Nitrogen molecule with PP and AE atoms.

[Vladimir_Konjkov]

Dear Casino developers.


I want to perform a test calculation of a nitrogen molecule in which one atom has a pseudopotential and the other does not.
I attach the input and output files for ORCA that were used for the calculation of gwfn.data.
In ORCA calculation dipole moment differs slightly from zero, which confirms the quality of the pseudopotential.

Code: Select all

-------------
DIPOLE MOMENT
-------------
                                X             Y             Z
Electronic contribution:      0.00000      -0.00000      -2.07568
Nuclear contribution   :      0.00000       0.00000       2.07927
                        -----------------------------------------
Total Dipole Moment    :      0.00000      -0.00000       0.00359
                        -----------------------------------------
Magnitude (a.u.)       :      0.00359
Magnitude (Debye)      :      0.00913

To reproduce this result in a Casino, I need to mark one nitrogen atom as AE and the other as PP but I can’t find such an option in the documentation.
I could create a file with an empty pseudopotential for AE atom, but then the cusp condition for that atom would not be satisfied.
I'm completely confused, maybe I should come up with another test?

Best wishes, Vladimir.

[Neil Drummond]

Perhaps you could look at a "nitrogen phosphide" molecule, where the nitrogen is all electron (no n_pp.data file is supplied) and the phosphorus pseudopotential in p_pp.data is actually just a renamed nitrogen pseudopotential?

To create this p_pp.data file you could just rename n_pp.data as p_pp.data, and change the atomic number at the top from 7 to 15.

(Sorry if I have misunderstood what you are trying to do.)

Best wishes,

Neil.

[Vladimir_Konjkov]

Hello Neil

I write tests for my program, and since the input is the same as yours, so for yours. I need such a test so that there are atoms with and without pseudopotential. I can take B2H6 and set the pseudopotential only for hydrogen, the energy should be the same as for B2H6 without the pseudopotential (also with Jastrow, backflow). And I wanted to take a diatomic molecule like (N2. O2, H2, F2) for the test, but it didn't work out that way.

A pseudopotential must be assigned to a group of atoms with the same atomic number, and there can also be a group of atoms with the same atomic number without a pseudopotential. Grouping is carried out through atomic number + 1000 * <group number>, and simply atomic number is the 0th group.
Accordingly, the x_pp.data pseudopotential belongs to the zero group, and the x<group number>_pp.data pseudopotential belongs to <group number>.

You need to do it in such a way that if there is no x<group number>_pp.data file, then assume that the group of atoms does not have a pseudopotential, and now this is regarded as: the user forgot to copy the file with the pseudopotential.

Best wishes.

Vladimir.

[Neil Drummond]

CASINO allows you to have both all-electron and pseudopotential ions present. For an example, see

CASINO/examples/molecule/silane/ground_state

which has a silicon pseudopotential, while the hydrogen atoms are all-electron.

Before 2014 there used to be a mishmash of CASINO keywords related to allowing AE calculations in plane-wave bases and allowing a mixture of AE and pseudopotential calculations, etc. Mike removed these keywords in CASINO v2.13.379 (884fd75db68bff).

Best wishes,

Neil.

[Vladimir_Konjkov]

CASINO allows you to have both all-electron and pseudopotential ions present. For an example, see

CASINO/examples/molecule/silane/ground_state

which has a silicon pseudopotential, while the hydrogen atoms are all-electron.

I can do that! I cannot make it so that in a nitrogen molecule one atom has a pseudopotential, and the second does not, although such a calculation makes sense in principle. Сan you show me the input for this calculation?

[Neil Drummond]

Does the trick of looking at a N"P" (nitrogen "phosphide") molecule, where the nitrogen is all-electron and the phosphorus is actually a pseudonitrogen atom not work?

Best wishes,

Neil.

[Vladimir_Konjkov]

Does the trick of looking at a N"P" (nitrogen "phosphide") molecule, where the nitrogen is all-electron and the phosphorus is actually a pseudonitrogen atom not work?

Best wishes,

Neil.

It seemed to me that this was some kind of puzzle, I need time to think about it.

Best wishes Neil. You make my day.

[Mike Towler]

Hi Vladimir,

Have a look at my DIARY entry for patch 2.13.379 as Neil says, which contains an extensive discussion of the issue. Precisely which atoms are
pseudoatoms should be flagged by the external code, not deduced by CASINO.

Note that you can flag a pseudoatom by adding 200 to the atomic number in the xwfn.data file, though I can't remember
if CASINO's internal logic allows for a mixture of atoms and pseudoatoms with the same atomic number - probably not.

Cheers,
Mike

[Vladimir_Konjkov]

Great to read you, Mike.
The manual says 6.1.2 Pseudopotentials:

On the rare occasions when you might want to use two or more different pseudopotentials for atoms
with the same atomic number (say in a surface, and in an atom or molecule absorbed on that surface),
then you may use additional pseudopotentials renamed as, e.g., xx2 pp.data. Different types of
pseudoatom are flagged in xwfn.data by adding multiples of 1000 to the original atomic number,
e.g., atomic numbers 12, 1012 and 2012 correspond to atoms using pseudopotentials mg_pp.data,
mg2_pp.data and mg3_pp.data, etc.

So
mg_pp.data maped to 12 in xwfn.data and
mg2_pp.data to 1012 and
mg3_pp.data to 2012

I do as Neil suggests and it works, but I didn't support this in my code.

I have such things going on here that I don’t have time for pseudopotential.
https://www.dw.com/en/why-are-so-many-r ... a-68025856

All the best, Vladimir