[Pw_forum] convergence issues

Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu
Wed Jan 28 02:12:21 CET 2009


On Tue, 27 Jan 2009, Jonas Baltrusaitis wrote:

JB> WIth all being said, how do I test pseudopotentials, grid and cutoff 
JB> for my system? My understanding is this:
JB> 
JB> select pseudopotential, use it on a real system, have grid fixed, 

using a real system at the beginning is serious waste of computer time.

an initial test would be to converge the lattice constant of 
a corresponding bulk cystal system for each element using E/V curves.

nicola marzari posted some examples on carbon a while ago, IIRC.
not sure whether they made it to the wiki, but a search through
the mailing list archives, should get you there.

for ultrasoft psps, you increase the wavefunction cutoff (leaving
density cutoff at the default of 4*ecutwfc) until convergence,
keep that density cutoff and lower ecutwfc until it starts to
diverge (usually at 25-35 ry).

JB> calculate total energy at , say, 20 Ry and several ecutrho values, 
JB> the increase cutoff value, calculate total energy at sevarl ecutrho 
JB> again etc. The take the lowest cutoff and ecutrho values (when it 
JB> doesn;t change anymore) and increase grid while having them 
JB> constant.

nope. grids are determined by the cutoffs.

JB> Am I thinking right? I am a cluster type of guy so it's somewhat out 
JB> of my understanding

get a good text book then!
asking questions on a mailing list is no replacement 
for learning the basics. sorry. TANSTAFL.

cheers,
   axel.


JB> thanks
JB> 
JB> JOnas
JB> University of Iowa
JB> 
JB> 

-- 
=======================================================================
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 users mailing list