Research Of Parallelization Of Dissipative Particle Dynamics Simulation

With the development of computer science and technology, the research on the molecular simulation with the high performance computing and the new algorithm are becoming a new hot topic. Dissipative Particle Dynamic (DPD), as an important tool of molecular simulation, has been a significant effect in Chemical researches and experiment. However DPD method needs many computing resources and long time to obtain result, and with the particles in DPD increasing, the computing time would be soar. Thus if the DPD serial program is paralleled by some parallel algorithms, the computing time would decrease and the resources would be saved. This paper uses the space-divided method in the parallel DPD program and the program is rewritten with MPICH. The results are ideal and the speed-up ratios are up to 1.635 and 5.8 by 2 and 8 computing nodes respectively. Besides because the traditional DPD method can not solve some nontrivial transient problems of capillary dynamics, this paper rewrites the DPD program to MDPD(Many-body Dissipative Particle Dynamics) program under the related theory and simulates the capillary dynamics to prove the rationality of MDPD serial program. And for improving the efficiency, the MDPD program is modified to parallel MDPD program. This paper simulates the Cassie module, which is one of Super-hydrophobic Interface, with parallel MDPD program and the speed-up ratio is 3.425 using 4 computing nodes, at same time obtaining an ideal load balance. Parallel computing, as one of High Performance Computing, applies for the molecular simulation, which can not only improve the molecular simulation but also largely promote the parallel computing technology itself. Therefore associating the two subjects are significant to scientific research.