What has changed concerning the meta GGA implementation in the new VASP version 5.4.1 in comparison to VASP 5.3.5?This question arised for example when calculating bcc-iron with the TPSS functional in both versions: We get an energy(sigma->0) of -30.88694853 eV with the new version, an energy of -30.85589926 eV with the old version (both after fully optimizing the structures). Moreover, when starting a run using the new version 5.4.1 with a structure fully converged with the TPSS functional in the old version 5.3.5, the new version starts doing a lot of structural optimization steps (>50) which should not be necessary at all.
So, the question is: was the old version buggy, should we throw away all old calculations?
Or is there maybe something wrong with the new version?
I didn't find the right solution from the Internet.
App video company