[Phonopy-users] Reducing drift in force constants

[Greene, S. <sam...@au...>]

Hello,

I understand that the “Max drift of force constants” reported by phonopy is an indication of how accurate the underlying DFT calculations are. I’ve been running some calculations in which the magnitude of the max drift seems quite large, at least relative to other examples I’ve seen online. In my calculations, the magnitude is between 0.2 and 1.3. I’ve tried increasing the accuracy of my DFT calculations (including the structure relaxations I perform before running phonopy) by increasing the k-point sampling density and plane wave basis energy cutoff, but this does not seem to help. In some cases, it increases the magnitude of the max drift. Similarly, increasing the supercell size in phonopy helps in some cases but not in others.

Should I be concerned about these large magnitudes of the max drift, and if so, what can I do to try to reduce them?

Thanks in advance for your help,
Sam