search

tdep-developers / tdep-tutorials

TDEP Tutorials
Creative Commons Zero v1.0 Universal
20 stars 18 forks source link

Using LAMMPS as the force calculator #57

Closed antonf-ekb closed 6 months ago

antonf-ekb commented 7 months ago

Dear community! I was trying to reproduce the tutorial n. 2 "sTDEP with DFT forces" for a system I am interested in. It is stated "...move the initial sample (contcar_conf0001) to that folder and compute forces using your force field or DFT of choice". I supposed I can use LAMMPS simulation as a force calculator since it is much faster than DFT. However, I find that --output_format for the canonical configuration still doen't support LAMMPS (although here I found some information on how to run LAMMPS simulation to obtain TDEP input data). I managed to convert CONTCAR format to LAMMPS input myself, however, the initial velocities of atoms from contcar_conf0001 are lost in this case. So my first question is - are these velocities essential in order to obtain correct data for TDEP? I suppose yes because after running the simulation, postprocessing as reccomended on the above mentioned page, and running extract_forceconstants -rc2 0 2>&1 | tee extract_forceconstants.log I obtain very low R^2 value, and unphysical elastic constants and dispersion curves. May be I do something wrong, so your advices are highly appreciated. I attach extract_forceconstants.log here and can provide additional files of the calculation if needed.

Best regards, Anton. extract_forceconstants.log

antonf-ekb commented 6 months ago

Issue is considered here https://github.com/tdep-developers/tdep/issues/58