New Global Optimization Algorithms For Structure Optimization Of Clusters And The Applicantion

Cluster geometry optimization is one of important subjects in the field ofcomputational physics. In cluster geometry optimization, the number of local minimatends to grow exponentially with the number of atoms in the system. It can be found thatcluster geometry optimization is is a kind of multivariable nonlinear problem and solvingmultivariable nonlinear problem is very easy to get into local extremum. So clustergeometry optimization is a very difficult global optimization problem and effectivestructure optimization methods play a crucial role. For cluster structure optimization, thispaper proposes a new method and applys it to the structure optimization of carbon cluster.The details are described as follows:Firstly, the source of the subject is introduced and the research significance of clustergeometry optimization method is explained. The history of cluster research at home andabroad, the global optimization algorithm and genetic algorithm applied in clustergeometry optimization are reviewed.Secondly, the basic theory of cluster geometry optimization is introduced.Optimization problem is described. Potential function and global optimization algorithmsfor cluster structure optimization are outlined. The basic theory and the process of geneticalgorithm applied to cluster structure optimization are summarized in detail.Thirdly, a new local optimization way is proposed based on Newton method .Thislocal optimization way and genetic algorithm stand or fall are analyzed. Considering thetwo algorithm’s complementary and traditional genetic algorithm "premature"phenomenon, a new algorithm named Adaptive hybrid genetic algorithm is put forward.Finally, the new optimization method is used to optimize carbon cluster of three totwenty. Compared to the results in other paper, It proves the new optimization methodreasonable. While compared the result with hybrid genetic algorithm’s, it shows that themethod can be very good to jump out of the local extreme value. And the adaptive hybridgenetic algorithm presents good stability over many times optimization of carbon clusterof twelve to twenty. Through comparing the result with adaptive genetic algorithm’s, it is found that the new local optimization way can effectively find local extremum.