My Community

I used default EDIFF and EDIFFG to relax the atoms.
But, when the total energy change between two relaxation steps is less than 1E-3(I am sure I check it carefully),the calculation still continues... I don't know why?

there is always one more ionic test step added after the the ionic convergence is reached according to EDIFFG. This is done in order to check whether the geometry is really converged really or whether the last step just was too small to cause significant changes in TOTEN. Only if the convergence criterium is fulfilled for that step additional as well, the ionic optimization procedure stops.