[Wien] Ghost band come from large charge fluctuation

Peter Blaha pblaha at zeus.theochem.tuwien.ac.at
Tue Mar 1 11:55:00 CET 2005

Did you "correct " case.in1. Do not use -in1new, but restore the old in1
file (manually).

You will have almost the same fluctuations for the 36 atom cell, so you must
have done something strange for the other case.
When PRATT is very small, the density (potential) must be identical to
the previous iteration and it should run. Of course, at some point you
need to increase the mixing, because otherwise you will never reach

> I also try it as you said, using PRATT and a very small mixing, but it still 
> failed. I found for a superlattice including 36 atoms, when we use PRATT and 0.01 mixing 
> parameters, we still can get a converged result, but for 48 atoms's superlattice it failed
> even we set the parameter to 0.00001. 

> By the way, when we ask the core level converged to 0.00001Ry which parameters are important.

For which "super-experiment" do you need a core-level accuracy of 0.1 meV ???
When you compare with experiment, you will see much larger differences

Continue your scf cycle (with better -cc or -ec limits) until
your core-eigenvalue stays constant the way you like it.

Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/

More information about the Wien mailing list