[QE-users] YBCO (001) slabs: "convergence" problem

Wed May 28 21:49:21 CEST 2025

Dear Michele,
as far as mixing_mode type and the type of diagonalization i usually use
the default ones, and try other possible choices only in the cases that
there is no means to make a calculation working.
I don't think that increasing the number of bands by 40% can help. When
using smearing, the code automatically adds maybe 20% more bands than
nelec/2 (nelec if using spin orbit). I would not say that your convergence
issues are due to the fact that bands at energy higher than the (nelec / 2
*1.2)-th do play a role. Using 40% is maybe just a waste of computational
time and might be considered only in nscf calculations if you need to plot
higher energy bands.
The ecutwfc you use seems very high. Provided that only convergence tests
can give you the right answer, within each pseudo file there is a suggested
ecutwfc, that is definitely lower than 80 Ry. You should use the largest
one found in these files and make convergence tests to check if higher
values should be considered.
ecutrho for PAW pseudopotential can range from 6 to 8 or even 12 times
ecutwfc. Again, there are some suggested values in the pseudo files, but
convergence test can confirm whether or not they are a safe choice.
For smearing I usually use 'marzari-vanderbilt', the value should be a
compromise such that you use the smallest degauss with the largest
affordable k-point grid, some good explanation can be found for example
here: https://theos-wiki.epfl.ch/Main/ElectronicTemperature

Hope this helps a bit.
Giovanni

-- 

Giovanni Cantele, PhD
CNR-SPIN
c/o Dipartimento di Fisica
Universita' di Napoli "Federico II"
Complesso Universitario M. S. Angelo - Ed. 6
Via Cintia, I-80126, Napoli, Italy
e-mail: giovanni.cantele at spin.cnr.it <giovanni.cantele at spin.cnr.it>
Phone: +39 081 676910
Skype contact: giocan74

ResearcherID: http://www.researcherid.com/rid/A-1951-2009
Web page: https://sites.google.com/view/giovanni-cantele/home

Il giorno mer 28 mag 2025 alle ore 16:39 michele de angelis <
michele.deangelis.01 at alumni.uniroma2.eu> ha scritto:

> Dear Giovanni,
> Thanks for the answer, I wanted to ask you what are the best options and
> combinations for this type of calculation, in terms of: smearing (type and
> value),  mixing_mode type and the type of diagonalization. Also the value
> I’m using for ecutrho is larger than the convergence value which is 200 Ry,
> I increased the number of bands by 40% and I left the convergence value for
> Ecutwfc and K-grid (obviously being a slab I’m using 6x6x1). These values
> seem to make sense but if you have advice I gladly accept them also in
> considering other command lines to activate on QE.
> Thanks again and sorry for the trouble.
> Have a good day,
> Michele
>
> Inviato da Outlook per Android <https://aka.ms/AAb9ysg>
>
> ------------------------------
> *Da:* users <users-bounces at lists.quantum-espresso.org> per conto di
> Giovanni Cantele <giovanni.cantele at spin.cnr.it>
> *Inviato:* mercoledì, maggio 28, 2025 12:19:10 PM
> *A:* Quantum ESPRESSO users Forum <users at lists.quantum-espresso.org>
> *Oggetto:* Re: [QE-users] YBCO (001) slabs: "convergence" problem
>
> Dear Michele,
>
> I have had the time for just a quick inspection to your files, and
> unfortunately I did not find possible mistakes.
> Maybe someone else will be more helpful in this regard.
> Both the total energy and the total force in the output are
> monotonically decreasing, with the exception of the
> 4th relaxation step.
> The only thing that might be odd (but it could be correct for the system
> you're studying) is that the
> comparison of the initial and the latest structure shows only an outwards
> relaxation of the topmost and lowest atomic
> layers.
>
> I can just suggest at the moment to restart from scratch a calculation in
> which you use as initial
> positions the latest (or the ones before the latest) printed by the code.
> It shouldn't change anything in principle, but
> there have been several cases in my experience in which that properly
> worked, as if at some point of the structural optimization the
> starting charge density were for some reason "wrong" or "inappropriate"
> (even if coming from the previous steps), and
> restarting from atomic/random wave functions prevents the code from
> remaining stuck. However, I don't have a clear explanation for that
> if not that this recipe has worked for me!
>
> Giovanni
>
> --
>
> Giovanni Cantele, PhD
> CNR-SPIN
> c/o Dipartimento di Fisica
> Universita' di Napoli "Federico II"
> Complesso Universitario M. S. Angelo - Ed. 6
> Via Cintia, I-80126, Napoli, Italy
> e-mail: giovanni.cantele at spin.cnr.it <giovanni.cantele at spin.cnr.it>
> Phone: +39 081 676910
> Skype contact: giocan74
>
> ResearcherID: http://www.researcherid.com/rid/A-1951-2009
> Web page: https://sites.google.com/view/giovanni-cantele/home
>
>
> Il giorno mer 28 mag 2025 alle ore 11:23 michele de angelis <
> michele.deangelis.01 at alumni.uniroma2.eu> ha scritto:
>
>> YBCO_001_CuO2_2layers.in
>> <https://uniroma2-my.sharepoint.com/:u:/g/personal/michele_deangelis_01_alumni_uniroma2_eu/EdqwsiZzdStJhJGXaYDSIOsB0WUkus6riva9xFNK4dvGrw>
>> YBCO_001_CuO2_2layers.out
>> <https://uniroma2-my.sharepoint.com/:u:/g/personal/michele_deangelis_01_alumni_uniroma2_eu/EWwLUiRWxFxLiOe-okzTGBkBJ5sQt7QCPbjEqCaKYWBDQA>
>> Dear all,
>> I am trying to relax YBCO (001) slabs with symmetric terminations (same
>> on both sides) and a vacuum spacing of 7.5 Å on each side (7.5 Å – slab –
>> 7.5 Å). I have tested all four possible surface terminations of YBCO, but
>> none of the relaxations converge properly. After m SCF cycles and n BFGS
>> cycles, the SCF calculation starts to exceed 100 iterations.
>> Moreover, the total energy values reported in the output file do not show
>> a monotonic trend but oscillate instead. In contrast, the total force does
>> decrease, from 0.9 down to about 0.09. I am using PBE-PAW pseudopotentials
>> with the GGA approximation. The initial geometry has been carefully
>> verified.
>> Spin and polarization effects are not included, as the slabs are
>> symmetric. I have attempted several adjustments to improve convergence,
>> including:
>> - Reducing mixing_beta to 0.1
>> - Increasing degauss to 0.05 and changing the smearing type
>> - Setting mixing_mode = 'local-TF'
>> - Switching the diagonalization method to cg
>>
>> However, the problem persists.
>> Could you kindly help me understand what might be going wrong? My input
>> and output files are attached below for reference.
>> Thank you in advance for your support.
>>  Best regards,
>>
>> Michele De Angelis
>> PhD student, Dept. of Chemical Sciences and Technologies, Univesity of
>> Rome Tor Vergata (Italy)
>>
>>
>>
>> _______________________________________________________________________________
>> The Quantum ESPRESSO Foundation stands in solidarity with all civilians
>> worldwide who are victims of terrorism, military aggression, and
>> indiscriminate warfare.
>>
>> --------------------------------------------------------------------------------
>> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
>> users mailing list users at lists.quantum-espresso.org
>> https://lists.quantum-espresso.org/mailman/listinfo/users
>>
>
>
> _______________________________________________________________________________
> The Quantum ESPRESSO Foundation stands in solidarity with all civilians
> worldwide who are victims of terrorism, military aggression, and
> indiscriminate warfare.
>
> --------------------------------------------------------------------------------
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20250528/5d4924d7/attachment.html>