A symbiotic algorithm for finding the lowest energy isomers of large clusters and molecules

A novel `symbioticʹ algorithm, based on the genetic algorithm, is presented for finding the structure and energy distribution of the lowest energy isomers of large clusters and molecules. This approach takes advantage of the strong coupling of nearest neighbor atoms through the fitness function (the binding energy) directing the selection process, due to the short range of the interatomic potential in comparison to the cluster size. Evolving locally in cells and then forming and evolving a symbiosis of the cells is substantially more efficient than employing the genetic algorithm on the full cluster. Application is made to Lennard-Jones clusters of 6, 18, 23, 38 and 55 atoms.