[Pw_forum] about convergence criterion in supercell

[Axel Kohlmeyer]

On Wed, 16 Jul 2008, lfhuang at theory.issp.ac.cn wrote:

LFH> Dear all friends:

dear <insert name here>,

LFH>   I have a question on convergence criterion in supercell, which is:
LFH> 
LFH>      For unit cell, I define "conv_thr" and "tr2_ph" to be 1*d-12 and
LFH>      1*d-14,respectively. Then I take a 4*4*4 supercell, and is it
LFH>      precise enough to define 4*4*4*d-12 and 4*4*4d-14 for "conv_thr"
LFH>      and "tr2_ph" respectively? If it is not, what is the reason?

perhaps you should first explain _your_ rationale for lowering
the convergence threshold when increasing the system size.

a convergence parameter that would depend this much on 
the system size would not be a good choice. 

in any case, the rationale to choose a specific convergence criterion
should be based on the convergence of the property that you are 
interested in. i.e. if you want to do a geometry optimization, the
forces have to be converged accurately enough, not only with respect
to wavefunction/density-cutoffs, k-point sampling but also relative
to the wavefunction convergence. similarly, for other calculation types.

cheers,
   axel.

LFH> 
LFH> Thanks a lot for your time to read my letter!
LFH> And I would like greatly appreciate the one who can give me some resonable
LFH> rationale!
LFH> 
LFH> Best Wishes!
LFH> 
LFH> 
LFH> _______________________________________________
LFH> Pw_forum mailing list
LFH> Pw_forum at pwscf.org
LFH> http://www.democritos.it/mailman/listinfo/pw_forum
LFH> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.