Heuristic Optimization Of The Binary-Morse Cluster

Locating the enery-lowest configuration of a cluster has been the focus of computational chemistry. In the search of energy-lowest configuration of a cluster, stable and efficient global optimization algorithm has played a vital role, so locating the energy- lowest configuration is also of great importance in the field of computer science. Cluster is composed of atoms, molecules or ions, which is relatively stable microscopic and submicroscopic aggregate bonding through physical or chemical force, and its physical and chemical properties change as the change of configuration. However, with increasing atomic number, cluster configuration space is exponential growth, finding the energy- lowest configuration is proved to be a NP-Hard problem, these problems has not yet been solved in polynomical time complexity, for it, people proposed a lot of heuristics algorithms. These alogorithms include Simulate Annealing, Basin-Hopping, Tabu Search, Heuristic algorithm with Surface and Interior, Simple Greedy method with Energy-based perturbation and Daynamic Lattice Searching Algorithm, etc.This article mainly focuses on the Binary Morse cluster. Binary Metallic cluster consists of two kind of metal atom, and the coefficients of the two kind of atom are 5 and 7 in this study. N umerous studies have demonstrated that binar y metallic cluster performances better chemical characteristics compared to single metallic cluster, however, due to the diversity of the configuration of binary metallic cluster, the searching space of energy- lowest configuration increases greatly, so the requirement to the algorithm is more strictly. In this paper, we have made two improvements on the Dynamic Lattice Algorithm(DLS). O n the one hand, in the period of Dynamic Lattice Construction, we construct the Dynamic Lattices with two types. On the ot her hand, on the Dynamic Lattice Searching phase, we are searching for the Dynamic Lattice on the base of the type and number of enery-highest atoms. And the experimental results show that the two improvements have great impact o n the efficiency and stability of the algorithm.