical social and economic forces of the modern world support this activity and stand to benefit from its developments. One is the situation of our nation’s nuclear deterrence and how this is to be maintained in an era without nuclear testing. Currently, the nuclear weapons program is committed to Science Based Stockpile Stewardship as a policy to replace nuclear testing. This means developing a predictive capability for computer simulation of nuclear explosions. Another socio-economic force is the modernizing trend in engineering practice emphasizing computer simulation design as a method to dramatically speed the design process and leading to shorter, more competitive, product design cycles.

To achieve a predictive capability we must confront the fundamental incompatibility between the mathematics of physical law — continuum differential equations — and the mathematics of computation — discrete, binary operations. When it comes to practical problems like the golf ball airflow, this difference in the mathematics is not simply a question of accuracy. The behavior of discrete approximations to the continuum laws can differ in a qualitative way from the physical and introduce spurious discretization artifacts. The applied physics simulation project is continuing to investigate research methods in discrete mathematics that can capture (and efficiently compute) the essential physics of the continuum.