Low Cost Generalized Coupled Cluster Theory for Static and Dynamic Correlations

点击翻译按钮获取中文摘要

Gustavo Scuseria of Rice University is supported by an award from the Chemical Theory, Models and Computational methods program in the Chemistry Division to develop novel, efficient and accurate methods to treat molecules and materials in which the motion of electrons is strongly correlated. Computational methods play a key role in understanding chemical and physical behavior and therefore can help guide the development of transformative technologies which rely on novel material properties. However, there are important problems for which the existing techniques are inadequate. Most prominently, current computational tools do not properly describe the strong correlations which are responsible for superconductivity and various important magnetic phenomena. This research seeks to remedy that deficiency by extending conventional methods in a new and unique direction. These new methods will be applied to study the strong correlations in graphitic systems, which serve as prototypes for systems displaying magnetic ordering. In the longer term, the tools developed in this proposal will be useful for other researchers in chemistry, physics, and materials science as well. A diverse group of graduate students and postdoctoral associates are involved in carrying out this research. In order to accomplish these goals, Scuseria and coworkers combine the recently developed pair coupled cluster theory (which is able to describe strong static correlations) with Lie algebraic similarity transformation (which is able to describe the remaining weak correlations needed to reach quantitative accuracy). The combination of these two techniques should describe both strong and weak correlations with reasonable expense while avoiding the unphysical symmetry breaking which more conventional methods require but which makes their predictions somewhat suspect. The applications to fullerenes, graphene nanoribbons, and nanodots will demonstrate the capabilities of this new computational approach and will help explain their unique properties.