动态格点搜索的改进算法用于团簇结构快速优化  

An improved dynamic lattice searching method for fast optimization of clusters

作　　者：吴夏[1,2] 程文[1] 

机构地区：[1]安庆师范学院化学化工学院,安徽安庆246011 [2]光、电、磁功能配合物和纳米配合物安徽省重点实验室,安徽安庆246011

出　　处：《计算机与应用化学》2014年第8期917-920,共4页Computers and Applied Chemistry

基　　金：国家自然科学基金资助项目(21203002、21171008);安徽省自然科学基金资助项目(1308085QB29)

摘　　要：结合高效的动态格点搜索(DLS)算法与扰动操作(Perturbation Operation)提出一种新的改进方法(DLS-PO),用于确定团簇的最低能量结构。针对一个特定构型,DLS算法总能给出其对应搜索空间的最规则结构。然而,一次失败的DLS优化将消耗大量的运算资源。为此,采取原子移动和结构旋转的扰动操作成功地改变了构型,再结合后续的DLS操作,提高了优化效率。将该算法用于原子数高达309的Lennard-Jones团簇及100原子NP-B函数铝团簇的结构优化。优化结果显示相比于DLS算法,DLS-PO算法更为高效。Global optimization is a challenging problem for structural optimization of clusters. An algorithm based on dynamic lattice searching （DLS） and perturbation operation （PO） is proposed, which is named as DLS-PO. For a defined configuration, DLS algorithm can give a regular configuration related to the searching space. However, much CPU time is consumed for one unsuccessful run of DLS. PO, e.g., atomic movement and rotate operation, can change a configuration, and the following DLS can improve the efficiency. The algorithm is applied to the optimization of Lennard-Jones （LJ） clusters up to 309 atoms and NP-B functional AI clusters up to 100 atoms. Results show that the method improves the successful rates of DLS.

关 键 词：全局优化 Lennard-Jones团簇 铝团簇 动态格点搜索 免疫优化算法 

分 类 号：TQ015.9[化学工程]