Question on this: definition of force constants (orbits) change discretely as volume changes

Dear whom it may concern,

thank you very much for the hiphive package! This is one of unique packages for TDEP, which can explore phonon at finite temperature with very user-friendly interface.

We calculate quasiharmonic approximation using hiphive, that is, many TDEP matrices are generated at different volume points. The structures are the same. The system is Al fcc.

Here, opt.parameters and the definitions of force constants (orbits) change discretely as volume changes. Orders and elements of eigentensors change discretely as volume changes. In the end, interestingly, the elements of hessian matrices change continuously as volume changes, and this is what we use for the next step.

Is this an expected feature? It seems that generation of reduced force constants using symmetry give different but equal definitions of orbits. Many equivalent symmetry reductions are possible?

We want to somehow understand this more, because at large volume, the calculated free energy F seem to behave differently from well-converged phonopy results. We did not check everything yet. d^F/dV^2 was in good agreement with phonopy results when cutoff included 1st nearest neighbor. However, d^F/dV^2 was somewhat different when the cutoff was increased to include 3rd nearest neighbors.

Thank you very much!