My Community

Is there a way to specify convergence criteria for the stresses in VASP? It would be nice to control the stresses in the calculation similar to EDIFF in energy for electronic steps and EDIFF in the forces (or energy) for ionic steps.

Oops! That second EDIFF should be EDIFFG in the previous message.

please set EDIFFG to negative values. this applies to both, the forces on the atoms and the stress on the unit cell.

I usually use negative values for EDIFFG, but I thought that the value used only applies to the forces (i.e. the maximum force on an atom that stops further ionic steps). I don't understand how EDIFFG applies to the stress tensor.