High-Accuracy and Low-Cost Electronic Structure Theory for Strongly Correlated Systems

Abstract

Electronic structure theory is a powerful tool to study chemical systems, but it is very challenging to apply accurately to strongly correlated systems. Despite significant recent progress, a high-accuracy and low-cost electronic structure theory for strongly correlated systems is not available. This is partly related to the scarcity of accurate reference data for developing and testing improved theories, but it is also due to insufficient fundamental understanding of the ingredients necessary for a theory of a strongly correlated system to be accurate. This thesis addresses these issues. It includes benchmark studies on the spin-splitting energy of transition metals and their use to test a variety of wave function theories and density functionals in Kohn–Sham density functional theory, providing guidance as to which electronic structure method might be accurate for practical calculations, as well as providing accurate reference data for future theory development. It also contains the development of several analysis tools for improving fundamental understanding of existing theories. The results provide insight into the sources of errors in correlation energies and into improvements of existing theories. Finally, based on the discoveries of this work, a new theoretical framework named multiconfiguration density-coherence functional theory (MC-DCFT) is presented. The new theoretical framework provides an alternative approach to combining multiconfiguration wave functions with density functional theory, making it a promising method for further development.