EPW modified source for the ~Quantum Espresso v7.0~ suite

[qiuzhanjun]

I am using the newest QE v7.0 and EPW v5.4. I have modified the EPW source code files according to the files of directory EPW/src/ but failed, due to the changes of ndegen_g. Is it possible to provide the modified EPW v5.4 code after the revision of this version?

[nakib]

Hi @qiuzhanjun. Thanks for using elphbolt and for your question! Would you be able to provide more information? For example, did you replace the whole file ephwann_shuffle.f90 of your version of EPW with the one provided with elphbolt? If yes, could you please try for this file adding just the lines 417-458 from elphbolt's version to your version of EPW right below line 389? Let me know if this solves your issue.

Best,
Nakib

[qiuzhanjun]

I appreciate your answer. According your advice, I recompiled the epw. Fortunately, it worked and didn't report an error. Next I'll test if it can print out the required files. I will give the feedback later.

[suan12]

Thanks for your reply@nakib. I know how to revise ephwann_shuffle.f90 according to the one provided with elphbolt.
However, I do not know how to modify rigid_epw.f90. THe source of rigid_epw.f90 in EPW v5.4 changes too much.
Is it possible to provide it for this version?
By the way, does the code take 4 phonons into considerations?

Best,
Pengfei Liu

[nakib]

Dear Pengfei Liu,

I took a quick look at the file rigid_epw.f90 of the EPW module shipped with QE version 7.0 and 7.1. For these versions, I don't think this file needs to be replaced for generating results for elphbolt. That said, it is always a good idea to compare the results of the band structure, phonon dispersions, and matrix elements plotted along some path between EPW and elphbolt. I hope this helps!

Related to the above, the next major version of elphbolt will likely be able to work with a new version of QE.

Regarding 4-phonons, yes I plan to interface with an external 4-phonon calculator, possibly the FourPhonon code (https://github.com/FourPhonon/FourPhonon). The tentative idea is to generate 4-phonon scattering rates on a relatively coarse q-mesh using this external tool. Then I will interpolate these onto a fine q-mesh mesh inside elphbolt. The linear interpolator already exists in elphbolt. I don't have a timeline for when this might become available. I always appreciate any contribution from the community though. So if you have time and human-power to spend on this project to make it happen quicker, let me know! :)

Best regards,
Nakib

[suan12]

Thank you for your reply. However, I have the same problem above in ephwann_shuffle.f90.
In QE v7.0, ndegen_g is defined as below:
INTEGER, ALLOCATABLE :: ndegen_g(:, :, :)

In the one provided with elphbolt, it is shown as below:
INTEGER, ALLOCATABLE :: ndegen_g (:,:,:,:).

Thus, an error occurred during compilation:
"ephwann_shuffle.f90(427): error #6351: The number of subscripts is incorrect. [NDEGEN_G]"

Best,
pfliu

[nakib]

I see. Thanks a lot for bringing this to my attention. Indeed the structure of ndegen_g has changed in EPW of QEv7 onward. One can see this by looking at this tensor in the subroutine ephwannblochp of wan2bloch.f90, for example.

I will need some time to figure out what is happening. In the mean time, you can continue using up to QEv6.7MaX-Release.

Best,
Nakib

[suan12]

Thanks. I am looking forward to your good news.
Best,
pfliu