A Genetic Algorithm for Solving Combinatorial Problems and the Effects of Experimental Error - Applied to Optimizing Catalytic Materials

A new form of Genetic Algorithm (GA) is introduced which has been developed to solve combinatorial problems. A combinatorial problem involves choosing the best subset of components from a pool of possible components in order that the mixture has some desired quality. This paper concentrates on applying the new technique to the optimization of catalytic materials. The new form of GA is compared to an evolutionary algorithm developed by Wolf et al. and shown to produce faster convergence. The paper also reports on the best GA parameter values (crossover technique, parent selection etc.) for problems such as these. Finally a statistical analysis of the effects of experimental error is performed, and the effects that these errors have on the convergence of the GA are reported.