Efficient self-consistent treatment of electron correlation within the random phase approximation.

Area of Science:

• Computational Chemistry
• Quantum Mechanics
• Electronic Structure Theory

Background:

• Accurate treatment of electron correlation is crucial for predicting molecular properties.
• Existing methods often involve approximations that limit accuracy or computational efficiency.

Purpose of the Study:

• To develop and evaluate a self-consistent Kohn-Sham (KS) method incorporating the adiabatic-connection fluctuation-dissipation theorem with direct random phase approximation (dRPA).
• To assess the performance of this method for calculating electronic correlation and its impact on various molecular properties.

Main Methods:

• Employed the direct random phase approximation (dRPA) within a self-consistent Kohn-Sham framework, neglecting the exchange-correlation kernel for correlation energy and potential.
• Included exact exchange energy and a local multiplicative KS exchange potential.
• Utilized Gaussian basis sets, achieving a favorable O(N^4) computational scaling for molecular applications.

Main Results:

• Self-consistent dRPA correlation potentials closely matched exact reference potentials.
• Eigenvalues of highest occupied molecular orbitals aligned well with experimental ionization potentials.
• Self-consistency offered limited improvement for total energies, reaction, and non-covalent binding energies in pure dRPA.
• Combining self-consistent dRPA orbitals with the EXXRPA+ approach significantly enhanced total energies, reaction energies, and non-covalent binding energies, achieving accuracy comparable to high-level methods like coupled cluster singles doubles.

Conclusions:

• The self-consistent dRPA method provides accurate correlation potentials and ionization potentials.
• While self-consistency alone has limitations for energy predictions, it serves as an excellent starting point for more advanced methods like EXXRPA+.
• The combination of self-consistent dRPA and EXXRPA+ offers a computationally efficient route to high-accuracy results for molecular energies and binding.