Professor Achi Brandt
Weizmann Institute of Science
Rehovot, Israel

Wednesday, July 21, 1999
3:00 pm (CT)
3269 Beckman Institute

Abstract

Even with the vastly simplified models of molecular dynamics, computing costs rise very steeply with problem size, restricting the modeling to relatively small systems and to time intervals many orders of magnitude smaller than needed. This steep rise can be radically reduced by combining several multiscale approaches, such as: fast multiscale summation of electrostatic forces; algebraic-multigrid-like energy minimization techniques; large time steps that blend statistical approaches in the small with deterministic dynamics in the large; and, most important, multiscale Monte-Carlo. The latter involves renormalization-type derivation of increasingly coarser molecular Hamiltonians based each on local simulations at the next finer level. Demonstrations of these approaches for simple models (a simple polymer and simple fluids) will be mentioned.

Tea and coffee will be served in R3151 Beckman Institute at 2:15pm.