Group 1 - Correlation

Task of WG 1: The description of electronic correlations

 WG leader: Jan Minár (DE), Deputy:  Gian-Marco Rignanese (BE)

 Most electronic structure calculations supplying the basis for spectroscopic investigations are performed in the framework of density functional theory (DFT). However, in many cases the local density approximation (LDA) is not sufficient to describe the electronic correlations properly. To overcome this many promising new developments have been made in DFT, for example the LDA+U method which incorporates correlation effects into LDA by an additional Coulomb term in a way that allows a straightforward link to subsequent spectroscopy calculations. There are also several implementations of the GW-method for moderately correlated materials that express correlation effects in terms of a wave vector dependent self-energy.

However the direct link to spectroscopic properties is so far rarely made. The task of working group 1 is to promote the development of new tools for the description of electronic correlations  and their implementation in spectroscopy codes.

For moderately and strongly correlated systems dynamical mean field theory (DMFT) has become a new and powerful tool. First applications in X-ray absorption and valence band photo emission were very successful and therefore supply a very promising starting point for further work in this direction. Recently, a unifying method has been proposed to account for electronic correlations both in the ground state and excited states of correlated systems in the framework of a configuration interaction (multi-channel), real-space multiple scattering scheme. It will offer a further, alternative possibility to tackle the correlation problem.

The activities of WG 1 will be based on the common data format introduced by the working group platform (WG 4) together with the corresponding input and output routines. Interfaces for these will be incorporated into the various program packages to allow first of all the transfer of the self-consistent potential created by any electronic structure code to any spectroscopy code. Additional interface routines will be developed to also transfer a complex and energy-dependent and possibly wave vector dependent self-energy being of interest in the cases of GW- and  dynamical mean field theory (DMFT)-based calculations. To implement DMFT in electronic structure codes as a new scheme to deal with electronic correlations a standard interface of DMFT- solvers will be provided to available packages. The various corresponding code developments will be mutually supported by providing data sets for testing and benchmarking. For example, using exact diagonalisation scheme as a DMFT-solver gives access to atomic multiplet effects in solids. This approach will be checked against schemes using an atomic multiplet description within the crystal field approximation and against accurate first principles quantum chemical embedded cluster methods. This way it will ultimately offer an alternative for the computation of L2,3-edge spectra of transition metals. In addition it will be exploited to provide the necessary input for the recently developed multi-channel multiple-scattering scheme.