Global Optimization Algorithms Of Clusters

The cluster is a relatively stable aggregate that is formed by several, dozens of, even thousands of atoms or molecules through their mutual force. For the clusters, the different atomic arrangements correspond to different structures, and the corresponding clusters will have different physical and chemical properties. The problem we studied is to find out which structure is most stable for a cluster of a certain number of atoms.Optimization of atomic clusters is a global optimization problem and has proved to be a NP-Hard problem. Previous studies showed that NP-Hard problems are unlikely to be solved in polynomial time. So, we proposed a variety of heuristic algorithms to solve such kind problems. Though there is no guarantee that the optimal solution can be found for a heuristic algorithm, it can provide a satisfactory solution in an acceptable time.First, the gradient descent method with variable step size and gradient descent method are used to obtain the local optima of clusters. The local optima of clusters is started from an initial pattern. To locate the local optimum near the initial pattern. Clusters of local optimization algorithm can not guarantee to find the global optimum.Based on the characteristics of cluster, the problem of designing efficient heuristic algorithms is targeted.Experimental results show that the efficiency of the algorithm is low only using local optimization algorithm. To improve the efficiency of the algorithm, we have considered the features of these problems such as the huge search space. Hence we have combined heuristic algorithm to improve the algorithm. After combining with heuristic algorithm, the efficiency of the algorithm is greatly improved.