Global Optimization Algorithms And Application For Structural Optimization Of Clusters

Global optimization problem is a subject of intense current interest. In chemical field, finding out the energy-lowest conformation of a structure is of great importance. However, the structural optimization problem is notoriously difficult because the number of local minima tends to grow exponentially with system size. In this dissertation, several novel and efficient global optimization methods were proposed and applied to cluster structural optimization problem. The energy landscape of clusters was also studied with a conformational analysis method.In this dissertation, basic knowledges, methods and significance of cluster studies were introduced, and the global optimization methods employed in structural optimization of atomic or molecular clusters were reviewed. The main works contained in the dissertation include:1. Based on the immune theory of biology, a novel evolutionary algorithm, adaptive immune optimization algorithm (AIOA), was proposed. In AIOA, density regulation and immune selection is adopted to control the individual diversity and the convergence adaptively. By applying the algorithm to the optimization of test functions, it was shown that the algorithm is a highly efficient optimization method compared with other stochastic optimization methods. The algorithm was also applied to Lennard-Jones (LJ) clusters, and optimal structure up to LJ80 was reproduced.2. A novel and effective cluster similarity checking method using the connectivity table (CT) was proposed. Because CT contains the topological information of a cluster, configurations at different funnels on the potential energy surface (PES) show great difference in their CTs. A new version of AIOA was utilized for optimization of LJ clusters up to LJ110 using the CT for similarity checking. It was proved that the method is very efficient, and the method is also capable for optimizations of larger clusters, e.g., LJ200.3. An energy-based perturbation and a new idea of taboo strategy were proposed for structural optimization and applied in a benchmark problem, i.e., the optimization...