search

TRIQS / triqs_0.x

DEPRECATED -- This is the repository of the older versions of TRIQS
Other
11 stars 9 forks source link

Implementation of add_dmft_intr and changes to fourir.F #146

Closed rainbac closed 10 years ago

rainbac commented 11 years ago

I am currently working on doing charge self-consistent LDA+DMFT with WIEN2k and a different DMFT code than TRIQS and was wondering about the changes to SRC_lapw2/fourir.F of Wien2k that were made (I hope it's ok to ask this here:)
As I understand it, for every k-Point, fourir.F FFTs, sums, and FFTs back all the interstitial coefficients below the Fermi energy to get the density. Because it only considers bands whose energies at the k-point are below EFermi, the index je of the topmost band can sometimes be below the bottom index of the bands considered in DMFT. Now the "QDMFT" changes make sure that dmftdm(ik)%mat weighted elements are added to the density no matter the initial value of je. Did I understand this correctly and how is it justified to add contributions from bands above EF?

Kind regards, Rainer

leopo commented 11 years ago

With the DMFT self-energy added the KS LDA eigenstates are not anymore true egienstates of the problem, and the states are not filled in accordance with KS eigenenergies up to the corresponding Fermi level as it was in the LDA case. Instead the DMFT density matrix is generally not zero for all bands included in DMFT calculations (i.e. in our approach those located within a certain energy window and used for the construction of Wannier orbitals representing correlated states). Hence, contributions from all those bands should be included into the renormalized DMFT charge density.

rainbac commented 11 years ago

Thank you very much for your answer