Project atomic displacement onto phonon modes

[lucydot]

Hi Jarvist,

I'd like to assess which phonon modes contribute to a given atomic displacement. It should be relatively straight forward for JuliaPhonons to read in two POSCARs, calculate the atomic displacement, and project this displacement onto a set of phonon eigenvectors.

I'm happy to code this up and submit as a PR, if this is something you'd also find useful in JuliaPhonons?

[jarvist]

Should be doable! The issue is that the phonon eigenvectors are 3N, with an atomic basis built as Phonopy chooses, so you need to construct a similar vector (which you then can do prodoct with the eigenvectors).
This should be just in a simple form of [A_x, A_y, A_z, B_x, B_y, B_z] etc. but I haven't really tested!

If you're just interested in which mode gives, e.g. atom A a kick, you can reuse basically all of the code from the mode-decomposition part of the code, and not aggregate the data by atom type.
E.g. if you look at the case for MAPI, you can immediately see which modes have lead motion https://github.com/jarvist/Julia-Phonons/blob/master/plot-mode-decomposition/MAPI_mode.png

Depending whether you re-apply the mass-weighting or not, you get an answer in terms of atomic displacement, or energy fraction of the phonon mode.

[hnwglsh]

Hi lucydot ，
Do you have solved the problem？

[lucydot]

Hello - It's a work in progress, on a branch in a forked repo here. It's completely untested, and sitting in a Jupyter Notebook. I'll keep this thread updated with any progress. It may take me a little time to think how best to integrate this into the existing code (I'm new to Julia) - Lucy