Doubt related to Acoustic sum rule (sumRuleFC2)

[AnishDas-2025]

Dear experts,

We generally use the ASR (sumRuleFC2) during the electron or phonon transport calculations, (appName = "electronWannierTransport" and "phononTransport"). However, during the transformation of the electron-phonon coupling matrix elements from the Bloch to the Wannier representation (i.e. appName = "elPhQeToPhoebe"), we also use 2nd order force constant (ex. phFC2FileName = "silicon.fc"). In that case while using appName = "elPhQeToPhoebe, should I use the ASR also (i.e. turn on sumRuleFC2 flag). I am asking this because, I generated the *.phoebe.elph.hdf5 file two ways viz. 1) not using sumRuleFC2 flag, and 2) using sumRuleFC2 = "simple". For both the cases when I calculated Sigma and Ke, on converged k/q grids (90x90x30 at 300 K). I am getting quite a lot deviation in the values (alomost 25%). In that case which values should I consider (applying ASR in elPhQeToPhoebe or "electronWannierTransport" or both)?? Below I attached the screen shot and also the scattering rate at 300 K.

Notre: The 2nd force constant file I did not apply any ASR in QE. The ASR is applied in the PHOEBE only. I noticed that scattering rate decreases when I apply the ASR additionally on "elPhQeToPhoebe", thus Ke and Sigma increases.

Thanking you
Anish

[jcoulter12]

Hi Anish,

Thanks for reporting this and showing me a clear demonstration of the effect.
I would apply it to both application inputs -- basically, the force constants are used to generate the phonon eigenvectors, which are used to rotate electron-phonon coupling into the basis of atomic displacements during the elPhQeToPhoebe application -- the same force constants should be used to regenerate the phonon eigenvectors to perform the reverse rotation during the transport calculation. If you're familiar with the Wannier interpolation procedure (see Giustino et al, PRB. 2007) this makes sense.

I should perhaps add a note or a warning about this, however, as it's easy to forget to put it in one or the other. I would suspect in most cases, it won't make such a large difference, though -- if the phonon force constants are a very good quality, the ASR won't change things so much.

Out of curiosity, can I see the phonon.in file used for this calculation?

Hope this helps,
Jenny

[AnishDas-2025]

Dear Dr. Coulter,

Thank you for your response and clarification. Certainly, I can share the inputs. "phonon.in" files for QE or Phoebe calculation or both? And is it OK if I share these in your personal email (given in your github account)? The discussion can be continued here.

Thank you
Anish

[jcoulter12]

Both is helpful if you don't mind -- I was mostly curious though to see the QE one, as I am wondering if maybe the phonons have a small number of points on the q-grid.

[AnishDas-2025]

Dear Dr. Coulter,
Thanks for your assistance. I sent those to your email.