Save previous ionic step CHGCAR, etc?

[d-farrell2]

Hi All,

Myself and some other members of my group noticed that if our relaxations die in the middle of an ionic step (during the electronic relaxation), that the WAVECAR and CHGCAR files are blank. This, of course, leads us to have to restart from just the CONTCAR from the previous step, which means that we don't have as nice a starting place as we would with the WAVECAR and CHGCAR.

I am curious - is there any easy way to keep these files from a previous step? Why are they cleared before the ionic step actually finishes?

[jlbettis]

Future Researchers,
VASP only prints supplementary output files (WAVECAR, CHGCAR) for fully converged calculations. In my experience with ionic relaxation calculations, first try to obtain convergence with less accurate electronic and force convergence criteria. In subsequent calculations increase the accuracy criterion for both the electronic and force convergence. Besides, the CONTCAR file which is updated after each ionic relaxation cycle is a perfect starting point for continuing any ionic relaxation calculation. However, I do understand d-farrell2's frustration as the initial convergence of the charge density is the most time consuming task of most ionic relaxation calculations.

[mforti]

The post is ancient but the topic is of the interest of all those that run vasp on hardware plugged to unstable electric systems.

the only way would be to modify the source code. Has anyone attempted this ?