Range-Separated Double-Hybrid Functional from Nonempirical Constraints

Abstract

On the basis of our previous developments in the field of nonempirical double hybrids, we present here a new exchange-correlation functional based on a range-separated model for the exchange part and integrating a nonlocal perturbative correction to the electron correlation contribution. Named RSX-QIDH, the functional is free from any kind of empirical parametrization. Its range-separation parameter is set to recover the total energy of the hydrogen atom, thus eliminating the self-interaction error for this one-electron system. Subsequent tests on some relevant benchmark data sets confirm that the self-interaction error is particularly low for RSX-QIDH. This new functional provides also correct dissociation profiles for charged rare-gas dimers and very accurate ionization potentials directly from Kohn-Sham orbital energies. Above all, these good results are not obtained at the expense of other properties. Indeed, further tests on standard benchmarks show that RSX-QIDH is competitive with the more empirical ωB97X-2 double hybrid and outperforms the parent LC-PBE long-range corrected hybrid, thus underlining the important role of the nonlocal perturbative correlation.

Range-separated double-hybrid density-functional theory with coupled-cluster and random-phase approximations

“…Other related double-hybrid approximations have been proposed which combine a long-range HF exchange term with a full-range MP2 correlation term [16], a fullrange HF exchange term with a long-range MP2 correlation term [17,18], or a HF exchange term based on the decomposition of Eq. (1) with a full-range MP2 correlation term [19].…”

Double-Hybrid DFT Functionals for the Condensed Phase: Gaussian and Plane Waves Implementation and Evaluation

“…Further, there are DHDFs with range-separated correlation functionals [56]. In this article, we will not focus on DHDFs with range-separated correlation functionals and refer to the literature [40,[57][58][59][60] for more details.…”

Computation of covalent and noncovalent structural parameters at low computational cost: Efficiency of the DH‐SVPD method

“…[11][12][13] Others are based on the range-separation exchange (RSX) scheme [14] to (partially) cure the self-interaction and delocalization errors [15] (SIE and DE, respectively). [16][17][18][19] However, all of them conserve similar features, that is, a fraction of EXX generally larger than 50% and a fraction of PT2 correlation ranging between 10% and 40%, which allow them to reach the chemical accuracy when modeling a large panel of properties ranging from ground-state energies [20] to structures [21,22] and extending to singlet-singlet vertical excitation energies. [23,24] Despite their undeniable robustness, DHs remain more computationally demanding than standard density functionals.…”