Restricted active space configuration interaction methods for strong correlation: Recent developments

In this review we outline the theory and recent progress of the restricted active space configuration interaction (RASCI) methodology within the hole and particle approximation. The RASCI is a single reference approach based on the splitting of the orbital space in different subsets, in which the target CI space is expressed by the concomitant number of electrons and empty orbitals in each subspace. Initially, the method was born as a spin complete version of the spin-flip ansatz able to deal with any number of unpaired electrons. Since then, the method has experienced several improvements related to its theoretical foundations, its efficient implementation, in the characterization of RASCI wave functions, and in the calculation of molecular properties. It has shown to be especially suitable for the characterization of medium to large molecular diradicals and polyradicals, and to the study of photophysical processes involving open-shell species and/or multiexcitonic states, like in the singlet fission reaction. But, despite this success, several limitations emerge from the sever truncation of the excitation operator, which might translate into inaccurate relative energies between different regions of the potential energy surface, and overestimated (or underestimated) excitation energies. These errors are associated with the lack of dynamic correlation, which has motivated the development and implementation of various improvements based on multi-reference perturbation theory and to blend the RASCI wave function with density functional theory through range separation of electron-electron interactions. Finally, we discuss some of the properties available from RASCI wave functions and give potential future developments. This article is categorized under: Electronic Structure Theory > Ab Initio Electronic Structure Methods Software > Quantum Chemistry Quantum Computing > Theory Development

Automated summary

The RASCI methodology is suitable for characterizing medium to large molecular diradicals and polyradicals, as well as studying photophysical processes involving open-shell species and multiexcitonic states. However, limitations arise from the truncation of the excitation operator, leading to inaccuracies and errors that require improvements to be addressed.