A Self-Adaptive Evolutionary Algorithm for Cluster Geometry Optimization

We propose a self-adaptive hybrid evolutionary algorithm for the optimization of Morse clusters. The approach relies on a two-phase local optimization method to efficiently guide search. Individuals encode its own penalty settings and the algorithm evolves them simultaneously with the search for low energy clusters. Results show that the approach is efficient, as it is able to discover all optimal solutions for Morse clusters between 41 and 80 atoms.