FITTING POTENTIAL-ENERGY SURFACES - A SEARCH IN THE FUNCTION-SPACE BYDIRECTED GENETIC PROGRAMMING

We propose new procedures by which genetic programming can be used to find the best functional form and the best set of parameters to fit th e energies and the energy derivatives provided by no initio calculatio ns. Our main contribution is a new procedure, which we call a directed genetic search, which is more efficient and more stable than a ''trad itional'' genetic program. (C) 1998 American Institute of Physics.