[CP2K:1794] Re: Total charge of the system

[Ondrej Marsalek]

Hi,

On Wed, Feb 25, 2009 at 00:27, Axel <akoh... at gmail.com> wrote:
>
>
>
> On Feb 24, 5:38 pm, Ondrej Marsalek <ondrej.... at gmail.com> wrote:
>> Dear all,
>>
>> let me say first, that I know the drill and will send a proper bugreport
>> and/or request for help, if needed. This is more of a quick sanity
>> check.
>
> it looks like the MOMENTS calculation/output does not support ROKS at
> all.
>
> the part of the output that matters for your total charge is this part
> (cut'n'pasted from a closed shell +1 charged system).
>
>  Total electronic density (r-space):        -23.9999999997
> 0.0000000003
>  Total core charge density (r-space):        25.0000000000
> 0.0000000000
>  Total charge density (r-space):
> 1.0000000003
>  Total charge density (g-space):
> 1.0000000003

I did not really think that the charge of the system in the simulation
is 128. It just looks like to moments calculation is not right. I have
the correct output here:

  Total electronic density (r-space):       -256.9999994523        0.0000005477
  Total core charge density (r-space):       255.9999999919       -1.0000000081
  Total charge density (r-space):                                 -0.9999994604
  Total charge density (g-space):                                 -0.9999994604

> i don't know how many people actually do use ROKS in cp2k, but it is
> very likely a very small number compared to the total number of
> users.
>

For starters, anyone who wants the self-interaction correction :-)

> as a consequence, you probably have to be quite paranoid (as your
> question already indicates) with almost everything that depends on or
> process the wavefunction. it is probably a good idea to have a minimal
> test system input at hand, so that potential problems can be tracked
> down and corresponding regression tests added to the distribution.
>
> with a package as large and as complex as cp2k, it is next to
> impossible to have tests for everything up front (or you would not get
> any programming done).
>

And I am not complaining, of course.

Ondrej


> cheers,
>   axel.
>
>
>>
>> In a ROKS calculation of a system with an overall charge of -1, I get
>> the following output, requested by the section:
>>
>> CP2K_INPUT / FORCE_EVAL / DFT / PRINT / MOMENTS
>>
>>  ELECTRIC MOMENTS
>>   Reference Point [Bohr]            18.98836680     18.80082624     18.91122759
>>   Charges
>>     Electronic=     -128.0000      Core=     256.0000      Total=      128.0000
>>   Dipoles are based on the traditional operator.
>>   Dipole moment [Debye]
>>     X=      1.8180    Y=      2.7291    Z=     -1.0386     Total=       3.43971
>>   Quadrupole moment [Debye*Angstrom]
>>                 XX=   -297.7171         XY=     11.7868         XZ=      3.6787
>>                 YY=   -266.3743         YZ=    -22.9058         ZZ=   -263.2844
>>
>> Is that just me or is there really half of my electrons missing? If so,
>> I'll create a proper (small and fast) input that reproduces this.
>>
>> Regards,
>> Ondrej
> >
>