Open ndaelman-hu opened 11 months ago
@ladinesa Should I add the test referenced in the original issue?
Thanks for the fix. Please also add a test for this, maybe one outcar and one vasprun, you can also use this for the atom positions fix.
Something to consider in the future, when we clean the VASP parser. The VASP mainfiles have very different structures. Just considering the case of the frequency analysis:
OUTCAR
: k-points and eigenvalues are repeated for each ionic update + k-points are repeated at the endvasprun.xml
: k-points (list) are repeated at each ionic update + at the beginning (under <incar>
). The (electronic) eigenvalues are only printed in the last calculations step (maybe also for under the projected
tag)When it comes to eigenvalues, we mostly just need the no. irreducible k-points (at that step) to restructure the eigenvalues. Atm, we are extracting all the k-points in full, at each step, and this might be overkill. Full extraction is only needed at the last step.
Tackle changing symmetries (and k-point set) between ionic updates in newer VASP versions (about 6). This solution packages
kpoints
into lists matching the no.calculations
. The cases of single-point calculations and the older format are also handled (forvasprun.xml
). No explicit information about the version is used.Closes #148