Research On Optimization Of Atom Clusters Structure Using Intelligent Algorithms

An atomic cluster is a steady congery, which is formed by mutual forces of several to several thousand atoms, and whose physical and chemical characters often change along with the number of its atoms. The study of clusters has become an interesting research in recent years, due to its particular geometric structures, unique physical and chemical properties and potential application. The prediction for the best steady structure of atomic clusters is a thorny issue in the such research fields as computer, physics, chemistry, biological and so on.Currently the main method to predict the structure of atomic clusters is to legitimately simplify its internal structure, and build its physical model, then change the physical model to mathematic model and solve the model. Studying the potential models between atoms has much effect on the prediction for the best steady structure of atomic clusters, and the widely used potential models are empirical potential, self-consistent potential and tight binding potential. The second step of optimization is searching steady structure on the potential energy surface. The main methods are Monte Karlo algorithm, Molecular Dynamics method, Simulated Annealing algorithm, genetic algorithm and so on.The paper mainly researched on the optimization of Ar and Cu clusters.The structure of Arn clusters was optimized by using genetic algorithm, particle swarm optimization algorithm, improved particle swarm optimization algorithm and differential evolution algorithm combined with Lennard-Jones potential in order to obtain a stable structure of the clusters. The Arn clusters with cluster size n=2,3,…,14 have been demonstrated to verify the effectiveness of the proposed method. Compared with genetic algorithm and particle swarm optimization algorithm, a better convergence characteristic and a shorter computing time can be achieved by the improved particle swarm optimization algorithm in the optimization of Arn cluster structure. Using differential evolution we can get the ground state energy and steady structure of Arn(n=15-30).The results show that the structure of Arn clusters obtained by the method has a high degree of symmetry.With Gupta potential we have got the ground state energy and steady structure of Cun by using genetic algorithm and the improved particle swarm optimization algorithm. The structure of Cun clusters obtained by the methods also has a high degree of symmetry.All the results show that intelligent evolution algorithms are effective in the optimization of clusters. The improved particle swarm optimization algorithm has the best convergence characteristic than others. We can use it for optimization of complex clusters.