[Pw_forum] Re: stress in CP and FFT grids

[Axel Kohlmeyer]

On 5/24/06, Konstantin Kudin <konstantin_kudin at yahoo.com> wrote:
>  Hi Axel!
>
> > what kind of cutoffs are you using?
> > does your system expand or shrink? and how much?
>
>  Ordinarily, I would use the cutoff of 25 Ryd, however, for the
> constant pressure I was using 30 Ryd with the modified kinetic energy
> functional
> [ecutwfc = 30.0, ecfixed = 25.0, qcutz = 25.0, q2sigma = 3.0]

ok, i have very little experience how well this works with ultra-soft
pseudopotentials, but i suggest you try with a higher plane wave
cutoff, e.g. 35/40 nevertheless and a too large system. this way
your basis set can only get 'better'  when the cell shrinks and not
the other way around. at least this is how it worked for me with
norm-conserving pseudopotentials (and a different code, too).

of course the 'softer' the potential curve with respect to the
box size, the better your stress tensor has to be converged
with respect to the basis set and the density grid.

>  I started with Z=22 A, the grid was:
>    Array leading dimensions ( nr1x, nr2x, nr3x )   =     90    90   192
>    Array leading dimensions ( nr1x, nr2x, nr3x )   =     72    72   150
>    Array leading dimensions ( nr1x, nr2x, nr3x )   =     24    24    24
>
>  It settled around Z=23 A (I optimized only Z), and if I restart a
> fresh calculation with all the same parameters, i get:
>
>    Array leading dimensions ( nr1x, nr2x, nr3x )   =     90    90   200
>    Array leading dimensions ( nr1x, nr2x, nr3x )   =     72    72   160
>    Array leading dimensions ( nr1x, nr2x, nr3x )   =     24    24    24
>
>  So the change in the grid is pretty much unaccounted for by the
> constant pressure simulation. I guess I could force the Z=23A run to
> use the same grid as before, however, that is not so clean.

no you want it the other way around. you cannot change the grid, so
you want the code to use the finest grid that you might need during the
run. i.e. something corresponding to the maximally expected cell size.

best regards,
     axel.

>
>  Thanks!
>  Kostya
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection around
> http://mail.yahoo.com
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>
>


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