Changing the paradigm for strong electron correlation in quantum chemistry

Funded by: National Centre of Poland (NCN), program OPUS 18

After decades of quantum chemistry methods development, efficient and accurate description of electronic structure of systems featuring strongly correlated electrons remains a challenge.

Recently, new ideas have emerged on how to treat multireference effects in chemical systems. The prominent example is density matrix renormalization group (DMRG) technique. The widespread use of new, potentially more accurate multireference theories is hindered by lack of the so-called dynamic correlation energy, which is crucial for making not only quantitative but also qualitative predictions. Until now, the paradigmatic method used to study multireference systems is still the CASPT2 approach, which is composed of the complete active space (CAS) self-consistent field method supplemented with the second-order correlation (PT2) energy correction. Despite its limitations and flaws, CASPT2 is considered a “gold standard” for many applications in chemistry and photochemistry.

The goal of the proposed project is to develop a novel theory for dynamic correlation energy, which would be complementary with most of the cutting-edge multireference methods and would be free from problems limiting perturbation theory methods. The latter includes: overcoming the bottleneck of PT2 in terms of the number of active orbitals that can be treated, assuring size-consistency, good accuracy for both ground and electronically excited states, and avoiding intruder state or instability problems.

The above requirements are expected to be fulfilled by the method based on adiabatic connection pairing matrix theory in the particle-particle approximation. Approaching the problem of predicting correlation energy for multireference wavefunctions by abandoning perturbation methods and using the adiabatic connection formalism is a promising pioneering field, which has been initiated in our lab. The adiabatic connection theory in the particle-particle approximation has been thus far applied within the framework of density functional theory to construct correlation energy functional for ground states. We go much further than this as we plan to formulate the particle-particle approximation in the adiabatic connection pairing matrix theory for ground and excited state multireference wavefunctions, which has been never attempted.

In the second part of the project we plan development of the correlation density functional, which adapts itself to both the underlying multireference wavefunction model and a given basis set. This will be the first dual-purpose correlation functional, with a locally adjusting range-separation parameter, which at the same time recovers the correlation energy missing in the multireference model and corrects for the basis set incompleteness error.

Successful conduction of the project would open new horizons for applications of the recently developed multireference computational methods in chemistry.