[QE-users] Momentum matrix element on PBE+U

[L Y]

Wow thanks Lurii for all these very helpful references and comments! I read
the section in your thesis and looked at the references. By commenting out
the part of the [r, V_nl] calculation in "compute_ppsi", I think I have
narrowed down the issue to the calculation of this commutator. Attached
are figures of velocity_{cv} with and without this commutator, along
symmetry lines. Near the Gamma point (ik=31), a dip is seen in the figure
with nonlocal part (in other calculations with USPP and PAW this
discontinuity) is much more pronounced.
I will now need to look at the implementation of this part, but the
implementation is quite complex with almost no comments. With your
experience, do you have any idea or hunch why there is a problem in the
calculation at the Gamma point? Can you refer me to references or a person
that possibly can help?

Best regards,
Lun Yue
Louisiana State University




It is now clear to me that "bug" in QE is in the part


Den fre. 15. maj 2020 kl. 04.13 skrev Timrov Iurii <iurii.timrov at epfl.ch>:

> > Did you mean [J. Chem. Phys. 120, 9935 (2004)] by Toby and Dal Corso?
>
>
> The correct reference (without mistakes) is this: J. Tóbik
> <https://aip.scitation.org/author/T%C3%B3bik%2C+Jaroslav>* and *A. Dal
> Corso <https://aip.scitation.org/author/Dal+Corso%2C+Andrea>, J. Chem.
> Phys. *120*, 9934 (2004).
>
>
> See Eq. (7) in this paper, which is for the ultra-soft pseudos.
> For norm-concerning pseudos, see Eq. (14) in S. Baroni and R. Resta, Phys.
> Rev. B 33, 7017 (1986).
>
>
> > I could not find an expression for the momentum matrix element in that
> reference, I think that reference is only about the extra term involving
> the ultrasoft pseudopotential?
>
>
> Let me try to explain for the corn-conserving case, which is simpler. The
> routine "compute_ppsi" has p - i [x, V_{nl}], where p is the momentum and V_{nl}
> is the nonlocal part of the pseudopotential. See the right-hand side of Eq.
> (15) in Phys. Rev. B 33, 7017 (1986). See also Eqs. (8) and (9) in
> A. Marini, G. Onida, and R. Del Sole, Phys. Rev. B 64, 195125 (2001).
>
>
> So actually the code computes not only the matrix element of the momentum
> p, but also it contains an extra contribution due to the non-local part
> of the pseudopotential. In the DFT+U case, there is yet another term given
> by Eq. (53) in Phys. Rev. B 101, 064305 (2020).
>
>
> Finally, some related discussion about this (how to compute an electron
> velocity using these "tricks") is given in Sec. 6.1.3(a) of my PhD thesis:
> https://pastel.archives-ouvertes.fr/pastel-00823758/
>
>
> HTH
>
>
> Iurii
>
>
> --
> Dr. Iurii Timrov
> Postdoctoral Researcher
> STI - IMX - THEOS and NCCR - MARVEL
> Swiss Federal Institute of Technology Lausanne (EPFL)
> CH-1015 Lausanne, Switzerland
> +41 21 69 34 881
> http://people.epfl.ch/265334
> ------------------------------
> *From:* users <users-bounces at lists.quantum-espresso.org> on behalf of L Y
> <xifntx at gmail.com>
> *Sent:* Thursday, May 14, 2020 9:15:47 PM
> *To:* Quantum ESPRESSO users Forum
> *Subject:* Re: [QE-users] Momentum matrix element on PBE+U
>
> Dear Lurii,
>
> Thank you for the quick response!
> Did you mean [J. Chem. Phys. 120, 9935 (2004)] by Toby and Dal Corso? I
> could not find an expression for the momentum matrix element in that
> reference, I think that reference is only about the extra term involving
> the ultrasoft pseudopotential?
> I ask because I have calculated the momentum matrix elements between
> valence and conduction bands along symmetry lines in MgO. Everything seems
> OK, except the fact that there is a noticeable discontinuity of the matrix
> element at the Gamma point. I did extensive debugging, but could find
> anything wrong (the band energies are fine and continuous without
> discontinuity). I suspect I need to do something special for the momentum
> matrix element, but I cannot figure out the expression.
>
> Best regards,
> Lun Yue
> Louisiana State University
>
> Den tor. 14. maj 2020 kl. 11.55 skrev Timrov Iurii <iurii.timrov at epfl.ch>:
>
>> Dear Lun Yue,
>>
>>
>> The reference is written in the header of that routine.
>>
>>
>> Greetings,
>>
>> Iurii
>>
>>
>> --
>> Dr. Iurii Timrov
>> Postdoctoral Researcher
>> STI - IMX - THEOS and NCCR - MARVEL
>> Swiss Federal Institute of Technology Lausanne (EPFL)
>> CH-1015 Lausanne, Switzerland
>> +41 21 69 34 881
>> http://people.epfl.ch/265334
>> ------------------------------
>> *From:* users <users-bounces at lists.quantum-espresso.org> on behalf of
>> lot fun <xifntx at gmail.com>
>> *Sent:* Thursday, May 14, 2020 6:45:16 PM
>> *To:* Quantum ESPRESSO users Forum
>> *Subject:* Re: [QE-users] Momentum matrix element on PBE+U
>>
>> Dear Lurii,
>>
>> I was wondering whether you have any reference on how the momentum matrix
>> element is implemented in QE?
>> Obviously the PP/src/compute_ppsi.f90 routine is crucial, but I dunno
>> which formula is used there. A reference to a paper would be
>> greatly appreciated!
>>
>> Best regards,
>> Lun Yue
>> Louisiana State University
>>
>> Den ons. 13. maj 2020 kl. 03.06 skrev Timrov Iurii <iurii.timrov at epfl.ch
>> >:
>>
>>> > I encounter this error: 'Error in routine write_p_avg (1): write_p_avg
>>> not working with LDA+U'.
>>>
>>>
>>> The LDA+U does not work in this case because there is a missing term,
>>> see Eq. (53) in PRB 101, 064305 (2020).
>>>
>>>
>>> > Is there a way to circumvent this on Quantum Espresso?
>>>
>>>
>>> I do not know any apart from adding the missing term.
>>>
>>>
>>> Iurii
>>>
>>>
>>> --
>>> Dr. Iurii Timrov
>>> Postdoctoral Researcher
>>> STI - IMX - THEOS and NCCR - MARVEL
>>> Swiss Federal Institute of Technology Lausanne (EPFL)
>>> CH-1015 Lausanne, Switzerland
>>> +41 21 69 34 881
>>> http://people.epfl.ch/265334
>>> ------------------------------
>>> *From:* users <users-bounces at lists.quantum-espresso.org> on behalf of
>>> Shivesh Sivakumar <shiveshsivakumar at gmail.com>
>>> *Sent:* Wednesday, May 13, 2020 8:14:25 AM
>>> *To:* Quantum ESPRESSO users Forum
>>> *Subject:* [QE-users] Momentum matrix element on PBE+U
>>>
>>> Hello users,
>>>
>>> Sorry for the repost, but I think my question went unnoticed.
>>> Is there a way to calculate the momentum matrix element in PBE+U on QE?
>>> I encounter this error: 'Error in routine write_p_avg (1): write_p_avg not
>>> working with LDA+U'. Is there a way to circumvent this on Quantum Espresso?
>>>
>>> Best,
>>> Shivesh Sivakumar
>>> University of Washington
>>> Seattle-WA, 98105
>>>
>>>
>>> _______________________________________________
>>> Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso
>>> )
>>> users mailing list users at lists.quantum-espresso.org
>>> https://lists.quantum-espresso.org/mailman/listinfo/users
>>
>> _______________________________________________
>> Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso)
>> users mailing list users at lists.quantum-espresso.org
>> https://lists.quantum-espresso.org/mailman/listinfo/users
>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso)
> 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/20200515/009f650c/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: p_without_nonlocal.png
Type: image/png
Size: 32904 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20200515/009f650c/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: p_with_nonlocal.png
Type: image/png
Size: 31788 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20200515/009f650c/attachment-0001.png>