[Pw_forum] Total energy does not coverge with ecut

dario rocca roccad at gmail.com
Sat Nov 15 02:32:06 CET 2008


It seems that now you are using pseusopotentials generated with different
functionals (pz and pbe).
In my experience it is hard to converge ultrasoft with respect to ecutrho:
at a certain point the total enegy starts
oscillating or it goes down very slowly.  This could be the reason of the
behavior that you see, namely the
results for different ecutwfc does not converge because none of them is
fully converged with respect to
ecutrho. From this point of view I found that rrkj ultrasoft have a much
better behavior. If I remember well they are optimized  to use a smaller
number of plane-waves to describe the augmentation
charge. I would follow the suggestion "So usually it is better to check the
convergence of the quantity you are interested in"
Dario

On Tue, Nov 11, 2008 at 12:05 PM, Timothy Mason <thmason24 at yahoo.com> wrote:

> I have not attempted your suggestion however I did trade out the PSPS with
> different PSPSs.   I'm now using
>
>  Li.pz-n-vbc.UPF
>  N.pbe-rrkjus.UPF
>  H.pbe-rrkjus.UPF
>
> and they seem to converge nicely.
>
> is there problem with the ones I was using?   here they are for reference
>
> Li 6.941000 Li.pbe-n-van.UPF
> N 14.006740 N.pbe-van_ak.UPF
> H 1.007940 H.pbe-van_ak.UPF
>
> I'm not sure of the best way to select them from the website.
>
> thanks for the help.
>
>
>
>
>
>
>
> --- On *Thu, 11/6/08, prasenjit.jnc at gmail.com <prasenjit.jnc at gmail.com>*wrote:
>
> From: prasenjit.jnc at gmail.com <prasenjit.jnc at gmail.com>
> Subject: Re: Re: [Pw_forum] Total energy does not coverge with ecut
> To: "Timothy Mason" <thmason24 at yahoo.com>, "PWSCF Forum" <
> pw_forum at pwscf.org>
> Date: Thursday, November 6, 2008, 4:10 PM
>
>
> Sometimes it is also difficult to converge the total energy. So usually it
> is better to check the convergence of the quantity you are interested in.
>
> But anyway it is really strange that total energy doesn't converge at such
> abnormally high cut off values.
>
> Did you test the pseudopotentials individually?? For eg. did you try
> calculating the bond lengths of say a nitrogen molecule & see how the
> results are??
>
> With regards,
>
> Prasenjit.
>
> On Nov 6, 2008 5:45pm, Timothy Mason <thmason24 at yahoo.com> wrote:
> > Dear Lorenzo,
> >
> > here are results for the lower energy range. Still doesn't seem to be
> converging.
> >
> > ecut energy energy differnce
> > 30 -48.19258627
> > 35 -48.19503183 0.00244556
> > 40 -48.197306 0.00227417
> > 45 -48.20217820 0.0048722
> >
> > thanks
> >
> > --- On Thu, 11/6/08, Lorenzo Paulatto paulatto at sissa.it> wrote:
> > From: Lorenzo Paulatto paulatto at sissa.it>
> > Subject: Re: [Pw_forum] Total energy does not coverge with ecut
> > To: thmason24 at yahoo.com, "PWSCF Forum" pw_forum at pwscf.org>
> > Date: Thursday, November 6, 2008, 10:08 AM
> >
> > Timothy Mason ha scritto:
> > >
> > > Gentlemen,
> > >
> > > My structures do not see to be converging with increased ecutwfc.
> > > I'm setting ecutrho to ecutwfc * 12 and using US PP. below is a
> > > table of ecut values vs energies vs difference from previous steps
> > > followed by my input file followed by part of the output file. It
> > > appears to begin to converge and 120 ecut but then jumps at 130 and
> > > then more at 140.
> > >
> > Dear Timothy,
> > these are ridiculous cutoff energy for ultrasoft pseudopotentials - they
> > are very high for norm-conserving too! I suspect the presence of a
> > high-energy ghost-state in one of the pseudopotential; these kind
> > of
> > states aren't usually harmful but can suddenly pop-up under some
> > specifics conditions like high pressure or very high cutoff.
> >
> > What is happening when you try some reasonable cutoff in the 20-50 Ry
> range?
> >
> > regards
> >
> > --
> > Lorenzo Paulatto
> > SISSA & DEMOCRITOS (Trieste)
> > phone: +39 040 3787 511
> > skype: paulatz
> > www: http://people.sissa.it/~paulatto/<http://people.sissa.it/%7Epaulatto/>
> >
> > Due to budget cuts approved by Italian Minister of Finance and Italian
> > Minister "for" Education, University and Research this may be the
> > last
> > email you receive from me.
> >
> > Please send any future correspondence, possibly via pigeon, to
> > mr. Lorenzo Paulatto
> > under Roiano Railway Bridge - Pillar 2
> > 34136, Trieste
> >
> > Donations in food, coins, booze and processor time are warmly welcome.
> >
> >
> >
> >
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.democritos.it/pipermail/pw_forum/attachments/20081114/4d139970/attachment.htm 


More information about the Pw_forum mailing list