Abstract
Comparative tests are made of the performance of several methods of minimization in applications to static lattice potential energy calculations made on defects in organic molecular crystals. Tests are made using an earlier reported cyclic procedure based on Newton's method,1 and an implementation by Powell of the Broyden-Fletcher-Goldfarb-Shanno (BFGS) quasi-Newton update2 applied simultaneously to all variables. The methods and the accuracy attainable by each are described. The results indicate that when not close to a potential energy minimum the methods seem to be equally valuable. However, when close to a potential energy minimum the well-known superior convergence of the BFGS method is exhibited.

This publication has 13 references indexed in Scilit: