Dear Mattia,
first step would be to have LDA+U working - not sure if the V problem is from your computer, or from the V in the pwscf database (if the former, ask a friend to run your calculation on a different machine, to see if it works. Also, if you plan to do comptuational research, you might want to try and get access to some machines bigger, if not better, than a Mac). Second - I don't think there is a canned solution ready, but what I would do is something akin to what we did a few years ago (Sit, Cococcioni, Marzari, PRL06) in adding a penalty functional to the total energy functional - to force or to gently push the electronic ground state towards occupying certain orbitals. The algorithms is not implemented in pwscf, but it is implemented in CP - Patrick, cc'd here, has a copy, and there was the hope/plan to put it in the CVS. Taking the subroutines from the CP code and putting them in PWscf would not be too difficult, but would require expertise on the two codes. The CP implementation does work, though (contact Patrick), albeit Gamma sampling only, forcing you to use large supercells to get decent BZ sampling. Some kind of workaround could be this, with PWscf: get LDA+U to work, define each vanadium atom as being different (V1, V2, V3...) but with the same pseudopotential, and modify the part of the code that calculates the occupation matrix n. By changing the n for V1 so that it has 1 only for the oribtals that you want occupied, and same for V2, V3, etc..., and by choosing "judiciously" the U, you might force the system to the orbital ordering that you want. Then, restart your calculation without any constraint, and see if the electronic-state is stable. Note, you might need a true U in this second case, applied equally to all V, so the steps are first to become an expert in LDA+U (Cococcioni's thesis and papers), and then move onf rom there. It's not easy, but very interesting. nicola Mulazzi Mattia wrote: > Dear QE community, > I am currently studying charge-density wave ordered materials form the > theoretical point of view. In my system, VS2, the Vanadium planes > order in an hexagonal 2D lattice and are sandwiched by two Sulfur > planes. As a basic approach, an hexagonal unit cell with one V and two > S atoms produces a band structure that is in fair agreement with the > experimental data and is consistent with other all-electron > calculations. But both theoretically and experimentally there are > hints that the system is orbital-ordered, i.e. that the d electrons on > the Vanadium sites occupy orbitals of different symmetry on different > Vanadium sites. > In order to calculate the electronic structure of an orbitally-ordered > system I think I have to use a supercell and specify different > occupation numbers for the different Vanadium sites. The problem is > that I don't know how to do that. I tried specifying the > starting_ns_eigenvalue variable on different sites, but the converged > calculation is identical to the calculation with > starting_ns_eigenvalue left unspecified. In the INPUT_PW it is written > that the occupation matrices can be specified only when the LDA+U > scheme is used. > However, I cannot use the LDA+U on my system (Mac OS X 10.5.6) because > every time I use the Vanadium pseudopotential (either the one I > generated, ether the one on the QE database) I find a "segmentation > fault" that I cannot solve. BTW, the segmentation fault error does > not appear for Co or Fe or Ni pseudopotentials. > > Thanks everybody for the help, > > Mattia Mulazzi > > > FPR Fellow of RIKEN at Spring8 > Excitation Order Research Team > 1-1-1 Sayo-cho Sayo-gun, Hyogo > Japan > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum > -- --------------------------------------------------------------------- Prof Nicola Marzari Department of Materials Science and Engineering 13-5066 MIT 77 Massachusetts Avenue Cambridge MA 02139-4307 USA tel 617.4522758 fax 2586534 marzari at mit.edu http://quasiamore.mit.edu