Calculation of electron-phonon matrix elements (Many-body pertubation theory approach)

[arta_schellhorn]

Hello everyone,

I've been trying to evaluate and save the electon phonon matrix elements
following the Many-body pertubation theory approach workflow. I've
completed the first step which is to create the displaced potentials using
phelel. However, I wasn't succesfull continuing creating the matrix
elements in the second step via calculating a physical observable, in my
case I tried "Bandgap renormalization due to electron-phonon coupling".
I've put ELPH_DRIVER = mels to store them.
Do you have a small example script/ Input files which you could provide
that sucseccfully lead to the calculation and saving of the matrix elements
in the vaspelph.h5 file; especuially the INCAR andKPOINTS_ELPH files?

In the INCAR file I have set the tags:
ELPH_MODE = renorm
ELPH_DRIVER = mels
ELPH_DECOMPOSE = VDPR

For the KPOINTS_MELS file I have chosen 3 points between two high symmetry
points.

Unfortunately none of my calculations succeeded. They always run out of
time after two weeks, however don't seem to update the outcar file during
the run time. Also no vaspelph.h5 file has been created.

Although the ELPH_DRIVER tag is set to the option that should lead to the storage of the matrix elements (if they get calculated..), it doesn't seem to do so and not even an empty storage file is created.

I would be very greatful for your support.
I'm looking forward to hearing back from you.

Many thanks in advance and best regards.

[jonathan_lahnsteiner2]

Dear Arta Schellhorn,

I was checking your INCAR file. I recognized the following things:

• The tag elph_writemels= .TRUE. does not exits.
• You don't have to specify ELPH_MODE = renorm.
• NELM = 0 should not be set. Just use the default value for the number of electronic minimization steps.
• Setting NSW=0 is also not necessary.

This are just some hints but most likely will not result in a solution to your problem. To give you further guidance please upload your OUTCAR file. Otherwise we can only guess what your issues might be.

All the Best Jonathan

[arta_schellhorn]

Hello Jonathan,

thank you for your quick reply!
Usually for this calculation the outcar file does get deleted when the calculation runs out of time after reaching the maximum calculation time in the cluster. However, I was able to copy the outcar file out of the calculation node for a very similar calculation compared to the one I have previously uploaded. The only difference lies in the INCAR file in the number of N_Bands, where in the INCAR file corresponding to the attached OUTCAR file the number of bands is not restricted since ELPH_NBANDS is not set (while in the originally attached INCAR file I have restricted the bands to "ELPH_NBANDS = 5"). Plus here I have not set:
ELPH_RUN = .TRUE.
elph_writemels= .TRUE.

Other than that the files are the same and the issues regarding not being able to create and store the matrix elements or even "finish" the calculation seem to be identical.
I attache the OUTCAR file for you here.
What I have noticed is that the file didn't seem to update at all viewing it alsmost directly after submission compared to viewing in six days into the calculation.

The reason I have added the tag "elph_writemels= .TRUE" in the previous INCAR file was due to it being shown as "elph_writemels= f" in the OUTCAR file. I thought that this might have been a reason why there was no file created which contains the matrix elements.

Is there a similar minimal vasp example/Input files and Output files where the matrix elements have been created and stored succesfully?

Thanks you for your support!

Best regards,
Arta Schellhorn