[Pw_forum] cell optimization

Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu
Sun Jun 15 19:57:34 CEST 2008


On Sun, 15 Jun 2008, hania djani-ait aissa wrote:

HDA> 
HDA> dear Ralph,
HDA> nice to read from you.

HDA> I am trying new calculation with ecutoff=40Ry, and will see if the 
HDA> discontinuous energy-volume persist. however for better 

yes, but i'd expect the discontinuities to shrink. if i may make
a suggestion to approach the problem more systematically and with
less computional effort:

how about doing a test run for bulk Bi, Ti first and see what the
required cutoffs for wavefunction and desity are to get converged
results. the cutoff requirements are in general a property of the
the individual pseudopotential and the proper choice is set of cutoff
that satisfies the minimal requirements for all of them. for the 
norm conserving potentials there is no need to raise the density
cutoff beyond 4x the wavefunction cutoff, but for the ultra-soft
potentials you may need up to 10x the cutoff. doing those calculations
with a small unit cell and plenty k-points is usually _very_ fast.
once you know how each potential's requirements are, you have a much
better understanding for what to choose for your compound.

HDA> understanding, i would ask if someone could send me:  M.Bernasconi 
HDA> et al. J.Phys.Chem.Solids 56,501(1995).

please note, that this technique is mainly intended for variable 
cell dynamics, where you are (for technical reasons) using a constant
number of plane waves (and not a constant cutoff). with individual
calculations, like you were doing, you are running calculations with
"constant cutoff" (within the limits of the available grids and matching
number of plane waves, hence the "jumps"). without the bernasconi trick
you need to choose an insanely high wavefunction cutoff to make sure 
the stress tensor is tightly converged wrt. cutoff, so that (small)
changes in the effective cutoff do not matter.

hope that helps,
   axel.

HDA> thanks 
HDA>  
HDA> Hania Djani-aitaissa
HDA> CDTA, Algiers, Algeria
HDA> _________________________________________________________________
HDA> Téléchargez le nouveau Windows Live Messenger !
HDA> http://get.live.com/messenger/overview

-- 
=======================================================================
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