Berlin 2008 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Downloads | Hilfe
DY: Fachverband Dynamik und Statistische Physik
DY 29: Poster II
DY 29.65: Poster
Donnerstag, 28. Februar 2008, 16:00–18:00, Poster C
A novel off-lattice cluster Monte Carlo algorithm for fluid simulation — •Markus Bellion1, Ludger Santen1, Heiko Rieger1, and Werner Krauth2 — 1Saarland University, Theoretical Physics, D-66041 Saarbrücken, Germany — 2CNRS-Laboratoire de Physique Statistique, Ecole Normale Supérieure 24, rue Lhomond, F-75231 Paris Cedex 05, France
The invention of Cluster Monte Carlo algorithms allowed for a much larger computational efficiency compared to local update schemes. By using cluster algorithms it was possible to reduce critical slowing down or even to avoid this problem. However, up to now most such cluster algorithms have been designed for classical and quantum mechanical models that are defined on a lattice. In fact, currently there are very few cluster algorithms that work for off-lattice models, because the identification of appropriate clusters while satisfying the detailed balance condition is even more challenging for continuous systems. Existing algorithms rely essentially on a geometric symmetry operation. Although these algorithms have been successfully applied to several model systems of complex fluids, their efficiency breaks down for higher densities. Here we report on an alternative algorithm: Particle clusters are identified by iterating a translational elementary move. The main advantage is the possibility to tune the size of generated clusters by varying the step size of the translation. As the trial step is not self-inverse, establishing detailed balance is a highly non-trivial task related to a well-known graph-theoretical problem. We show how this problem can be tackled or possibly circumvented for systems of hard particles.