Using a genetic algorithm to simulate hydrocarbon and hydrosilicon structures

CHED 1049

Charles Rareshide,, Stetson University, 421 N. Woodland Blvd., Unit 6646, DeLand, FL 32723 and Kai-Ming Ho, Department of Physics and Astronomy, Iowa State University, Condense Matter, A502 Physics, Ames, IA 50011-3160.
A geometric genetic algorithm is used to simulate hydrocarbon and hydrosilicon clusters. The algorithm produces possible candidates for hydrocarbons using Brenner's empirical potential and the Hansen-Vogel potential for the hydrosilicons. The candidates are then mated to form new candidates. The candidate pool is updated and relaxed. This technique is repeated until no new structures with lower energies are being produced within a few hundred iterations. The total energies of the structures are calculating using MP2 and DFT methods. The calculated energies for the hydrocarbons are then compared to the experimental formation enthalpies found in the NIST database. The genetic algorithm is found to be an efficient optimization tool for finding low energy structures. MP2 calculations for the hydrocarbons appear to correlate more with the experimental formation enthalpies than the DFT energy calculations.