A new optimization method for shell model interatomic potential parameters of perovskite ferroelectrics

An effective interatomic potential is crucial to molecular dynamics simulation and hence an efficient method for optimizing potential parameters is necessary. A new method to find the key parameters in shell model potential of perovskite ferroelectrics based on sensitivity analysis is presented. Whereafter, the genetic algorithm is applied to optimize the key parameters, and the insignificant parameters are kept constant in optimization. This approach can substantially reduce the dimension and computer time of optimization. Simulation results using the optimized potential parameters show that the crystalline structures and physical properties of perovskite ferroelectrics agree well with the experimental data.