Accurate prediction of heats of formation for c1-c16 alkanes: The genetic algorithm and neural network approach with simple input descriptors

Recently,the combination of genetic algorithm and neural network approach(GANN) has been carried out to improve the calculation accuracy of density functional theory. In the present work, the GANN approach with three simple input descriptors is applied to improve the accuracy of B3LYP calculation for C₁-C₁₆ alkanes. The prediction result shows that GANN is a more effective and economical techniques. The mean absolute deviations of the heats of formation of C₁-C₁₆ alkanes are 13.92, 1.05 and 0.20 kal/mol for the B3LYP, G3 and GANN methods, respectively.