I'm doing a relaxation now and find that the pressure oscillates by +-0.5kbar
or so with a period of maybe 6 iterations. The oscillation doesn't damp out
quickly so it's taking forever to converge. It's a structure with 104 atoms
run on an old Linux box with 8 processors (I'm using 7 to be safe), so it's been
4 days and still going.
Is there interesting physics behind the oscillation? Is there a parameter
which could be tuned to make it converge faster?
My strategy is to pick reasonable values for the convergence parameters, relax
the structure, then converge the relaxed structure and re-relax it if
I find that I need to use different parameters than the ones I guessed. The
idea is that the convergence tests are best performed on something close to
the actual system, which I take to mean with the atoms in more or less the
right places. Does this make sense?
--
Sincerely,
Matthew Marcus
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso)
users mailing list [email protected]
https://lists.quantum-espresso.org/mailman/listinfo/users
