APPLICATION OF CELL LIST IN MOLECULAR DYNAMICS

Abstract

In this project, we investigated the application of cell list method in molecular dynamics simulation. We adopted the java API, Etomica, from University at Buffalo in our simulations and implemented the java package on the open source Java platform, Eclipse. In normal molecular dynamics simulation, computation time is approximately scaled by N2 where N is total number of atoms. However, with cell list method, the computation for force and energy is scaled with N at the cost of additional time for setting up the cells with molecules. Extensive numerical simulations for hard spheres in a cubic box with periodic boundary have been carried out. The experimental data obtained demonstrates the time saving by cell list method. We found the computation time does not decrease infinitely. The cell size also plays a crucial role in efficiency and accuracy of simulation. We have also attempted to extend the cell list method to rectangular box and confined hard boundary in one direction. Neighbour criteria are written to take care of neighbour cells for various boundary conditions. Preliminary results show cell list method still demonstrates 30-40% time saving in rectangular box case.