[gpaw-users] RuO2 NiO

[Isabela Man]

Hi Niels,
You can try to run the first iteration in the LCAO mode and than to switch
to fd mode. For me this worked for TiO2 with transitional metals.

Here is the way you should do:

calc = GPAW(h=0.2,
xc='RPBE',
txt=outid,
kpts=(4,4,1),
mode = 'lcao',
#eigensolver='cg',
width=0.10,
#verbose=True,
mixer=mixer,
poissonsolver = DipoleCorrectionPoissonSolver(p,2),
spinpol=True,
nbands = -20,
usesymm=False,
maxiter = 500)

#sir quasi newton run
calc.attach(calc.write,1,gpwdat)
slabs.set_calculator(calc)
slabs.get_potential_energy()
calc.set(mode = 'fd')

Good luck

Isabela



On Thu, May 3, 2012 at 8:41 PM, Christopher O'Grady
<cpo at slac.stanford.edu>wrote:

>
> Niels wrote:
>
> >I am looking into a rutile RuO2 structure with doped Ni atoms on the
> >surface. However when I add adsorbates (O and OH to a Ru cus site on
> >the surface the calculations fail to converge.
>
> Hi Niels,
>
> If I'm not mistaken, one of our visitors (Ansgar Schaefer from BASF)
> spent quite a bit of time looking at convergence behaviour/performance
> for rutile RuO2.  You can find a summary in the large table on the
> last page here:
>
>
> https://confluence.slac.stanford.edu/download/attachments/94833934/ansgar_102610.pdf?version=1&modificationDate=1288133189000
>
> He looked at O defects, not adsorbates, but perhaps it will give you
> some ideas to try.
>
> For some rutiles, I believe he had to go to the cg eigensolver, which
> was much slower.  I've heard rumors that that could perhaps be made as
> fast as rmm-diis with some work.
>
> chris
> _______________________________________________
> 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/20120506/0c757eff/attachment.html