Development of an Automatic Code Generator and Implementation of Multireference Equation of Motion Coupled-Cluster Theory in the ORCA Program Package
Huntington, Lee Michael
MetadataShow full item record
Over the past 50 years, single-reference coupled-cluster theory has emerged as a cornerstone of quantum chemistry. While it is an accurate methodology for the calculation of the properties of the electronic states of many systems, there are still many strongly-correlated (multireference) systems which cannot be adequately treated with single-reference coupled-cluster theory. Hence, in the past four decades, there have been significant efforts to develop multireference generalizations of coupled-cluster theory to treat such systems. In this work, we review some of the major developments in single-reference and multireference molecular electronic structure theory. We discuss the details of the Multireference Equation of Motion (MR-EOM) coupled-cluster approach, developed in the Nooijen group, and introduce a new variant which makes use of a Hermitizing transformation. The MR-EOM methodology constitutes a transform and diagonalize approach to electronic structure theory, that is applicable to both ground and excited states. A major topic of this thesis concerns the development of an automatic code generation tool, that has been used to implement the MR-EOM approach in the ORCA quantum chemistry software package. The implementation in ORCA is employed for the characterization and calculation of the excitation energies of transition metal complexes. We also introduce an orbital selection scheme which can be used to extend the applicability of the MR-EOM approach to larger systems for the calculation of excitation spectra. A variety of MR-EOM approaches are then considered in benchmark applications to organic molecules and the various approximations, introduced in the ORCA implementation of MR-EOM, are studied for several transition metal complexes. Finally, we discuss how the implementation in ORCA might be improved in the future, in order to push applications to larger systems and larger active spaces.