Static Correlation Density Functional Theory

Abstract

Over the years, several schemes have been proposed to describe multireference systems with Kohn-Sham Density Functional Theory. Problematic is the combination of two aspects: the Kohn-Sham reference wavefunction is usually taken to be a single Slater determinant, and approximate exchange-correlation functionals are typically derived form the local density approximation. In this work, we develop a theoretical framework that foregoes the single Slater determinant and instead employs thermal states as reference states for zero-temperature interacting systems. We provide convenient definitions of static and dynamic correlation functionals via an adiabatic connection approach. The formalism and computational results indicate that the entropic term of the thermal reference state is a good approximation to the static correlation functional. Hence, this work validates several reported results in the literature and motivates additional developments of static correlation density functionals.

Unveiling the Low-Lying Spin States of [Fe3S4] Clusters via the Extended Broken-Symmetry Method

“…The Broken-Symmetry (BS) method is a common approach within DFT. This method calculates the energy difference between the high-spin state and the BS state, allowing for an estimation of the magnetic coupling constant, J, between spin centers [29]. However, because the BS state is not an eigenstate of the total spin operator Ŝ2 [30], some errors may occur when using the BS method to calculate the magnetic coupling constant J. Additionally, due to the strong magnetism of transition metal clusters and the high degree of dynamic and static correlation among the 3d orbital electrons of the central atom [31], clusters exhibit multiple degenerate states, with the ground state (GS) typically being a low-spin state.…”

nablaDFT: Large-Scale Conformational Energy and Hamiltonian Prediction benchmark and dataset

“…[14][15][16][17][18] The third direction is to use neural networks to model the XC functional for high accuracy DFT. [19][20][21][22][23][24][25][26] The general framework of ab initio molecular property prediction consists of two steps: first compute the electron structure of a specific molecular conformation or a set of conformations, and then calculate desired properties based on the results of the first step. The second step is relatively simple, but total computational complexity could be too high depending on the method used on the first step.…”

Application of neural network for exchange-correlation functional interpolation

“…NN was first utilized as a functional form for XC potential by Tozer et al [20]. After that, several studies have been addressed the possibility of using NN to approximate XC functionals form [21][22][23][24][25][26]. Work by Nagai and co-authors [22] is especially worthy of note.…”