GTH pseudopotential from CP2K to abinit format

129 views
Skip to first unread message

Dan_M

unread,
Feb 8, 2021, 1:05:04 PM2/8/21
to cp2k
Dear all,

I am interested in performing some calculations with Abinit employing the same GTH pseudopotentials I am using now in CP2K. These GTH pseudos are different from those provided with CP2K and thus I have not the option of downloading them from the 'GTH pseudos' library where these are in different formats. I naively assumed I could simply compare both formats and then rewrite my GTH's to Abinit format,  but when comparing the same pseudo of the library in both formats, I see that the Abinit one contains a couple of parameters that are not present in the CP2K format, e.g. for O it would be:

O GTH-PBE-q6
2 4
0.24455430 2 -16.66721480 2.48731132
2
0.22095592 1 18.33745811
0.21133247 0

in CP2K format versus

Goedecker pseudopotential for O
8 6 070301 zatom,zion,pspdat
10 11 1 2 2001 0 pspcod,pspxc,lmax,lloc,mmax,r2well
0.24455430 2 -16.66721480 2.48731132 rloc nloc c1 c2
2 nnonloc
0.22095592 1 18.33745811 rs ns hs11
0.21133247 1 0.00000000 rp np hp11
             0.00755927 kp11

for Abinit. So, in this case the Abinit pseudo contains a parameter 'mmax' in the second line and another parameter 'kp11' in the last line, which are not present in the CP2K one. Can these be straightforwardly derived from any GTH pseudo in CP2K format to write directly the Abinit one?

Thanks in advance,
Daniel

fabia...@gmail.com

unread,
Feb 8, 2021, 2:42:19 PM2/8/21
to cp2k
Dear Daniel,

You can download the pseudo potentials in the Abinit format (among others) here: https://www.cp2k.org/static/potentials/

Cheers,
Fabian

Dan_M

unread,
Feb 8, 2021, 2:59:58 PM2/8/21
to cp2k
Dear Fabian,
yes, but the question is that the PP's I want to translate from CP2K to abinit are *not* any of those in the library, but instead some PPs that a coworker created for a functional that is not there in that list. The thing is then to know how to find those missing parameters that are present in the abinit format but not in the CP2K one.
I understand that the 'mmax'' parameter is always 2001 for all elements and functionals, but I don't know how to compute the k parameter/s that are present in the abinit files... (and the abinit documentation does not clarify this).
Thanks!
D.

Krack Matthias (PSI)

unread,
Feb 8, 2021, 5:11:49 PM2/8/21
to cp...@googlegroups.com

Hi Daniel

 

The GTH PP psp.par files created by the legacy stand-alone GTH PP generator code are converted to ABINIT, CP2K, and CPMD format using this code:

 

https://github.com/cp2k/cp2k-data/blob/master/potentials/Goedecker/tools/gth_pp_convert.f90

 

Matthias

--
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...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/00ff500e-2e1f-4ea1-b3a2-a723c7db998bn%40googlegroups.com.

Dan_M

unread,
Feb 9, 2021, 5:27:09 AM2/9/21
to cp2k
Thanks a lot Matthias!
This seems quite helpful, but I don't have the .par files corresponding to the pseudos I want to convert. I only have the pseudos in CP2K format. Is it possible to back-engineer the pseudos from the CP2K to abinit format?
Best
Daniel

Krack Matthias (PSI)

unread,
Feb 9, 2021, 6:35:58 AM2/9/21
to cp2k

Hi Daniel

 

As described in Hartwigsen et al., the GTH PP consists of a local and a non-local part. The GTH parameters were optimized based on relativistic atomic all-electron wave functions. The non-local part is composed of a weighted average potential and a difference potential which describes the spin-orbit coupling (see eqs. 16-19). The h(i,j) and k(i,j) parameters refer to the average and the difference non-local potential, respectively. CPMD and CP2K do not support spin-orbit coupling with GTH PPs, i.e. the difference potential is neglected and thus the k(i,j) parameters are dropped, which results in a scalar relativistic description. I do not know, if ABINIT supports spin-orbit coupling for GTH PPs. At least the ABINIT GTH PP parameter sets would allow for that. You may ignore the k(I,j) parameters, e.g. try to set them zero, if you are not interested in spin-orbit coupling effects anyway. It is not possible to back-engineer the k(i,j) parameters from a scalar relativistic GTH PP parameter set.

 

HTH

 

Matthias

 

Dan_M

unread,
Feb 9, 2021, 7:52:13 AM2/9/21
to cp2k
Oh I see, thanks a lot for the very detailed answer! Since I am not interested in spin-orbit, the current info in the GTHs should suffice. Thanks for clarifying everything!
Best
Daniel

Emiliano Isaías Alanis Manzano

unread,
Sep 20, 2022, 3:50:32 PMSep 20
to cp2k
Dear all, I am interested in performing some calculations with the GTH PP but with Nwchem or Orca. Nevertheless, I do not know how to export the GTH PP into the Nwchem Format. I tried direct comparison, but I have not succeeded. Any help would be highly appreciated.

Best
Emiliano

Krack Matthias (PSI)

unread,
Sep 21, 2022, 4:01:57 AMSep 21
to cp...@googlegroups.com

Hi Emiliano

 

As far as I know, neither Orca nor NWChem support GTH pseudopotentials, i.e. have them implemented.

 

Best

 

Matthias

Reply all
Reply to author
Forward
Message has been deleted
Message has been deleted
0 new messages