In this paper we propose a numerical method to localize many-electron excited states. To characterize the electronic structure of the electronic excited states of a system, quantum chemistry methods typically yield a delocalized description of the excitations. Some a priori localization methods have been developed to provide an intuitive local picture of the excited states. They typically require a good strategy to separate the system of interest from its environment, or a set of a priori localized orbitals, that may reduce their computational accuracy. Here, we introduce an a posteriori method to localize delocalized many-body excited states directly obtained from quantum chemistry calculations. A localization metric for the excited states is defined from their representation as electron–hole pairs, which is encoded in the transition density matrix. This novel a posteriori strategy thus allows to localize excitons within a volume around selected fragments of a complex molecular system without tempering with its quantum chemical treatment. The method is tested on π-stacked oligomers of phenanthrenes and pyrenes. It is found to efficiently localize and separate the excitons according to their character while preserving the information about delocalized many-body states at a low computational cost.

中文翻译：

---

通过同时对角化对多体激发态进行后验定位

在本文中，我们提出了一种定位多电子激发态的数值方法。为了表征系统电子激发态的电子结构，量子化学方法通常会产生对激发的离域描述。已经开发了一些先验定位方法来提供激发态的直观局部图像。他们通常需要一个很好的策略来将感兴趣的系统与其环境或一组先验局部轨道分开，这可能会降低他们的计算精度。在这里，我们引入了一种后验方法来定位直接从量子化学计算中获得的离域多体激发态。激发态的定位度量是根据它们作为电子-空穴对的表示来定义的，它被编码在过渡密度矩阵中。因此，这种新颖的后验策略允许在复杂分子系统的选定片段周围的体积内定位激子，而无需对其量子化学处理进行回火。该方法在菲和芘的π堆叠低聚物。发现可以根据激子的特性有效地定位和分离激子，同时以低计算成本保留有关离域多体状态的信息。