2006 Volume 7 Pages 163-167
For prediction of molecular property and elucidation of physical mechanism it is important to use molecular dynamics(MD) simulation. However, there is necessity for speeding up MD simulation because these simulations of large size molecules such as dendrimer expend huge calculation cost. For that purpose, it is most effective to improve the part for calculating Coulomb interactions which dominates in the entire simulation process. The mutipole algorism ,which has same effect as converting far many particles to pseudo one particle, is one of the most powerful methods for solving the problem. In this research, we develop and improve the Multipole method for MD simulation. The Fast Multipole Method(FMM), which is one of the Multipole algorisms, is often used in the MD simulation. The drawback of this method is the high cost of the improvement in the accuracy. In this research, we develop a new FMM. For improving the accuracy efficiently, we employed the least mean square method on the FMM local expansion, instead of Taylor expansion. In order to compare our FMM with conventional FMM, we calculate Coulomb energy among the particles generated at random and that among electronic charges of fifth generation dendrimer. It is shown that the accuracy of our FMM is twice as much as that of the conventional method under the same condition, and that the calculation cost of our FMM is almost equivalent to the conventional FMM.
