Interpolation of Hessian in new version #97
-
Dear developers, we have already talked about this topic here, and I'm referring to the H3S example treated there: Since there have been a few bugfixes and branch merges, I'd have some follow-up questions for the new code version (SSCHA 1.2, CC 1.1). I have finished a calculation on H3S in the 222 supercell (SSCHA, 3rd and 4th Hessian), and I would like to interpolate the Hessians to a higher grid like 444 or 666. I managed to interpolate the SSCHA auxiliary matrices to the higher grids with the help of the harmonic support matrices and I have the real-space 3rd order file coming from the 222 calculation.
Thanks in advance. Best wishes, |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
Hi Roman,
Method 2 is more precise and does a better job as you are interpolating the 2nd and 3rd order force constants to evaluate the hessian in each specific q point. However, as it is implemented, method 2 does not allow interpolating the 4th order force constant. So if you see differences between using get_free_energy_hessian with and without include_v4, you must use method 1. The interpolation of the fourth-order force constants is much more complex and probably requires a brand new algorithm to compute the Hessian matrix with CC.Spectral.get_static_correction_along_path. We are working on it and will release this feature as a major update of the code, like version 2.0, but it will likely take some time. Bests, |
Beta Was this translation helpful? Give feedback.
Hi Roman,
There are two ways.
CC.Spectral.get_static_correction_along_path
Method 2 is more precise and does a better job as you are interpolating the 2nd and 3rd order force constants to evaluate the hessian in each specific q point. However, as it is implemented…