Question About Differences Between Phonopy and SSCHA Harmonic Phonon Dispersions

[masoumehmahmodi-debug]

Dear SSCHA team,

I hope you are doing well. I am writing to ask about an issue I encountered when comparing the harmonic phonon dispersion from Phonopy with the harmonic dispersion generated inside SSCHA.

My workflow was as follows:

I computed FORCE_SETS using Phonopy.

I converted FORCE_SETS to FORCE_CONSTANTS using Phonopy’s built-in tool.

I then provided these FORCE_CONSTANTS to SSCHA, defining the harmonic force-constant matrix so that SSCHA could plot the harmonic phonon dispersion.

I used the same K-path in both Phonopy and SSCHA.

However, the resulting dispersions do not match. The black curves (Phonopy) and the dashed blue curves (SSCHA harmonic) show noticeable differences, even though they should, in principle, be identical because they are based on the same force-constant matrix and the same K-path.

Could you please advise why these discrepancies may appear?
Is there any additional step needed for ensuring that SSCHA interprets the FORCE_CONSTANTS in exactly the same way as Phonopy?
Or is there any format/normalization/units issue that I may be missing?

Thank you very much for your help.
Masoumeh

[mesonepigreco]

Hi, unfortunately the subroutines that read the FORCE_CONSTANTS file in the sscha code do not work. The new version of the code raise an exception if the code is called. You should convert in the quantum espresso dynamical matrix format before processing it further with the sscha.

[masoumehmahmodi-debug]

hank you very much for your reply. I apologize — I did not fully describe my workflow in my earlier message.

Here are the exact steps I followed:

I generated FORCE_SETS using VASP and Phonopy.

I converted FORCE_SETS → FORCE_CONSTANTS using Phonopy.

I then converted these FORCE_CONSTANTS → QS dynamical matrices required by SSCHA.

After that, I started the SSCHA calculations using this harmonic matrix as input.

I have attached the relevant files here, including the dynamical matrix produced in step 3.

Despite following these steps, the harmonic phonon dispersion inside SSCHA (blue dashed lines) still differs from the Phonopy harmonic dispersion (black lines), even though they originate from the same force constants and the same K-path.

Could you please check whether there is anything incorrect in the way I converted or supplied the dynamical matrices, or if there is an additional requirement I may be missing?

Thank you again for your help and support.

Best regards,
Masoumeh
Archive 5.zip
Archive 5.zip

[DjordjeDangic]

Hello,

I can not see the graph you are describing. In Methods.py there is a function sscha_phonons_from_phonopy(phonon) where phonon is Phonopy object. Have you tried that one?