The ABINIT Users Mailing List ( CLOSED )

[Anglade Pierre-Matthieu <anglade@gmail.com>]

Hi St Druid,

I may mistake on the following but this may be part of the answer to
your question on convergence criterion :

The density functional theory is used to produce a density based
ground state. That is to say the result of a DFT calculation is an
electronic density, directly related to the potential via Poisson's
equation. Therefore, convergence of a ground state calculation is
quite adequatly monitored through the change of density or potential
along the algorithmic search for self-consistent solution. Then for a
ground state calculation, the most straight forward  (and accurate)
convergence criterion is based on the change of hartree potential
(better than density since potential changes faster and is then less
susceptible to numerical errors). It is however mandatory for ground
state calculations to check that the wave functions are converged,
since in some cases the algorithms may encounter a "sticky" potential
that doesn't change (and matches the convergence criterion)
eventhough the wavefunction are unconverged, which in turns means that
the density is wrong ...

For response function calculation, one keeps the ground state (i.e.
density) and tries to evaluate the ratio of change of energy with
respect to a perturbation of the hamiltonian. In the limits where the
perturbation is negligible, the density is unchanged and only the
wavefunctions are perturbed to first order (see Phys. Rev. A 52, 1086
- 1095 (1995) for an accurate explanation). So response function
calculations are non-selfconsistent and consist in evaluation of the
response of the wave-function to a perturbation. Therefore, the single
accurate convergence criterion for such calculation is the change of
wave-functions.

Best regards

PMA



On Fri, Jun 5, 2009 at 9:40 AM, matthieu verstraete
<matthieu.jean.verstraete@gmail.com> wrote:
>> I am just wondering why the scf calculation are different between
>> phonon and el-phon.
> you attach an abinit input file to calculate phonons (this is fine).
> The electron phonon calculation is done later in anaddb.
>
> I still do not know what you mean by bad convergence etc... The phonon
> perturbation calculation is different from a ground state, there is no
> reason it should converge the same way, but it should converge...
>
> Again, please follow _all_ of the nettiquette
> http://wwwold.abinit.org/community/?text=netiquette
>
> Matthieu
>
>
>
> --
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Dr. Matthieu Verstraete
>
> European Theoretical Spectroscopy Facility (ETSF)
> Dpto. Fisica de Materiales,
> U. del Pais Vasco,
> Centro Joxe Mari Korta, Av. de Tolosa, 72,   Phone: +34-943018393
> E-20018 Donostia-San Sebastian, Spain        Fax  : +34-943018390
>
> Mail : matthieu.jean.verstraete@gmail.com
> http://www-users.york.ac.uk/~mjv500
>
>



-- 
Pierre-Matthieu Anglade