Using WAVECAR from prior ionic steps after a crash

[argon214]

I am running relaxations in VASP and occasionally the job will crash. The issue is that the first SCF step is always extremely costly unless I start from a WAVECAR, but I don't know when VASP will crash. Right now, I'm thinking about running the job in a loop and setting nsw to something small (like nsw = 20) and having it write the WAVECAR every 20 steps. Then if it crashes at step, say, 7 I can just restart using the CONTCAR but the WAVECAR from 7 steps prior (which hopefully isn't too different). Is there any better method than this?

[admin]

yes

[admin]

yes, you can restart if only the positions of the ions are updated.
If include the volume and the shape of the cell in your relaxation, the
FFT grids might not be compatible.