[gpaw-users] beef error
Peter Bjerre Jensen
pbjen at dtu.dk
Mon Mar 2 11:48:18 CET 2015
Hi both
Thank you for the comments I will try to do a bit of testing.
I will let you no if I come to any conclusion. All I will do will be in
grid mode using full PBC.
Best regards
Peter
> Date: Sun, 1 Mar 2015 18:02:19 +0100 From: Marcin Dulak
> <Marcin.Dulak at fysik.dtu.dk> To: Jess Wellendorff
> <jess.wellendorff.dtu at gmail.com> Cc: gpaw-users at listserv.fysik.dtu.dk
> Subject: Re: [gpaw-users] beef error Message-ID:
> <54F3461B.3080808 at fysik.dtu.dk> Content-Type: text/plain;
> charset="windows-1252"; Format="flowed" Hi, On 03/01/2015 12:09 PM,
> Jess Wellendorff wrote:
>> Peter: I certainly think starting from a GGA electron density may in
>> general save computational time. If you want to go all the way and
>> also start BEEF-vdW geometry optimization from a GGA geometry, PBE is
>> probably a good choice.
> one would have to verify if starting from PBE helps on more examples.
> As Manuel noticed the SCF convergence is fast for vdW functionals for
> non-periodic BC.
> I can add that the convergence problem seems to be present only in grid
> mode, planewave mode converges fast for both periodic and non-periodic BC.
> I have opened a ticket with a script attached:
> https://trac.fysik.dtu.dk/projects/gpaw/ticket/266
>
> Best regards,
>
> Marcin
>> - Jess
>>
>> On 02/28/2015 08:56 PM, Manuel ?aric wrote:
>>> Regarding using the geometry converged with PBE or other xc.
>>>
>>> I remember I had to do something similar for methanol. It simply wasn't converging with BEEF-vdW no matter what I tried.
>>> What I did is I optimized the structure with PBE and used it as the start for the BEEF-vdW optimization (or maybe it was LCAO).
>>> I did the same with LCAO for dimethylcarbonate and dimethyloxalate. It worked.
>>>
>>> I would recommend doing this for organic molecules and using it as a backup plan for molecules that aren't converging.
>>> I'm not sure if I would do it always. Some molecules converge nicely.
>>>
>>> Note: In the LCAO calculator I used basis = 'dzp'
>>>
>>> Best regards,
>>>
>>> Manuel
>>> ________________________________________
>>> From:gpaw-users-bounces at listserv.fysik.dtu.dk [gpaw-users-bounces at listserv.fysik.dtu.dk] on behalf of Peter Bjerre Jensen [pbjen at dtu.dk]
>>> Sent: Saturday, February 28, 2015 5:31 PM
>>> To:gpaw-users at listserv.fysik.dtu.dk
>>> Subject: Re: [gpaw-users] beef error
>>>
>>> Hi Jess..
>>>
>>> Den 28-02-2015 kl. 12:00 skrevgpaw-users-request at listserv.fysik.dtu.dk<mailto:gpaw-users-request at listserv.fysik.dtu.dk>:
>>>
>>> Message: 1
>>> Date: Sat, 28 Feb 2015 09:06:53 +0100
>>> From: Jess Wellendorff<jess.wellendorff.dtu at gmail.com><mailto:jess.wellendorff.dtu at gmail.com>
>>> To:gpaw-users at listserv.fysik.dtu.dk<mailto:gpaw-users at listserv.fysik.dtu.dk>
>>> Subject: Re: [gpaw-users] beef error
>>> Message-ID:<54F1771D.10804 at gmail.com><mailto:54F1771D.10804 at gmail.com>
>>> Content-Type: text/plain; charset=windows-1252; format=flowed
>>>
>>> One could also try to start from a converged GGA calculation, eg. PBE or
>>> revPBE. This should give a fast and very good initial guess for the
>>> BEEF-vdW electron density. Might even save time since the GGA SCF steps
>>> are faster than the BEEF-vdW ones. It's very easy to switch to BEEF-vdW
>>> after a converged GGA run:
>>>
>>> h2o.calc.set(xc='BEEF-vdW')
>>>
>>> Do you think one should do this generally to save computation time, and improve the success of convergence?
>>>
>>> Best regards
>>> Peter
>>>
>>>
>>> Peter Bjerre Jensen
>>> Postdoc
>>> DTU Energy
>>>
>>>
>>> Technical University of Denmark
>>> [cid:part1.09000908.08040102 at dtu.dk]
>>>
>>> Department of Energy Conversion and Storage
>>> Fysikvej, Building 307
>>> 2800 Kgs. Lyngby
>>> Direct +45 45253204
>>> pbjen at dtu.dk<mailto:pbjen at dtu.dk>
>>> www.ecs.dtu.dk<http://www.ecs.dtu.dk>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> - Jess
>>>
>>> On 02/27/2015 11:23 AM, Manuel ?aric wrote:
>>>
>>>
>>> Regarding H2O(g) with BEEF-vdW.
>>>
>>> One should be able to do this:
>>> """
>>> from ase import Atoms, Atom
>>> from ase.optimize import QuasiNewton
>>> from gpaw import GPAW, FermiDirac, Mixer
>>> from ase.visualize import view
>>> from gpaw.eigensolvers.davidson import Davidson
>>> from ase.io import read, write
>>>
>>> name = 'H2O'
>>>
>>> cell = (12, 12, 12)
>>> periodic = (0, 0, 0)
>>>
>>> h2o = Atoms([Atom('O', position=(0, 0, 0)),
>>> Atom('H', position=(1, 0, 0)),
>>> Atom('H', position=(0, 1, 0))], cell=cell, pbc=periodic)
>>> h2o.center()
>>> view(h2o)
>>>
>>> h = 0.18
>>> nbands = -10
>>> xc = 'BEEF-vdW'
>>> mixer = Mixer(0.05, 5 , weight=50.0)
>>> eigensolver = Davidson(niter=2)
>>>
>>> calc = GPAW(h=h,
>>> xc=xc,
>>> mixer=mixer,
>>> eigensolver=eigensolver,
>>> nbands=nbands,
>>> txt='%s.txt' % name)
>>>
>>> h2o.set_calculator(calc)
>>>
>>> dyn = QuasiNewton(h2o, trajectory='%s.traj' % name, logfile='%s.log' % name)
>>> dyn.run(fmax=0.05)
>>>
>>> calc.write('%s.gpw' % name)
>>> """
>>> Should work for O2(g) as well.
>>>
>>> I remember having problems when I tried using a larger cell with pbc = ((1, 1, 1))
>>>
>>> Hope it helps
>>> ______________________________________
>>> From: Marcin Dulak
>>> Sent: Wednesday, February 25, 2015 1:02 PM
>>> To: Logi Arnarson;gpaw-users at listserv.fysik.dtu.dk<mailto:gpaw-users at listserv.fysik.dtu.dk>; Manuel ?aric
>>> Subject: RE: beef error
>>>
>>> Hi,
>>>
>>> the calculation scf does not converge.
>>> I believe this is a common problem with the BEFF-vdw functional for molecules,
>>> seehttp://listserv.fysik.dtu.dk/pipermail/gpaw-users/2014-May/002686.html
>>> I would try to set width=0.1 eV and check you don't get fractional occupancies in the solution.
>>> Manuel: how do you converge molecules with BEEF-vdw?
>>>
>>> Best regards,
>>>
>>> Marcin
>>> ________________________________
>>> From: Logi Arnarson [logi at au.dk<mailto:logi at au.dk>]
>>> Sent: Wednesday, February 25, 2015 9:24 AM
>>> To:gpaw-users at listserv.fysik.dtu.dk<mailto:gpaw-users at listserv.fysik.dtu.dk>
>>> Subject: beef error
>>>
>>> Dear GPAW Users,
>>>
>>> I am trying to estimate the error of a formation energy calculation using the the BEEF-vdw functional. The calculation is for a Vanadia Oxide species on a surface (VO3H/TiO2) where the VO3H is calculated with respect to V2O5(b) and H2O(g).
>>> In the script (see script.txt) I therefore read the respective gpw files in order to get the potential energies. Then I get KohnShamConvergenceError for the H2O calculation (see error.txt). I have done analogous beef calculation for a VO3 species where instead of having H2O(g) as the reference but O2(g) which doesn't give rise to an error. The calculations for H2O(g) and O2(g) are done in exactly the same way.
>>>
>>> Is there someone out there who might know how to fix this?
>>>
>>> Regards,
>>> Logi
>>>
>>> --
>>> Logi Arnarson
>>> PhD Student
>>> iNANO, AArhus University
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> gpaw-users mailing list
>>> gpaw-users at listserv.fysik.dtu.dk<mailto:gpaw-users at listserv.fysik.dtu.dk>
>>> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>>>
>>>
>>>
>>>
>>>
>>> ------------------------------
>>>
>>> _______________________________________________
>>> gpaw-users mailing list
>>> gpaw-users at listserv.fysik.dtu.dk<mailto:gpaw-users at listserv.fysik.dtu.dk>
>>> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>>>
>>> End of gpaw-users Digest, Vol 61, Issue 22
>>> ******************************************
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> gpaw-users mailing list
>>> gpaw-users at listserv.fysik.dtu.dk
>>> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>>
>>
>> _______________________________________________
>> gpaw-users mailing list
>> gpaw-users at listserv.fysik.dtu.dk
>> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20150301/5c354825/attachment.html>
>
> ------------------------------
>
> _______________________________________________
> gpaw-users mailing list
> gpaw-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>
> End of gpaw-users Digest, Vol 62, Issue 2
> *****************************************
More information about the gpaw-users
mailing list