[ase-users] DftbPlus calculator

[Jens Jørgen Mortensen]

On 04/13/2018 10:48 AM, Ben Hourahine wrote:
> Dear Jens,
>
>
>>> Just been looking at the ASE bindings to DFTB+ and noticed that the
>>> internal (total_energy), not the Mermin free energy (mermin_energy), is
>>> being collected from results.tag.
>>>
>>> This would mean that at finite electron temperatures there can be a
>>> system dependent discrepancy between the forces and the energy that ASE
>>> uses.
>> Most of the calculators in ASE work like that.  They return the energy
>> extrapolated to zero temperature, which is normally what you want.
>> You can get the free energy if you ask for it with force_consistent=True:
>>
>> https://wiki.fysik.dtu.dk/ase/ase/atoms.html#ase.Atoms.get_potential_energy
>>
>>
>> and structure optimizers do that - I think.
> Perhaps I'm misunderstanding the routine in the ASE repository, but it
> doesn't collect the zero temperature extrapolated energy from DFTB+
> either (don't think we export that in our results.tag file at the moment
> anyway). To extrapolate the exported energy to T=0K, doesn't this
> requires the entropy term?

That is correct.  I would say that our DFTB+ interface *should* return 
the T=0K energy, so this should be fixed.

Jens Jørgen

>
> Regards
>
> Ben
>