ANALYSIS OF THE GENETIC ALGORITHM METHOD OF MOLECULAR-CONFORMATION DETERMINATION

Many important problems in chemistry require knowledge of the 3-D conf ormation of a molecule. A commonly used computational approach is to s earch for a variety of low-energy conformations. Here, we study the be havior of the genetic algorithm (GA) method as a global search techniq ue for finding these low-energy conformations. Our test molecule is cy clic hexaglycine. The goal of this study is to determine how to best u tilize GAs to find low-energy populations of conformations given a fix ed amount of CPU time. Two measures are presented that help monitor th e improvement in the GA populations and their loss of diversity. Diffe rent hybrid methods that combine coarse GA global search with local gr adient minimization are evaluated. We present several specific recomme ndations about trade-offs when choosing GA parameters such as populati on size, number of generations, rate of interaction between subpopulat ions, and combinations of GA and gradient minimization. In particular, our results illustrate why approaches that emphasize convergence of t he GA can actually decrease its effectiveness as a global conformation search method. (C) 1993 by John Wiley & Sons, Inc.