[Wien] Large oscillations in total energy

[Pavel Ondračka]

Laurence Marks píše v Čt 07. 12. 2017 v 07:37 -0600:
> Probably overlapping response from Peter...
> 
> Based upon the Check-mixing output, the density is very well
> converged and did not change with the jump in energy. The most likely
> way for :DEN alone to change, and nothing else, is that there was a
> change in how EXC is calculated, for instance using a metaGGA or
> changing case.in0, without a change in VXC.
> 
> Do "grep :POT *scf" to check this.
> 
This is a standard PBE calculation:
:POT  : POTENTIAL OPTION EX_PBE EC_PBE VX_PBE VC_PBEfor all the cycles.
In fact the surprising thing was that this started out of the blue
after at least 15 iterations when the calculation was already looking
quite converged.

BTW as mentioned in the beginning this might have to do something with
the atomic spheres, I converged this structure with the almost touching
spheres (0% reduction) and later I did reduce_rmt_lapw -r 3.5 to
produce this case, hence it was already pre-converged and I only wanted
to fine-converge it, however the energy oscillations showed...
Dunno of this might play a role.

Best regards
Pavel