NextComp - Molecular Dynamics for Long-Range Interacting Systems - Possible Connection with Nonextensive Statistical Mechanics

This project has as main objective to test the validity of nonextensive statistical mechanics for the a-XY Hamiltonian model. This will be achieved through molecular dynamics (MD) simulations using a computational distributed environment (Grid). Statistical physics is a fundamental area of physics with strong connections with complex systems and condensed matter. It describes systems with large number of similar interacting elements and is based on the remarkable contributions made by Boltzmann and Gibbs in the XIX century. The Boltzmann-Gibbs (BG) theory is intended to describe systems with short-range interaction and therefore has limitations to produce correct predictions in long-range systems. Nonextensive statistical mechanics, proposed by C.Tsallis (one of the authors), aims to generalize the BG theory and is expected to produce useful results for these types of systems. The physical problem in this work is a system of classical planar rotators with a specific interaction whose range is controlled by a parameter \alpha. This paradigmatic model contains the limiting cases of infinite-range and short-range (\alpha=0 and \alpha=\infty, respectively). The computing resources necessary to perform MD simulations typically diverge when calculating long-range interactions. Another difficulty is that the physical limits of interest are commonly very high. We have developed a program to determine the applicability of the nonextensive theory in this specific model. Our program was built for parallel computation using Charm++ language and has been tested in our Linux cluster environment. Our interest is focused on a grid environment in order to overcome the limiting factors of the simulations.