Strange kinks in phonon dispersions for acoustic modes near gamma point when including bubble correction

[vivekgusain09]

Dear Dr. Tadano,
Warm greetings. I am writing this query regarding the strange features i am getting in the phonon dispersions near gamma point when I am adding the bubble corrections to the dynamical matrix. The dispersions obtained using the SCPH only calulation show the expected trends and exhibit the hardenning of the phonon modes in the acoustic region. However, when i am adding the bubble corrections on top of the scph dynamical matrix I am getting strange results in the small frequency region near the gamma point with the dispersion exhibitting softening and then hardening with kinks in the phonon dispersion. I have attached the results for both scph and scph+bubble here with this question. I have used KMESH_INTERPOLATE = 2 2 2 MESH_SCPH = 12 12 12 for scph and KMESH_INTERPOLATE = 2 2 2, KMESH_SCPH = 8 8 8 for my calculations. I have tried it using with and without the non analytic term corrections and the kinks in the dispersions are unaffected. I have tried some tests with changing EPSILON and also the MAXALPHA but i could not obtained a smooth phonon dispersion.
My question are: 1. Is it an artifact of numerical calculations or something is wrong with my methods?
2. Can i trust the lattice thermal conductivity computed using the scph+bubble second order renormalized force constants (which are giving such strange kinks in the phonon dispersions)?
I kindly request for your help and guidance regarding this issue. Any response/hint is highly appreciated.

Thank you in advanced.

[ttadano]

Probably, the phonon frequencies get lower after the bubble correction, making the interpolation to small-q more difficult (or can indicate a signature of some sort of phase transition).

If the number of q points that sample these small-q unstable phonon region is small, the resulting thermal conductivity values may be reasonable.