[CP2K:53] GAPW method

Marcella Iannuzzi marcella.ia... at psi.ch
Mon May 7 19:12:23 CEST 2007


GAPW is based on the partitioning of the total charge density, as PAW  
It can be used with both PP or full potentials (AE), also in a mixed  
way: some atoms PP some others AE.
If you use PP, it is more like GPW, i.e. it uses GTO basis sets  
optimized for  GTH PP.
There are no ultrasoft PP in cp2k, and the PW representation of the  
density is used only for the calculation
of the density dependent parts of the potential (hartree and xc)
The advantage of using GAPW rather than GPW with PP should be the  
possibility to
run with a lower PW cutoff, because the hardest part of the density  
is never collocated on the grid.
However, GAPW is  mainly intended to perform full potential  
calculations, where the core electrons are
considered and never-frozen.  This should be useful to address  
properties that depend also on the relaxation of the cores.
The basis sets in this case are the standard GTO all-electron basis  
sets, as used also
  by other codes like Gaussian, obviously no PP is
required.  Therefore if you want to compare results you can compare  
with Gaussian.
As far as I know, PAW is often used with frozen cores. In the present  
implementation, for GAPW this is never the case.
One important difference with respect to PAW is in the construction  
of the local densities.
In GAPW these are expanded in the primitives centred on the specific  


On May 7, 2007, at 18:38, Nichols A. Romero wrote:

> Hi,
> I am trying to understand the GAPW method as described in the article:
> Theor. Chem. Acc (1999) 103:124-140
> The basic idea is to reduce the amount of work performed in  
> calculating
> E^H[n(r)] and E^XV[n(r)] by decomposing the density into soft and hard
> contributions, i.e.
> n = \tilde{n} - \tilde{n}^1 + n^1
> This methods can be applied using AE or PP approach. Though applying
> GAPW with PP would be more common.
> Is this method really more like PAW (Bloechl et. al) or is more  
> like an US-PP
> approach? For example, GAPW with PP has a frozen core and it would not
> be possible to use it an relax-core PAW type of method?
> Thanks and pardon my ignorance.
> -- 
> Nichols A. Romero, Ph.D .
> 1613 Denise Dr. Apt. D
> Forest Hill, MD 21050
> 443-567-8328 (C)
> 410-306-0709 (O)
> >

More information about the CP2K-user mailing list