[Pw_forum] FFT grid and cutoff energy.

[Nicola Marzari]

On 6/1/11 2:02 AM, Hongsheng Zhao wrote:
> //Hi all,
>
> //I've learned that, for all of the PW based DFT codes, such as,
> VASP/PWscf/Abinit/CASTEP, the
> FFT grid is generated based on the cutoff energy. So if we have tested
> the convergence total energy vs cutoff energy, then the
> FFT grid used for our calcultions should also be OK based on the
> selected cutoff energy. In other words, it is unnecessary to do a
>
> Independent FFT grid convergence test. Am I right?
>
> Regards.


Yes, indeed. In the old days one would use slightly
coarser meshes than what prescribed, and get away with it (especially
if that allowed for the FFT dimension to be a power of 2, for
which typically FFT used to be much faster); nowadays noone bothers.

FFTs dominate the cost of small calculations (relevant if you need
to do a lot of them) and can be tricky in massively parallel 
calculations, but in general I wouldn't bother.

The CP code in Q-E is slightly more sophisticated, and for ultrasoft
pseudopotentials it uses different FFT grids to transform the
softer wavefunctions, and harder charge density - not sure if other 
codes do the same (the technique is all described in Pasquarello 1993 PRB).

				nicola

----------------------------------------------------------------------
Prof Nicola Marzari    Department of Materials    University of Oxford
Chair of Materials Modelling  Director, Materials Modelling Laboratory
nicola.marzari at materials.ox.ac.uk     http://mml.materials.ox.ac.uk/NM