[Wannier] [Posible SPAM] Re: Problem reading *.mmn file

Tue May 26 13:58:21 CEST 2020

Dear Stepan

Thanks a lot for your message. 

Yes, I used nosym=.true. in the nscf step. I checked the nnkps entry in the *nnkp file, and the code started to write them (Nbands*Nbands lines per kpoints, that should be indeed Nbands*Nbands*Nkpoints*Nnb). After some number of lines, it no longer wrote anything else, but the calculation did not abort or anything like that. 

The file was like 3.5 Gb or so with more or less half the number of lines that it should contain. 

This is quite weird to me. I ran pw2wannier.x in other systems, and it did not fail. 

Thanks again

Juanjo

----- Mensaje original -----
De: "Stepan Tsirkin" <tsirkinss at gmail.com>
Para: wannier at lists.quantum-espresso.org
Enviados: Martes, 26 de Mayo 2020 13:37:05
Asunto: [Posible SPAM] Re: [Wannier] Problem reading *.mmn file

Dear Juanjo,

the .mmn should contain Nbands*Nbands*Nkpoints*Nnb, where Nnb in your case is 6 (number of neighbours in k-space)

As a guess, did you use nosym=.true.  in the nscf calculation of pwscf ? If not, then only the symmetry-irreducible kpoints were stored, not the full grid.

How big is the mmn file? could it be some problem with memory? (however I doubt that it could pass silently on the pw2wannier step)

Regards,
Stepan.

> On 26 May 2020, at 13:05, Juan Jose Melendez Martinez <melendez at unex.es> wrote:
>
> Dear Giovanni
>
> Thanks a lot for your response. I think that I found where the problem is: for some reason, the code is not writting the right number of lines. If I am right, the file .mmn should contain N_bands*N_bands data per block from the nnkp file. In my case, it wrote only a bit more than half. Surprisingly, the output from pw2wannier90.x (attached) did not show any error message.
>
> I ran the calculation on 72 cores. Could it be something related to the parallelization? Or is it something related to the size of the system?
>
> Take care
>
> Juanjo
>
> ----- Mensaje original -----
> De: "Giovanni Pizzi" <giovanni.pizzi at epfl.ch <mailto:giovanni.pizzi at epfl.ch>>
> Para: "Juan Jose Melendez Martinez" <melendez at unex.es <mailto:melendez at unex.es>>
> CC: wannier at lists.quantum-espresso.org <mailto:wannier at lists.quantum-espresso.org>
> Enviados: Martes, 26 de Mayo 2020 11:45:17
> Asunto: [Posible SPAM] Re: [Wannier] Problem reading *.mmn file
>
> Hi,
> The error is raised in this file
> https://github.com/wannier-developers/wannier90/blob/2ab7a904e8d9b8d8eb771ad4ce6fe2d415c895b0/src/overlap.F90
> (Error “103”).
> This is due to a problem in the file content.
>
> Could you open the file and check if you notice something strange, e.g. ***** instead of numbers, or NaN, or other unexpected values?
> I fear this might be the problem, and if this is the case, one needs to understand where this happened, fix the pw2wannier90 code and then re-run the code… (but maybe just the Mmn if the problem appeared only in the Mmn file?)
>
> Or maybe you will spot some other issue.
>
> Best,
> Giovanni
>
> --
> Giovanni Pizzi
> Theory and Simulation of Materials and MARVEL, EPFL
> http://people.epfl.ch/giovanni.pizzi
> http://nccr-marvel.ch/en/people/profile/giovanni-pizzi
>
> On 26 May 2020, at 09:05, Juan Jose Melendez Martinez <melendez at unex.es <mailto:melendez at unex.es><mailto:melendez at unex.es <mailto:melendez at unex.es>>> wrote:
>
> Dear all
>
> I am using Wannier90 to study a heterostructure made upon InSb and Ge. The system is large, and I need 848 wannier functions. I am facing a quite weird issue. After the DFT step with QE (successful, nothing strange there), I generated the *nnkp file (no problem) and then run pw2wannier.x to get the amn and mmn files. Again, nothing apparently strange there. But upon launching wannier90.x I get the error
>
> Error: Problem reading input file si_insb_111_2-10.mmn
>
> I have not been able to find any piece of info for this error message. It usually means that it is either very simple or very tough to solve. I cross my fingers for the former, since calculating the projections took more than 4 days.
>
> I am attaching the *win and *wout files, just in case they may help.
>
> Thanks a lot in advance
>
> Juanjo
>
> Juan J. Meléndez
> Department of Physics
> University of Extremadura
> Avda. de Elvas, s/n 06006
> Badajoz (Spain)<si_insb_111_2-10.wout><si_insb_111_2-10.win <https://protection.puc.rediris.es/fmlurlsvc/?fewReq=:B:JVk/NDE1PyNzODQrNSNsYTg1ND81NCN2bGJrZHFwd2A4ZD1mPTQyNz1hYzU0NTYyPDc2PTI9ZGY9MmBhPWZkNWFjNGBjYWBkPCNxODQwPDUxPDY3MjQjdGxhODUxVEdnZElENTU0MTM8KDUxVEdnZElGNTU0MTM8I3dmdXE4aGBpYGthYH9FcGtgfStgdiNmODw1&url=http%3a%2f%2fsi_insb_111_2-10.win%2f><https://protection.puc.rediris.es/fmlurlsvc/?fewReq=:B:JVk/NDE1PyNzODQrNSNsYTg1ND81NCN2bGJrZHFwd2A4NWYzPTFjYGE2MWA0NjxgPD0wNTFkMDFnZGM1Nz09ZmFmNmcwZz1nZyNxODQwPDUxPDY3MjQjdGxhODUxVEdnZElENTU0MTM8KDUxVEdnZElGNTU0MTM8I3dmdXE4aGBpYGthYH9FcGtgfStgdiNmODw1&url=http%3a%2f%2fsi_insb_111_2-10.win <https://protection.puc.rediris.es/fmlurlsvc/?fewReq=:B:JVk/NDE1PyNzODQrNSNsYTg1ND81NCN2bGJrZHFwd2A4ZD1mPTQyNz1hYzU0NTYyPDc2PTI9ZGY9MmBhPWZkNWFjNGBjYWBkPCNxODQwPDUxPDY3MjQjdGxhODUxVEdnZElENTU0MTM8KDUxVEdnZElGNTU0MTM8I3dmdXE4aGBpYGthYH9FcGtgfStgdiNmODw1&url=http%3a%2f%2fsi_insb_111_2-10.win%2f>>>_______________________________________________
> Wannier mailing list
> Wannier at lists.quantum-espresso.org <mailto:Wannier at lists.quantum-espresso.org><mailto:Wannier at lists.quantum-espresso.org <mailto:Wannier at lists.quantum-espresso.org>>
> https://lists.quantum-espresso.org/mailman/listinfo/wannier <https://lists.quantum-espresso.org/mailman/listinfo/wannier><si_insb_111_2-10.pw2wannier.out>_______________________________________________
> Wannier mailing list
> Wannier at lists.quantum-espresso.org <mailto:Wannier at lists.quantum-espresso.org>
> https://lists.quantum-espresso.org/mailman/listinfo/wannier <https://lists.quantum-espresso.org/mailman/listinfo/wannier>

_______________________________________________
Wannier mailing list
Wannier at lists.quantum-espresso.org
https://lists.quantum-espresso.org/mailman/listinfo/wannier