[QE-users] relax : convergence NOT achieved after 80 iterations: stopping (Lucian D. Filip)

[Lucian D. Filip]

I agree that using scf_must_converge=.false. is not the best solution if 
you start with it without testing! But the question was why the 
calculation stopped after 80 iterations. The default for 
electron_maxstep is 100 and who knows maybe it was stopped too soon and 
it would have converged at 92. In any case it stopped because it was 
told to stop at 80 iterations. Why it didn't reach convergence is 
another matter.


Using scf_must_converge=.false. is not that bad if you test the system 
beforehand. There are systems that will not converge in the first step 
and then the algorithm moves the atoms a bit and then convergence is 
achieved as it should.


All the best,


Lucian



On 20/07/2022 22:26, Tiana, Davide wrote:
> That (using scf_must_converge=.false. is a very vary bad advice. If 
> the system is not converging after 80 iterations most likely there's 
> something wrong. This could be the pseudo, the geometry, or any thing.
> You should check the convergence rate between each step and see if it 
> fluctuate and of how much. Then, maybe it is almost converged so you 
> simply need to extend the number of cycles from 80 to 100 or 150.
> Last but not least, you can know if it's not converging because your 
> threshold is too tight, or it's not converging at all. If the latter 
> you need to find why. By using scf_must_converge=.false. you'll simply 
> risk to end in a wrong state.
>
> All the best
> Davide
> ------------------------------
>
> Message: 3
> Date: Wed, 20 Jul 2022 10:40:09 +0300
> From: "Lucian D. Filip" <lucian.filip at infim.ro>
> To: users at lists.quantum-espresso.org
> Subject: Re: [QE-users] relax : convergence NOT achieved after 80
>         iterations: stopping
> Message-ID: <85bedcbc-8cde-3bad-ba05-2e38d6e80063 at infim.ro>
> Content-Type: text/plain; charset="utf-8"; Format="flowed"
>
> here is the problem! you told QE to only do 80 iterations in a scf step.
> After the 80 iterations there was no convergence and the calculation
> stopped. If you want it to continue to the next step and let QE move the
> atoms even if the first scf iteration did not converge you should use
> the following flag together with the electron_maxstep:
>
> scf_must_converge = .false.
>
> --
> *Dr. Lucian Dragos Filip*
> National Institute of Materials Physics
> Atomistilor str. 405A, PO Box MG. 7
> Magurele, 077125
> Bucharest, Romania
> E-mail: lucian.filip at infim.ro
> Website: https://lucianfilip.wordpress.com/
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: 
> <http://lists.quantum-espresso.org/pipermail/users/attachments/20220720/546df053/attachment-0001.html>
>
> ------------------------------
>
> Subject: Digest Footer
>
> _______________________________________________
>
> The Quantum ESPRESSO community stands by the Ukrainian
> people and expresses its concerns about the devastating
> effects that the Russian military offensive has on their
> country and on the free and peaceful scientific, cultural,
> and economic cooperation amongst peoples.
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu 
> <http://www.max-centre.eu>)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
>
> ------------------------------
>
> End of users Digest, Vol 180, Issue 16
> **************************************
>
> _______________________________________________
> The Quantum ESPRESSO community stands by the Ukrainian
> people and expresses its concerns about the devastating
> effects that the Russian military offensive has on their
> country and on the free and peaceful scientific, cultural,
> and economic cooperation amongst peoples
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> users mailing listusers at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
-- 
*Dr. Lucian Dragos Filip*
National Institute of Materials Physics
Atomistilor str. 405A, PO Box MG. 7
Magurele, 077125
Bucharest, Romania
E-mail: lucian.filip at infim.ro
Website: https://lucianfilip.wordpress.com/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20220721/9079a757/attachment.html>