[Pw_forum] "charge is wrong"

Paolo Giannozzi giannozz at nest.sns.it
Wed Apr 23 19:08:57 CEST 2008


Eyvaz Isaev wrote:

> As you can see you got charge deficieny, meaning that
> bands number is not sufficient. You did not specify
> any occupations keyword, so your system is considered
> as insulator (or semiconducting) for which only
> nbnd=(number of valence electros divided by 2) is
> used. It might be that during relaxation process your
> system turns to be metallic for which one should use
> more bands number. 
> So, try nbnd=490/2 + 10= 255 (in &system)

this might be a good suggestion, and one that is valid in
any case: having a few more bands and a small broadening
is a a good insurance against nasty phenomena such as bad
or no convergence.

However:  I run the same job (using gamma instead of 0,0,0
so that it runs faster) with the current development version:
it seems to work. So it might also be a problem with the
previous version of the parallel subspace diagonalization.

Paolo
-- 
Paolo Giannozzi, Democritos and University of Udine, Italy


More information about the Pw_forum mailing list