[Pw_forum] FW: cell optimization results seems to be a little strange

Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu
Thu Jul 3 19:21:07 CEST 2008


On Fri, 4 Jul 2008, vega lew wrote:

VL> > Think about, why you set etot_conv_thr so tight (don't mix it up with 
VL> > conv_thr, which seemed to be sloppy for this calculation ), while you set 
VL> > forc_conv_thr to be really sloppy? (Btw, is there any reason to set the 
VL> > cutoffs/thresholds to such precision?)
VL> 
VL> Many litterature doing so. The force threshold is usually set as 0.03eV and 0.05eV
VL> for larger system. the cutoff in literature can be 380 or 400eV sometimes.
VL> I'm only doing repeat work and learning how to do a perfect calculation.
VL> Also the conv_thr and etot_conv_thr is taken from CASTEP calculation directly. Others

please note that convergence parameters can be _very_ arbitrary
between different codes and they can be in very strange units.


VL> doing so of the same system is the only reason. I'm a beginner of 
VL> PWscf and DFT calculation. So I think doing repeat work of others 
VL> and discuss with skilled people may be the best way for me. What I'm 
VL> doing now is something more like a homework rather than the research 
VL> work. I'm learning to using PWscf. Thank you for your help.

before going any further with running big calculations. try
smaller ones. and have you checked the requirements for each 
pseudopotential? is your stress tensor converged wrt. wavefunction 
_and_ density cutoff? ...and k-points? you are using vanderbilt 
ultrasoft pseudopotentials and they can require unexpectedly high 
density cutoffs. if i remember correctly. you are also not mentioning 
whether the reference you are comparing to is using the _same_ 
functional that you are using. different functionals 
tend to over/underestimate lattice parameters differently.

cheers,
   axel.

VL> 
VL> 
VL> Vega Lew
VL> PH.D Candidate in Chemical Engineering
VL> 
VL> State Key Laboratory of Materials-oriented Chemical Engineering
VL> College of Chemistry and Chemical Engineering
VL> Nanjing University of Technology, 210009, Nanjing, Jiangsu, China
VL> 
VL> 
VL> 
VL> _________________________________________________________________
VL> Connect to the next generation of MSN Messenger 
VL> http://imagine-msn.com/messenger/launch80/default.aspx?locale=en-us&source=wlmailtagline

-- 
=======================================================================
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.


More information about the Pw_forum mailing list