[Wien] Re:ghost bands come from large charge fluctuation

[Peter Blaha]

Hi,

I did rerun your testcase and I could verify your problem. Aparently the 
density from dstart is not correctly normalized and at the moment I do not 
know why.

Anyway, for your calculations you can proceed as follows:

x dstart -c
run_lapw -i 1
Check case.scf, if you have normalization problems. If yes:
rm case.clmsum_old
x mixer
run_lapw

This should work.

>       You suggest are some part correct. As you suggest I recalculate some structures,and accept the struct created by sgroup, such for the superlattices along [110] direction the electrons normalized in the iteration, but for superlattices along [111] direction it still lost three electrons in the first iteration, and got a ghost band, even I accept the struct created from sgroup. The input files are in the attachment, Could you please check it for me?


                                      P.Blaha
--------------------------------------------------------------------------
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/
--------------------------------------------------------------------------