Regensburg 2013 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

TT: Fachverband Tiefe Temperaturen

TT 39: Focused Session: Frontiers of Electronic Structure Theory 4 (jointly with HL and O)

TT 39.2: Talk

Wednesday, March 13, 2013, 11:00–11:15, H36

DFT+U(ω): A simplified approach for dynamical Hubbard corrections to DFT: — •David D. O'Regan and Nicola Marzari — Theory and Simulation of Materials, EPFL, Switzerland.

Numerous successful techniques have been developed to date, such as DFT+U, in which the screened Coulomb interactions, underestimated by approximate density functionals, are described more accurately via a mapping onto the Hubbard Hamiltonian. Charge screening is a dynamical process, generally, and so to fully realise the capability of such methods for improving optical and quasiparticle spectra, the Hubbard U describing these interactions must gain a frequency dependence.

We introduce a simple and inexpensive approach, named DFT+U(ω), and readily implementable within an existing DFT+U or constrained-DFT code, in which the dynamical U tensor appropriate to the rotationally-invariant DFT+U functional is computed and used to correct DFT or static DFT+U spectra perturbatively. The rotationally-invariant DFT+U(ω) self-energy interpolates between static DFT+U and GW. We recast the density-functional linear-response approach for the static U, where it is defined as an energy curvature, within the language of many-body perturbation theory. Here, its dynamical generalisation, and its relationship to methods such as constrained RPA, becomes readily apparent. A plasmon-pole type model is used for the inverse dielectric function, whereby low-energy parameters are computed using the appropriately renormalised density-functional linear-response, and high-energy parameters are inexpensively approximated via independent-particle RPA or ALDA.