Parallel Simulation of Polymers on the Cray T3E
Maximilian Putz, Alexander Kolb, Burkhard Dünweg · 1997
. In this article we present a way of simulating polymer systems on the mesoscopic scale using massively parallel computers. We use a continuum bead-spring model and molecular dynamics with statistical noise and friction to propagate our systems in time. For simple flexible polymer melts and networks our model contains only pair interactions. In this case spatial decomposition of the simulation box onto several processing units is the ideal way of parallelizing the problem. We describe the algorithm we use and the data structures involved in considerable detail. The communication overhead is quite small, giving rise to an almost perfect scaling of the algorithm up to 256 Cray T3E processing elements, the only requirement being that our dense systems exceed a particle number of roughly 400 particles per processor. Furthermore we have developed a hybrid approach using both spatial and force decomposition for more complicated interactions along the backbone of the polymer chains. This app...