[gpaw-users] [gpaw-developers] Functional specific convergence problem

[Lasse Vilhelmsen]

Hi Jens Jørgen,

I'm normally using version 7113 and I now tried also with the newest version from the SVN (7336) and the behavior is unchanged. The only difference is that the steps taken are a bit different, but the energies that it starts oscillating between are the same. 

Best Regards
Lasse

On Nov 19, 2010, at 11:17 AM, Jens Jørgen Mortensen wrote:

> On Mon, 2010-11-15 at 09:34 +0100, Lasse Vilhelmsen wrote:
>> Hi developers,
>> 
>> I tried submitting the below question to the users list, but without any luck. I'm hoping that some of you experienced developers perhaps are able to help.
> 
> I think all gpaw-developers subscribers also subscribe to gpaw-users.
> 
> What version of GPAW are you using?  At revision 7223, the way MGGA
> calculations are done was changed a bit.  If you are using an earlier
> revision, then try one after 7223 -  if you are using an newer revision,
> then try one from before 7223.
> 
> JJ
> 
>> I have a problem converging a system with the M06-L functional. The system causing the problem is a MgO(100) slab with an oxygen vacancy. The system converges without any problems if I have an adsorbate on the slab or if the vacancy isn't present. 
>> 
>> I have tried changing the following parameters:
>> - Spin polarisation
>> - Number of k-points
>> - Mixing (I've tried a lot of very conservative combinations)
>> - Eigensolver (I've tried them all)
>> - Translating the slab a bit in the cell
>> - Changing the size of the slab (both number of layers and the surface size)
>> - Different starting geometries (small displacements around the equilibrium structure)
>> - Different PBE starting densities (changing the convergence)
>> - Number of bands to converge
>> - Fermi temperature (0.0, 0.05, 0.1, 0.5)
>> 
>> The slab converges without any problem with both PBE and TPSS.
>> 
>> Attached is both a sample script which shows this problem and a log from running the script. As can be seen from the log the slab starts oscillating between two different energies and it seems like it can't decide on which minimum to choose.
>> 
>> I hope that some of you have a good suggestion or two for other parameters I can change to get the slab converged. Perhaps some parameters to the eigensolver can be tuned?
>> 
>> Best Regards
>> Lasse Vilhelmsen
>> 
>> _______________________________________________
>> gpaw-developers mailing list
>> gpaw-developers at listserv.fysik.dtu.dk
>> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-developers
>