Exogenous parameter selection in a real-valued genetic algorithm

Abstract

To evaluate the performance of a real valued genetic algorithm (GA) exploiting domain knowledge, we systematically evaluate the effect of exogenous parameters using analysis of variance. The GA platform used for this study is Genocop-III, a real valued, co evolutionary algorithm implementation for numerical optimization. We use the protein structure prediction (PSP) problem as our test domain. Nearly all PSP research assumes the native conformation of a protein corresponds to its global minimum free energy state. Thus, our application integrates Genocop-III with our implementation of the CHARMM energy model as the objective function. Results and conclusions drawn from an extensive experiment set using the polypeptide [Met]-Enkephalin are presented from an exogenous parameter selection perspective.