| 摘要: | 我們應用平行電腦(PC-Cluster with MPI)以及各式最佳化演算法來計算分子叢集的結構,當分子叢集的數量在上百顆以上時,其已漸漸逼進奈米尺度,這些研究資料,為奈米實驗提供了許多初步的理論基礎,我們亦在此課題中驗證平行化計算的實用行,我們所發展出來的演算法均可以利用多顆CPU來分擔運算,使得計算效能突飛猛進,目前我們的計算已經可以進展至100顆以上,並且亦驗證其他領域的實用性,如蛋白質折疊與基礎態結構等,在本篇論文中,我們首先呈現兩個經典的最佳化演算法,隨後並介紹我們新發展的演算法,在數值計算方面,我們提供一系列初步的系統化結構分析,大小均在56顆以內,總共6個材料,分別是鹼金屬Na,K,Rb,Cs,以及四價金屬Pb,最後我們再以總數38金銅合金,並套用新發展之演算法作為結束 We present detailed numerical results on the ground state structures of metallic clusters and alloy clusters. The Gupta-like many-body potential is used to account for the interactions between atoms in the cluster. Both the genetic algorithm technique and the basin hopping method have been applied to search for the global energy minima of clusters. The good agreement found in both schemes for the global energy minima gives credence to the optimized energy values obtained. Our calculations for the ground state energies of alkali metallic clusters show regularities in the energy differences, and the cluster growth pattern manifested by this same group of clusters is generally icosahedral which is quite different from the close-packed and decahedral preferentially exhibited by the tetravalent lead clusters. Considering the inherent disparities in the electronic structures and the bulk structures in these metals (body-centered cubic for alkali metals and face-centered cubic for the lead metal), it is not unreasonable to say that the valence electrons do play a subtle role in the conformation of metallic clusters.And last,we introduce a new minimization method which contains all the advantages of the above algorithms,to examine the nanoalloy cluster,Cu-Au for total number N=38.To check all the permutation and the structure variation.In evidence,there exists the most stablize structure in an alloy with total number N=38.Our algorithm also shows the reliability to compete Basin-Hopping and Genetic Algorithm. |
