Finding Molecules with Specific Properties: Simulated Annealing vs. Evolution

Abstract

We compare the ability of a simulated annealing program and an evolutionary algorithm to find molecules with large molecular average hyperpolarizabilities. This property is an important component of nonlinear optical materials. Both optimization programs represent molecules as SMILES strings, a method that is widely used by chemists to describe molecular structure using short ASCII strings. Our results suggest that both approaches are comparable and can be used to solve a variety of more realistic problems of interest to chemists and material scientists.

Figures (3)

• Figure 1: Average molecular hyperpolarizability (and its standard deviation) vs the number of generations. This illustrates the effect of the mutation/crossover ratio on the convergence. The values for this figure are summarized in Table 2 and have been averaged over 5 different random number seeds.
• Figure 2: Average molecular hyperpolarizability (and its standard deviation) vs the number function evaluations. This illustrates the relative convergence of simulated annealing and the evolutionary algorithm from the perspective of the number of function evaluations. The values for this figure are summarized in Table 2 and 3 and have been averaged over 5 different random number seeds.
• Figure 3: Molecule produced by our 10/90 evolutionary program with $\beta$=649417 atomic units.