Research On Small Cluster Structures Using Intelligent Optimization Algorithms

An atomic cluster is a steady congeries formed by the 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, because of its particular geometric structures, unique physical and chemical properties and potential application. Nowadays, the prediction of the best steady structure of atomic clusters becomes a research hotspot in physics and chemistry fields.The interaction between atoms can be expressed as a potential function, which is an essential physical quantity for the prediction of matter structure. At present, the widely used potential functions are Sutton-Chen potential, Lenard-Jones potential and so on. The main methods to research clusters are molecular dynamics method, Monte Karol algorithm, density functional theory, etc. However, it often falls into the local minimum when these methods are applied to research the cluster ground state structure. So the new methods based on swarm intelligence algorithms such as genetic algorithm, particle swarm optimization algorithm, differential evolution algorithm, are praised highly by scientists in resent years. These swarm intelligence algorithms are more perfect and efficient, also they make more convenient to predict and calculate the clusters. In this thesis, the main contents as follows:(1) The structure optimization problem of iridium atom clusters was solved by means of the particle swarm optimization algorithm combined with Sutton-Chen potential,(a) In the case of the atomic number of2to20for the iridium atom clusters, the state energy, ground-state structure, average bound energy and second-order difference value were obtained by the proposed method. Also, the variation curve about the average bound energy of iridium and the second order difference value with the number of particles in the algorithm was drawn.(b) The magic number of iridium clusters was achieved, through the analysis of the second-order difference value.(2) The structure optimization problem of nickel atom clusters was solved by several intelligent optimization algorithms combined with Sutton-Chen potential.(a) In the case of the atomic number of2to20for the nickel atom clusters, the state energy was calculated by the particle swarm optimization algorithm and genetic algorithm, where the consumed time has been compared for both algorithms. According to the calculation results, the structure projection map of the nickel atom clusters was drawn,(b) With the differential evolution algorithm, the spatial structure chart of nickel atom clusters was obtained, whose symmetry was discussed,(c) In order to compare the optimize performance of different algorithms, the optimization processes of the atomic number of8and15for the nickel atom clusters were also discussed.The results can be summarized as follows:All nickel cluster structures obtained by the different intelligent optimization algorithms present good symmetry. And it is feasible and effective to apply intelligent optimization algorithms to predict the cluster structure of the transition metal atoms, which is verified by several examples.The obtained structure and ground-state energy are identical with the published literature; overall optimization performance is better than other traditional methods.Compared with the genetic algorithm and particle swarm optimization algorithm, the differential evolution algorithm has the advantages of good convergence characteristics and short computing time when the structure of the nickel atom clusters is optimized.The results obtained in this thesis provide some guidance and reference for the experimental investigation of the transition metal atom cluster structures. Furthermore, the intelligent optimization algorithms are expected to become an effective investigation method for complex cluster structures.