本計晝擬探討在不同温度下，計算金屬叢集之各原子功率光譜密度(power spectral density)，以及歸一化之結構相似度函數(normalized similarity score function)，其中後者排除掉計算功率光譜密度時相同之各原子，再配以超快狀似(ultra-fast shape recognition)方法，來暸解此類叢集之温度性質。從電腦所收錄的數據，本計晝試著取得叢集之微觀動態行為。另外，我們也希望發展出一套演算法可在極小温度範圍內，比如說間隔10 K，研究顯現瀲烈變化之物理量。利用本演算法我們將得以診斷及分析更多金屬叢集之原子勳態變化。 The temperature dependences of the power spectral density of individual atoms in a cluster and the normalized similarity score function minus the same individual atoms calculated by the ultrafast shape recognition method are proposed in this project to study the thermal properties of metallic clusters. We plan to extract from the simulation data the microscopic dynamics. With an ambitious goal to studying physical quantities whose thermal variations display anomalous behaviors within a narrow temperature window, we plan to develop a refined algorithm so that simulations can be performed at a very small temperature interval, say 10 K. The time-resolved atomic dynamics can therefore be understood at considerable details. 研究期間：10008 ~ 10107
