Frozen density embedding with hybrid functionals

Abstract

The Kohn-Sham equations with constrained electron density are extended to hybrid exchange-correlation (XC) functionals. We derive the frozen density embedding generalized Kohn-Sham (FDE-GKS) scheme which allows to treat the nonlocal exact-exchange in the subsystems. For practical calculations we propose an approximated version of the FDE-GKS in which the nonadditive exchange potential is computed at a semilocal level. The proposed method is applied to compute the ground-state electronic properties of small test systems and selected DNA base pairs. The results of calculations employing the hierarchy of XC functionals BLYP/B3LYP/BHLYP and PBE/PBE0 are presented, in order to analyze the effect of nonlocal exchange contributions, and compared with reference coupled-cluster singles and doubles results. We find that the use of hybrid functionals leads to a significant improvement in the description of ground-state electronic properties of the investigated systems. The semilocal version of the FDE-GKS correctly reproduces the dipole and the electron density distribution of the exact GKS supramolecular system, with errors smaller than the ones obtained using conventional semilocal XC functionals.

Subsystem density functional theory with meta-generalized gradient approximation exchange-correlation functionals

“…is the embedding potential [44] with I = A, B and II = B, A, respectively (this convention will be used throughout). In this work we consider for F HK the following partition:…”

“…At this point, following the GKS scheme [44,77], we introduce, for each subsystem (for example I) an auxiliary system of particles having the following properties: (i) it has the same ground-state density as our original embedded subsystem I; (ii) it is described by a single Slater determinant Φ I ; (iii) the ground-state energy is the minimum of the energy functional…”

“…43. Moreover, several works have extended the subsystem formulation of DFT beyond the conventional Kohn-Sham (KS) framework, considering e.g., hybrid functionals [44], embedded interacting wave functions [45], orbital-dependent effective exact exchange methods [21,46], or density matrix [47]. Nevertheless, to date, no attempt has been made to consider subsystem DFT calculations using meta generalized gradient approximation (meta-GGA) XC functionals.…”

The electronic spectrum of CUONg4 (Ng = Ne, Ar, Kr, Xe): New insights in the interaction of the CUO molecule with noble gas matrices

“…68) xc for this contribution, following earlier work on orbital-dependent functionals. 46,[78][79][80] For the kinetic energy component of the embedding potential we tested out a number of different functionals: the Thomas-Fermi [81][82][83] functional, the NDSD functional of Wesołowski and co-workers 84 (which contains a TF component but was developed especially for FDE), the PW91K (Ref. 85) functional, and PW91K with the long-distance correction proposed by Jacob and Visscher 78 (PW91K-CJCORR).…”

eQE: An open‐source density functional embedding theory code for the condensed phase

“…[31][32][33] In addition, Turbomole [34] has its own implementation by the Della Sala group. [35][36][37][38][39] We also mention here that other embedding methods, which can be categorized as exact density embedding, exact orbital embedding, or electrostatic embedding, are now found in ADF, [40] MOLPRO, [41][42][43][44][45] Q-Chem, [46,47] CP2K, [48] NWChem, [49] and GAMESS. [50] In this work, we present a novel implementation of the FDE approach that aims at filling the following gap that has persisted over the years, namely, the absence of a code that: (1) has a proven strong parallel efficiency that consistently outperforms semilocal KS-DFT, (2) has the ability to Figure 1 [12] The k-point grids and simulation cells (basis sets) are subsystem-specific, achieving the best performance.…”