[CP2K:10716] Unphysical results with custom Mg basis set

[Krack Matthias (PSI)]

Hi Richard

Firstly, there are basis sets for Mg available in the data file cp2k/data/BASIS_MOLOPT_UCL prepared by S. Ling. These might be a good starting point for further customisation. The ATOM module of CP2K provides tools to optimise GTH PPs and Gaussian basis sets compatible with CP2K. You may want to have a look at this tutorial<https://www.cp2k.org/exercises:2015_cecam_tutorial:basis_set_optimisation_using_optimize_basis>. 4 or 5 primitive Gaussians should be sufficient for Mg GTH-q2 PP to obtain a converged atomic energy and roughly 8 primitive Gaussian functions should be fine for a Mg GTH-q10 PP with 10 valence electrons. More primitive Gaussians will only increase the computational cost and introduce linear dependencies into the basis set. Note, that the accuracy of the Mg GTH-q2 PP with just 2 valence electrons is clearly inferior compared to the GTH-q10 PP. That might matter when Mg plays any role in your simulation beyond being just a counterion.

Matthias

From: cp... at googlegroups.com <cp... at googlegroups.com> On Behalf Of Richard Fogarty
Sent: Dienstag, 11. September 2018 12:32
To: cp2k <cp... at googlegroups.com>
Subject: [CP2K:10716] Unphysical results with custom Mg basis set

Dear all,

I'm trying to use a custom basis-set for Mg in CP2K but am getting unphysical results; I suspect the reason is that i'm making a mistake with the basis set formatting. The simplest unphysical result is for a calculation on a single atom, whereby the ground state has p-orbitals (rather than s) fully occupied (see attached *.inp and *.cpout files).

The basis set is generated in an external program using the following steps:
1) A calculation is carried out on an Mg atom with the s-shell occupied (neutral atom, no net spin). This calculation yields a list of grid points (distances from the nucleus, due to the spherical symmetry), and the value of each shell at these grid points (which I'll call \phi_{r}). I'm using the same pseudopotential ("Mg GTH-PBE-q2") in this calculation as in the subsequent CP2K calculations.
2) The value of \phi_{r} is multiplied by sqrt((2l+1))*(r**l)
3) Gaussians are fit to the data from step 2.

Im loading the exponents and coefficients from step 3) into CP2K. One example is attached (file MG_TEST_BASIS). Note that the basis set attached leads to reasonable results for a single atom in the external program (where i know im entering it in the correct format).

Any help would be greatly appreciated.

Thanks in advance,
Richard


Misc info:
This is using cp2k-6.1 in serial mode (cp2k.sopt executable)
--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns... at googlegroups.com<mailto:cp2k+uns... at googlegroups.com>.
To post to this group, send email to cp... at googlegroups.com<mailto:cp... at googlegroups.com>.
Visit this group at https://groups.google.com/group/cp2k.
For more options, visit https://groups.google.com/d/optout.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180911/6e013699/attachment.htm>