[CP2K:1705] Some minor issues

[Teodoro Laino]

Dear Ondrej,
point (1) and (2) have been fixed.

Point (3) has been fixed as well in this way:
-) for ghost atoms you don't need to specify any pseudo. This was just 
an overhead in the input since internally
    those information were not used. Thanks for reporting this annoying 
problem.
-) vdw correction for ghost has been disabled (early this morning, 
independently from these fixes).
-) You need still to define the element. This was a choice made some 
time ago. It may look like unhappy in cases
like this one or in classical MD runs (since even there you don't need 
to know the element.. you only need the kind).
Element information are anyway used in quite few places, so we preferred 
some time ago, to have always the
full element specification for each KIND. This applies as well to ghost 
atoms.

Thanks again.
Teo

Ondrej Marsalek wrote:
> Hi everyone,
>
> I have some minor issues with CP2K. Nothing too bad, but perhaps someone
> would be interested.
>
> 1) In the manual, in the section
>
>    CP2K_INPUT / FORCE_EVAL / DFT / XC / VDW_POTENTIAL / PAIR_POTENTIAL
>
>    it says for the ATOMPARM keyword: "This required keyword can be
>    repeated and it expects a list of words." However, a simulation runs
>    just fine with defaults without a single ATOMPARM keyword.
>
> 2) Manual again, section
>
>    CP2K_INPUT / FORCE_EVAL / DFT / QS
>
>    EXTRAPOLATION_ORDER: "Order for the PS extrapolation..."
>
>    However, the setting is also relevant for the ASPC extrapolation.
>    While this is strictly speaking also PS extrapolation (if I
>    understand it correctly), it would probably be more clear if it
>    referred directly to the EXTRAPOLATION setting (and therefore
>    mentioned PS and ASPC).
>
> 3) Ghost atoms. It seems strange to me that you have to specify a
>    pseudopotential and chemical element for a ghost atom. It
>    specifically says at the GHOST keyword:
>
>    "This keyword makes all atoms of this kind ghost atoms, i.e. without
>    pseudo or nuclear charge.Useful to just have the basis set at that
>    position (BSSE calculations),or to have a non-interacting particle
>    with BASIS_SET NONE"
>
>    So whatever you specify for pseudo and element gets (hopefully)
>    ignored.  A somewhat bigger annoyance is ghost atoms seem to have a
>    dispersion correction by default. Is this really the case? I
>    certainly had to resort to the following ugly hack: specify an
>    element that is not present in my system for the ghost atoms and then
>    set the dispersion parameters to zero for that element (using
>    ATOMPARM). It seems to work, but it is not pretty.
>
> If any of the developers find this relevant, good. If any decide to
> change stuff, great and thanks in advance. I am especially interested in
> point 3), 1) and 2) are just details, but maybe useful for total
> newcomers.
>
> Best,
> Ondrej
>
> >
>