Search papers, labs, and topics across Lattice.
This paper introduces a four-component multireference second-order perturbation theory (4C-MRPT2) within the small tensor product distributed active space (STP-DAS) framework, enabling efficient perturbative treatments of large external spaces while maintaining compatibility with various relativistic Hamiltonians. The method demonstrates significant improvements in recovering all-electron dynamic correlation, particularly highlighting the rapid increase of the Breit contribution to correlation energy with atomic number, reaching about 4% for xenon. Additionally, it reveals that the choice of multiconfigurational reference orbitals has a weak influence on second-order correlation energies, allowing for computational savings through frozen-core and frozen-virtual approximations with minimal accuracy loss.
The Breit contribution to correlation energy can reach 4% for heavy elements like xenon, revealing critical insights into relativistic effects in quantum chemistry.
We present a four-component multireference second-order perturbation theory (4C-MRPT2) within the small tensor product distributed active space (STP-DAS) framework. The formulation is compatible with the Dirac-Coulomb (DC), Dirac-Coulomb-Gaunt (DCG), and Dirac-Coulomb-Breit (DCB) Hamiltonians, and inherits the memory-efficient and massively parallel STP-DAS algorithm, enabling perturbative treatments over very large external spaces. Benchmark calculations on noble-gas and group-13 atoms demonstrate that 4C-MRPT2 efficiently recovers all-electron dynamic correlation while providing new insight into relativistic correlation effects. The calculations show that the Breit contribution to the correlation energy increases rapidly with atomic number and reaches approximately 4% of the total correlation energy for Xe. The method also shows that second-order correlation energies are only weakly dependent on the choice of multiconfigurational reference orbitals, and that frozen-core and frozen-virtual approximations substantially reduce computational cost with minimal loss of accuracy.
