GPW: strange behavior of cations

Axel akoh... at
Sun Jan 25 18:47:03 UTC 2009

On Jan 25, 4:56 am, Lukasz Cwiklik <cwi... at> wrote:
> Dear Ilya,
> I had the same problems with sodium atom/cation in water (almost 2
> years ago with an old version of cp2k). At that time I was discussing
> it with Fawzi and it seems that the problem was in too small cutoff
> value (as far as I understand it is simple - the small sodium cation
> needs a finer grid). I had to go up to 400 Ry to get rid of


just to clarify something. it is not the size of the sodium,
that requires the high (density) cutoff, it is the fact, that
you have a semi-core pseudopotential, i.e. there are 9 valence
electrons (not only one). that makes the pseudopotential very
'hard', i.e. very deep and thus you have a high "spike" of density
that requires a lot of plane waves to describe them well.

for cationic sodium, there is not much of an alternative
than this kind of pseudopotential. the regular one can only
be used for metallic sodium.

one remedy that cp2k offers in this case, would be to use
GAPW instead of GPW. GAPW is however less commonly used
(=> higher chances of running into 'undocumented features')
and requires more care in setting it up.

there also are options in cp2k to apply density smoothing,
but i have no idea how well that would work in this extreme
case. those options seem to depend too much on error cancellation
for my personal taste.


> unphysical 'shaking' of sodium. And yes - it made my simulations terribly long
> :-(
> If you find any other reason/solution of these problems with sodium,
> please, let me or the group know.
> Best,
> Lukasz
> --
> Lukasz Cwiklik

More information about the CP2K-user mailing list