[Wien] Energy, forces and muffin tin radii

[Stefaan Cottenier]

> Let me pleae ask you a further question:
> when the muffin tin sphere become almost touching ( within a  structure
> optimization) and produce inconsistency between energy differences and
forces,
> can we still rely on the forces vaules while discarding the energy
difference
> ones?

I don't know, as a matter of principle I would be cautious with anything
resulting from a calculation with almost touching spheres. Two things:

1) You can look at your forces as a function of optimization step: do they
behave continuously? If so, is your last force following that trend, or not?
You could also do this for other properties (maybe you have some :EFG's in
your structure?). If everything behaves 'normally' in the last step with
touching spheres, then probably you can trust these last forces.

2) But probably you don't need to do this effort. Restore the last step,
reduce the muffin tin spheres, use clminter (UG sec. 9.8), and restart
min_lapw (see http://www.wien2k.at/reg_user/textbooks/Optimization-Notes.pdf
on how to do that most efficiently). That will bring you to the real
optimized structure.

Stefaan