Development of efficient means to describe dynamic electron correlation effects in strongly correlated molecular systems

This project aims at the development of multireference
electronic structure methods that provide access to accurate and reliable
electronic energies and structures for molecular systems with many strongly
correlated electrons. In the first part of the enivsaged work, a partially
uncontracted reformulation of the well-established N-electron valence
perturbation theory will be developed and implemented. Through usage of
the heatbath CI-rationale that has been succesfully applied in the context of
selected configuration interaction (SCI) methods, the computational cost of the
proposed will be reduced while an efficient implementation will allow for
favorable parallelization properties.
The second part of this project encompasses the development and implementation
of the recently introduced adiabatic connection approach to calculate dynamic
electron correlation effects on top of selected CI reference wavefunctions. With a
working implementation at hand, the method will be extended to compute nuclear
gradients at the same level of theory.
Importantly, the proposed developments will result in openly available computer
codes that allow for the investigation of molecular systems with many strongly
correlated electrons on a physically sound basis. A set of initial test calculations
on chemically relevant compounds will demonstrate the capabilities and limitations
of the developed programs.