The ABINIT Users Mailing List ( CLOSED )

[Prasenjit Sen <psen@unity.ncsu.edu>]

Dear ab-init usres:
     I am having trouble converging the eigenvalues and eigenfunctions in
a band sturcture calculation. Any help to sort this out will be highly
appreciated.

I did a self-consistent calculation for a bulk solid system, and read in
the converged charge density for the purpose of band structure
calculation. In the band structure calculation from \Gamma to X point, I
use a total of 41 k-points. As convergence criterion I use
tolwfr  1.0d-12  and I allow
nstep   80

What I typically find is that for the first ten k-points starting from
\Gamma, the wavefunctions converge with these parameters. However, for the
remaining k-points, even 80 steps are not enough to get convergence. It's
a few states at the top that do not converge. (Of course they converge
most slowly). So my question is:

Should I do something different like use a different convergence
criterion, or should I simply increase nstep parameter?

Thank you so much.

Sincerely,
Prasenjit Sen.

        *********************************************************
        *  Prasenjit Sen             *                          *
        *  Dept. of Physics(Box 7518)*                          *
        *  North Carolina State Univ.*  psen@unity.ncsu.edu     *
        *  1005 Capability Drive     *  Phone: (919)513-4839(W) *
        *  Room# 244                 *  Fax:   (919)515-4496    *
        *  Raleigh NC 27695  USA     *                          *
        *********************************************************