Correspondence between AIMD and static relaxations

[apolet]

Hello!
Am a bit perplexed to see substantive differences in forces (as taken from OUTCAR with ASE) computed when I take an AIMD frame and run a static electronic relaxation on it. All conditions are the same: EDIFF=1e-4, ALGO=VeryFast. All other parameters are identical. Electronic convergence is reached in both cases. The calculation is collinear spin-polarized, but all magnetic moments are rigorously zero (only Ca, O, and H atoms are in the calculation). r2SCAN+rVV10, 2x2x2 k-points, again everything identical.

This occurs independently of the thermostat (Langevin / npt vs Nosé / nvt). Changing POTIM for the electronic relaxation does not matter. Changing the algorithm introduces only minute differences and does not change the overall discrepancy.

Here is an example plot of the difference in forces: https://drive.google.com/file/d/1ORjgOc ... sp=sharing

What could be the reason for this?
Are there forces from the changing box walls in npt or otherwise thermostatting that need to be accounted for?
Am I making some form of a mistake?

[ferenc_karsai]

Please upload your calculations according to the forum guidelines (POSCAR, KPOINTS, INCAR, POTCAR, OUTCAR and OSZICAR for both).