Reduced-Order Physics Models For Computationally Efficient Charged Particle Transport

MON, 02/14/2011 - 4:00PM TO 5:00PM
Spring 2011 Colloquium Series

High-energy charged particles (electrons, light and heavy ions, from 10's keV to GeV and above) are ubiquitous in nuclear science and engineering, medical physics, space science, and advanced materials applications, but the computational complexity associated with the transport of charged particles can greatly exceed that of neutrons, photons and other neutral particles. The primary reason is that long range, Coulomb-field mediated elastic and inelastic collisions of energetic charged particles with target nuclei and electrons are characterized by extremely small collision mean free paths and near-singular differential cross sections. This extreme physics renders the computational modeling of the analog or true problem prohibitively expensive in both stochastic (Monte Carlo) and deterministic numerical settings. The condensed history (CH) Monte Carlo method, widely employed in electromagnetic and hadronic shower codes, attempts to circumvent this practical difficulty by advancing the particle in fixed large steps but inherent flaws limit the ultimate accuracy possible with this method.

In this talk, a new approach will be presented that obviates the need for CH-like approximations for computational expediency yet retains the “look and feel” of the single-event or event-by-event analog simulation. The essence of this method is the construction of a pseudo-transport problem with de-singularized collision operators, which are constrained to preserve certain moments of the corresponding analog collision operators. This moment-preserving or reduced-order physics model is developed through a variety of projection-based strategies that include stabilizing asymptotic higher order Fokker-Planck expansions by renormalization methods and using purely discrete as well as hybrid discrete-continuous kernel representations. The result is a systematic and robust, particle species independent methodology, which can achieve high accuracy and computational efficiency for energy straggling, angular spreading, and dose calculations. After some background material, the new formalism will be described in detail, followed by a presentation of several illustrative numerical results from a Monte Carlo implementation, and concluding with a discussion of some outstanding challenges.

About the Speaker:

Dr. Anil K. Prinja is currently Professor and Associate Chair of the Chemical and Nuclear Engineering Department at the University of New Mexico. He also holds the positions of Co-Director (former Director) of the Center for Nuclear Nonproliferation Science and Technology, and Co-Director of the interdisciplinary Medical Physics Program at UNM. Dr. Prinja obtained his Ph.D. (1980) and B.Sc. (1st Class Honors, 1976) in Nuclear Engineering from the University of London, UK, and was a Research Engineer at UCLA (1980-1987) prior to joining UNM. Dr. Prinja is a Fellow of the American Nuclear Society, and a recent recipient of the NNSA Defense Programs “Award of Excellence”.