We develop a new Lagrangian material particle -- dynamical domaindecomposition method (MPD^3) for large scale parallel molecular dynamics (MD)simulation of nonstationary heterogeneous systems on a heterogeneous computingnet. MPD^3 is based on Voronoi decomposition of simulated matter. The map ofVoronoi polygons is known as the Dirichlet tessellation and used for gridgeneration in computational fluid dynamics. From the hydrodynamics point ofview the moving Voronoi polygon looks as a material particle (MP). MPs canexchange particles and information. To balance heterogeneous computingconditions the MP centers should be dependent on timing data. We propose asimple and efficient iterative algorithm which based on definition of thetiming-dependent balancing displacement of MP center for next simulation step. The MPD^3 program was tested in various computing environments and physicalproblems. We have demonstrated that MPD^3 is a high-adaptive decompositionalgorithm for MD simulation. It was shown that the well-balanced decompositioncan result from dynamical Voronoi polygon tessellation. One would expect thesimilar approach can be successfully applied for other particle methods likeMonte Carlo, particle-in-cell, and smooth-particle-hydrodynamics.
展开▼
