forum@abinit.org
Subject: The ABINIT Users Mailing List ( CLOSED )
List archive
- From: paul.boulanger@umontreal.ca
- To: forum@abinit.org
- Subject: Re: Re: Re: [abinit-forum] geometry optimization
- Date: Mon, 11 Jun 2007 09:58:47 +0200
Hi,
I peroused your input file and I found a possible error. Your SCF convergence
criteria, toldfe 1d-5, is probably not sufficient to get a reasonnable result
for forces. Hence, when relaxing the system with a criteria of tolmxf 5d-4,
you are probably sampling noise.
This can be checked in your output file. Look at the convergence of the
forces in the SCF results. I think a better parameter for the SCf
convergence should be toldff 1d-6. But, if your atoms are in highly symetric
positions (forces are all zero), you will be forced to use toldfe close to
1d-10. But, I don't think this is your case.
Paul
- Re: Re: [abinit-forum] geometry optimization, xd0533, 06/11/2007
- Re: Re: [abinit-forum] geometry optimization, Anglade Pierre-Matthieu, 06/11/2007
- Re: [abinit-forum] geometry optimization, Masayoshi Mikami, 06/11/2007
- <Possible follow-up(s)>
- Re: Re: Re: [abinit-forum] geometry optimization, paul . boulanger, 06/11/2007
- Re: Re: [abinit-forum] geometry optimization, Anglade Pierre-Matthieu, 06/11/2007
Archive powered by MHonArc 2.6.16.