Optimization of Semiempirical Quantum Chemistry Methods via Multiobjective Genetic Algorithms: Accurate Photodynamics for Larger Molecules and Longer Time Scales

Excited-state photodynamics is important in numerous varieties of important materials applications (e.g., liquid crystal display, light emitting diode), pharmaceuticals, and chemical manufacturing processing. We study the effectiveness of multiobjective genetic and evolutionary algorithms in multiscaling excited-state direct photodynamics via rapid reparameterization of semiempirical methods. Using a very limited set of ab initio and experimental data, semiempirical parameters are reoptimized to provide globally accurate potential energy surfaces, thereby eliminating the need for expensive ab initio dynamics simulations. Through reoptimization, excited-state energetics are predicted accurately via semiempirical methods, while retaining accurate ground-state predictions. In our initial study of small photo-excited molecules, our results show that the multiobjective evolutionary algorithm consistently yields solutions that are significantly better—up to 384% lower error in the energy and 87% lower error in the energy-gradient—than those reported previously. As verified with direct quantum dynamical calculations, multiple high-quality parameter sets obtained via genetic algorithms show near-ideal behavior on critical and untested excited-state geometries. The results demonstrate that the reparameterization via evolutionary algorithms is a promising way to extend direct dynamics simulations of photochemistry to multi-picosecond time scales and to larger molecules, with promise in more application beyond simple molecular chemistry.