普适的基于能量的分块局域激发态聚类算法

在普适的基于能量的分块(GEBF)方法的框架下, 大体系的局域激发(LE)能可通过一系列活性子体系激发能的线性组合近似得到, 从而有效降低了计算的时间标度. 然而, 在体系的局域激发具有多个激发态的情形下, 如何有效识别所有活性子体系的激发特征并将其组合是一个挑战. 提出了一种基于局域激发态聚类的算法. 该方案基于空穴-电子分析和基于密度的聚类(DBSCAN)机器学习算法, 可以自动地聚合不同子体系中最相似的激发态并组合得到相应的局域激发态能量或激发能. 结合该算法改进的LE-GEBF方法在荧光分子衍生物、 荧光染料-水团簇及绿色荧光蛋白模型体系的计算中均获得了令人满意的结果. 该算法有望大大提升LE-GEBF方法在计算局域激发时的稳定性和准确性, 并可以有效处理吸收光谱具有多重峰的大体系.

Generalized Energy-based Fragmentation Clustering Algorithm for Localized Excited States

Generalized energy-based fragmentation（GEBF） approach can approximately express the local excitation（LE） energy of large system as a linear combination of the excitation energies of a series of active subsystems. Thus， it reduces the computational scaling of the excited-state calculations for large systems. Howe-ver， it is still a challenge to effectively identify the excitation characteristics of all active subsystems and combine the excitation energies when the total system has multiple localized excitations. Here， a clustering algorithm for localized excited states was proposed. This scheme is based on hole-electron analysis and density-based spatial clustering of applications with noise（DBSCAN） algorithm in machine learning， which can automatically collect the excited states with maximum similarity in different subsystems and combine the corresponding localized excited-state energies or excitation energies. With this algorithm， the improved LE-GEBF approach has shown satisfactory results in various systems including derivatives of fluorescent molecules， fluorescent dye-water clusters， and green fluorescent protein models. This algorithm is expected to greatly improve the stabilities and accuracies of the LE-GEBF approach for calculating localized excitations， and can effectively treat large systems with multiple peaks in the absorption spectrum.