inverse dielectric matrix (task=180)

  • Hello,
    I cannot find instructions on how to specify the grid of \omega and q points to calculate the inverse dielectric matrix with task=180. Will be very grateful for any hint.
    Vladimir Nazarov

  • Hi Vladimir,

    For task=180, the q-point set is the same as the k-point set. Currently, there is only one frequency (w=0) because we're still figuring out the best way to do this.

    In the meantime, you can change the subroutine 'init3' and generate any set of complex frequencies you want, and so obtain time-ordered, retarded, advanced, or Matsubara correlators.


  • Marty Blaber
    Marty Blaber

    Hi Kay, Vlad,

    So a while back Aryasetiawan wrote this paper on how to decouple the frequency and state dependence. The final idea being a peice of code which runs  just as fast on 10000 frequencies as it does on 1. (PRB 61 7172 (2000))
    Now, having said that, I thought I understood how this would be implemented, and really.. I have no clue.

    Kay, is the problem you are trying to address the one about writing down the momentum matrix elements (from Claudia's paper below) in a scheme like the one above, or am I clutching at straws?

    Ambrosch-Draxl, Sofo : Computer Physics Communications 175 (2006) 1–14

    Anyway, I'm just curious really, my favorite thing to do at the moment is q-dependent dielectric functions, and the fact that elk does them now makes me happy.


  • Hi Vladimir and Marty,

    The new version (1.2.20) has a much improved dielectric function generator, where the Coulomb regulator has been removed, and the q->0 limits are computed analytically.