[QE-users] Possibly too few bands at point 2 message in DFT+U+V calculation

Iurii TIMROV iurii.timrov at epfl.ch
Thu Nov 25 11:04:02 CET 2021


> Well, pw.x just takes 20% more bands. It could not be enough, for example if you have a bunch of f-orbitals piling up around the Fermi energy, which is a typical case where you may want to use DFT+U. Double checking is always a good idea.


Ok, I agree. In this specific case the system contains Fe and O atoms, so no f electrons. But still good to check.


Thanks,

Iurii


________________________________
From: Lorenzo Paulatto <lorenzo.paulatto at sorbonne-universite.fr>
Sent: Thursday, November 25, 2021 10:57:37 AM
To: Iurii TIMROV
Cc: Quantum ESPRESSO users Forum
Subject: Re: [QE-users] Possibly too few bands at point 2 message in DFT+U+V calculation


 But if nbnd was specified explicitly in the input then one should increase nbnd and see what happens.

Well, pw.x just takes 20% more bands. It could not be enough, for example if you have a bunch of f-orbitals piling up around the Fermi energy, which is a typical case where you may want to use DFT+U. Double checking is always a good idea.

Dr. Lorenzo Paulatto
IdR @ IMPMC - CNRS UMR 7590 & Sorbonne Université
phone: +33 (0)1 442 79822 / skype: paulatz
http://www.impmc.upmc.fr/~paulatto/<https://link.getmailspring.com/link/61CD862F-ADD7-4B6F-8AB9-AFAC6904024B@getmailspring.com/0?redirect=http%3A%2F%2Fwww.impmc.upmc.fr%2F~paulatto%2F&recipient=aXVyaWkudGltcm92QGVwZmwuY2g%3D> - https://anharmonic.github.io/<https://link.getmailspring.com/link/61CD862F-ADD7-4B6F-8AB9-AFAC6904024B@getmailspring.com/1?redirect=https%3A%2F%2Fanharmonic.github.io%2F&recipient=aXVyaWkudGltcm92QGVwZmwuY2g%3D>
23-24/423 B115, 4 place Jussieu 75252 Paris CX 05
On Nov 25 2021, at 10:40 am, Iurii TIMROV <iurii.timrov at epfl.ch> wrote:

Hi Lorenzo,

Thanks a lot for clarifications!


> The only way to be sure is to repeat the pw.x calculation increasing nbnd, but if pw.x did not complain during scf about the total integrated charge being slightly off, the calculation should be ok.


Ok. So it depends how the pw.x calculation was performed. If nbnd was not specified explicitly in the input then I guess all should be fine because pw.x determines nbnd automatically. But if nbnd was specified explicitly in the input then one should increase nbnd and see what happens.

Iurii

--
Dr. Iurii TIMROV
Senior Research Scientist
Theory and Simulation of Materials (THEOS)
Swiss Federal Institute of Technology Lausanne (EPFL)
CH-1015 Lausanne, Switzerland
+41 21 69 34 881
http://people.epfl.ch/265334<https://link.getmailspring.com/link/61CD862F-ADD7-4B6F-8AB9-AFAC6904024B@getmailspring.com/2?redirect=http%3A%2F%2Fpeople.epfl.ch%2F265334&recipient=aXVyaWkudGltcm92QGVwZmwuY2g%3D>
From: Lorenzo Paulatto <lorenzo.paulatto at sorbonne-universite.fr>
Sent: Thursday, November 25, 2021 10:34:05 AM
To: Iurii TIMROV; Quantum ESPRESSO users Forum
Subject: Re: [QE-users] Possibly too few bands at point 2 message in DFT+U+V calculation


> In the HP output file, I am getting below two messages:
> Possibly too few bands at point    2
> If this is really related to QE input file then why I do not see any such message in QE output file?

I am not totally sure but I presume that this might be related to the fact that you use too many cores for your system. I think no need to worry about this.

Hello Iurii, I hope you don't mind if I jump in and comment on this.

The premise is that the Sternheimer equation of DFPT can be solved using only the valence states, which is it greatest advantage w.r.t standard QM perturbation theory, which requires a sum over the entire spectrum of virtual states. In the case of metal, where electronic occupation can be fractionary, the Sternheimer equation can be adjusted to keep this into account (PRB 51, 6773 (1995)).

The completely empty states are still not required and the linear-response solver tries to throw them away to save time. The algorithm only includes states up to a ~3 sigmas (the smearing width) up from the Fermi energy. If during this search the algorithm runs out of bands before reaching the threshold, its warns that maybe there were not enough bands to begin with, i.e. that if there had been more bands their occupation may have been non-negligible.
The only way to be sure is to repeat the pw.x calculation increasing nbnd, but if pw.x did not complain during scf about the total integrated charge being slightly off, the calculation should be ok.

hth


Dr. Lorenzo Paulatto
IdR @ IMPMC - CNRS UMR 7590 & Sorbonne Université
phone: +33 (0)1 442 79822 / skype: paulatz
http://www.impmc.upmc.fr/~paulatto/<https://link.getmailspring.com/link/61CD862F-ADD7-4B6F-8AB9-AFAC6904024B@getmailspring.com/3?redirect=https%3A%2F%2Flink.getmailspring.com%2Flink%2F30EFE2C7-FCF2-4279-A941-421485ED269D%40getmailspring.com%2F0%3Fredirect%3Dhttp%253A%252F%252Fwww.impmc.upmc.fr%252F~paulatto%252F%26recipient%3DaXVyaWkudGltcm92QGVwZmwuY2g%253D&recipient=aXVyaWkudGltcm92QGVwZmwuY2g%3D> - https://anharmonic.github.io/<https://link.getmailspring.com/link/61CD862F-ADD7-4B6F-8AB9-AFAC6904024B@getmailspring.com/4?redirect=https%3A%2F%2Flink.getmailspring.com%2Flink%2F30EFE2C7-FCF2-4279-A941-421485ED269D%40getmailspring.com%2F1%3Fredirect%3Dhttps%253A%252F%252Fanharmonic.github.io%252F%26recipient%3DaXVyaWkudGltcm92QGVwZmwuY2g%253D&recipient=aXVyaWkudGltcm92QGVwZmwuY2g%3D>
23-24/423 B115, 4 place Jussieu 75252 Paris CX 05
On Nov 25 2021, at 10:11 am, Iurii TIMROV via users <users at lists.quantum-espresso.org> wrote:



> In the HP output file, I am getting below two messages:
> Possibly too few bands at point    2
> If this is really related to QE input file then why I do not see any such message in QE output file?

I am not totally sure but I presume that this might be related to the fact that you use too many cores for your system. I think no need to worry about this.

> Should we set nbnd in QE input file or we can ignore it?

If you do not specify nbnd, pw.x will automatically include only occupied states for insulators and +20% more for metals.

> Number of k-points >= 100: set verbosity='high' to print the bands.
> I do not find q verbosity tag in "hp.x input description" file instead I found iverbosity.

HP prints only irreducible q points if you use symmetry.

> Anyway, it should not affect the final U+V value as we do not see any effect of verbosity on band structure calculation. RIght?

Right

> or I should set iverbosity=2 in HP input file?


No need. But if you want to see more info you can do that.

HTH

Iurii

--
Dr. Iurii TIMROV
Senior Research Scientist
Theory and Simulation of Materials (THEOS)
Swiss Federal Institute of Technology Lausanne (EPFL)
CH-1015 Lausanne, Switzerland
+41 21 69 34 881
http://people.epfl.ch/265334<https://link.getmailspring.com/link/61CD862F-ADD7-4B6F-8AB9-AFAC6904024B@getmailspring.com/5?redirect=https%3A%2F%2Flink.getmailspring.com%2Flink%2F30EFE2C7-FCF2-4279-A941-421485ED269D%40getmailspring.com%2F2%3Fredirect%3Dhttp%253A%252F%252Fpeople.epfl.ch%252F265334%26recipient%3DaXVyaWkudGltcm92QGVwZmwuY2g%253D&recipient=aXVyaWkudGltcm92QGVwZmwuY2g%3D>
From: users <users-bounces at lists.quantum-espresso.org> on behalf of Dr. K. C. Bhamu <kcbhamu85 at gmail.com>
Sent: Thursday, November 25, 2021 3:47:50 AM
To: Quantum Espresso users Forum
Subject: [QE-users] Possibly too few bands at point 2 message in DFT+U+V calculation

Dear Dr. Timrov,

I have set verbosity = high in QE input file without specifying nbnd.
In the HP output file, I am getting below two messages:

  *
Possibly too few bands at point    2

If this is really related to QE input file then why I do not see any such message in QE output file?
Should we set nbnd in QE input file or we can ignore it?

  *
Number of k-points >= 100: set verbosity='high' to print the bands.

I do not find q verbosity tag in "hp.x input description" file instead I found iverbosity.  Anyway, it should not affect the final U+V value as we do not see any effect of verbosity on band structure calculation. RIght? or I should set iverbosity=2 in HP input file?

Thank you for guiding me.

Regards

K C Bhamu
Postdoctoral Fellow
School of Chemical Engineering
University of Ulsan
South Korea




_______________________________________________
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/20211125/913d81a3/attachment.html>


More information about the users mailing list