A new paradigm for petascale Monte Carlo simulation: Replica exchange Wang-Landau sampling

25th IUPAP Conference on Computational Physics

物理学;计算机科学

Li, Ying Wai^1,2 ; Vogel, Thomas^1,3 ; Wüst, Thomas^4 ; Landau, David P^1

Center for Simulational Physics, University of Georgia, Athens, GA 30602, United States^1

National Center for Computational Sciences, Oak Ridge National Laboratory, Oak Ridge, TN 37831, United States^2

Theoretical Division, Los Alamos National Laboratory, Los Alamos, NM 87545, United States^3

Swiss Federal Research Institute WSL, Zürcherstrasse 111, CH-8903 Birmensdorf, Switzerland^4

关键词: Densities of state;    Energy spaces;    Free energy landscape;    Massively parallels;    Replica exchange;    Topological constraints;    Wang-Landau method;    Wang-Landau sampling;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/510/1/012012/pdf
DOI  :  10.1088/1742-6596/510/1/012012

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

We introduce a generic, parallel Wang-Landau method that is naturally suited to implementation on massively parallel, petaflop supercomputers. The approach introduces a replica-exchange framework in which densities of states for overlapping sub-windows in energy space are determined iteratively by traditional Wang-Landau sampling. The advantages and general applicability of the method are demonstrated for several distinct systems that possess discrete or continuous degrees of freedom, including those with complex free energy landscapes and topological constraints.