An internal-coordinate Monte Carlo method for searching conformational space

TLDR: In this article, an internal coordinate random-search method for finding low-energy conformations of organic molecules is described, which is biased toward the low energy regions of conformational space by choosing starting geometries for each step in the conformational search.

Abstract: An internal coordinate, random-search method for finding the low-energy conformations of organic molecules is described. The search is biased toward the low-energy regions of conformational space by choosing starting geometries for each step in the conformational search from among previously stored low-energy conformers. Tendencies of the search method to concentrate in certain regions of conformational space at the expense of others are reduced by uniform usage of stored structures as starting geometries and by using varying numbers or torsional rotations in each step. Tests of the procedure's ability to find all low-energy conformations of several acrylic, medium- and large-ring molecules are described. Direct comparisons of the method with Cartesian coordinate random methods and systematic internal coordinate grid searches are carried out with symmetrical and unsymmetrical molecules. The method makes it possible to carry out global conformational searches on symmetrical and unsymmetrical molecules having up to a dozen variable torsion angles.