Follow-up Qns. to {Re: [Wien] zero energy} + cohesive E

[Stefaan Cottenier]

> Does that mean we can compare energies of different systems by calculating
> the average of the potential in the interstitial region? ...  as opposed,
> say, doing one atom in a large cell calculations for all constituent
> atoms... which BTW does not work at 30 bohr, as the FAQ suggests. NN
> crashes with anything more than 28.25 bohr (tested only with Al & Cu).
> Has anyone else run into the same problem while doing cohesive or free
> atom energy calculations?

I think you are mixing two things now. The question of Kevin was for the
eigenenergies (energy of a particular state). The zero of the energy scale
on which these energies are measured is the average of the potential in the
interstitial region (for instance the E_l in case.in1 or the fermi energy
are given with respect to that energy).

The energy that plays a role for calculating cohesive energies is the total
energy :ENE, for which the zero has an absolute meaning: if you would bring
all electrons and all nuclei of your unit cell at infinite distances from
each other, then you would need :ENE Ry.

You're right about nn: it crashes as soon as the interatomic nn-distance is
larger than 20 bohr, which for an fcc lattice happens for a lattice constant
larger than 28.25. Strange, wasn't this limit increased in one of the
updates?

Stefaan