[CP2K-user] [CP2K:22275] Spin polarized calculation with xTB
Jürg Hutter
hutter at chem.uzh.ch
Tue Jun 9 09:21:45 UTC 2026
Hi
1) The implemented xTB Hamiltonians don't have a spin dependence.
You will always get the same orbital energies for alpha and beta electrons.
2) Occupation of the orbital levels depends on a combination of input parameters.
a) No smearing: Number of alpha and beta electrons is determined from the Multiplicity keyword.
b) With smearing: default is a single Fermi energy and electrons are allowed to move from alpha
to beta. This is default because this is what most people want for DFT Mermin functional calculations.
If you want to keep the number of electrons to be fixed, you need to specify
FIXED_MAGNETIC_MOMENT 2
This together with a FD temperature of 300 is the default in tblite (if I'm correct).
regards
JH
________________________________________
From: cp2k at googlegroups.com <cp2k at googlegroups.com> on behalf of Dmitry Ryndyk <dmitry.ryndyk at tu-dresden.de>
Sent: Tuesday, June 9, 2026 7:23 AM
To: cp2k
Subject: Re: [CP2K:22273] Spin polarized calculation with xTB
Dear Jürg, Michael, Thomas,
Thank you for the answers. I use the master git branch, and the UKS tblite simply doesn't work as UKS. I hope to get a reply from someone who cares about tblite integration into CP2K just now.
But meanwhile, I do not understand the internal spin-dependent now. Please have a look at the attachment. I tried
Multiplicity=3 and got a strange result. There is no spin difference at the end!
And it has nothing to do with xTB; the same is for DFT PBE.
Only with OT do I get the expected result.
What do I not understand?
Thank you in advance,
Dmitry
Thomas Kühne schrieb am Dienstag, 9. Juni 2026 um 00:30:36 UTC+2:
Hi everybody,
there have been some recent new developments. In conjunction with tblite 0.6.0
CP2K now supports GFN2 for periodic systems incl. k-points and via analytic
gradients and stress tensor for RKS and UKS. Have a look at the recent preprint,
as well as the recent regtests and https://github.com/DCM-Uni-Paderborn/Periodic-GFN2-Benchmarks
for representative sample inputs.
Greetings,
Thomas
<https://chemrxiv.org/doi/abs/10.26434/chemrxiv.15003939/v1>
chemrxiv.org<https://chemrxiv.org/doi/abs/10.26434/chemrxiv.15003939/v1>
[X]<https://chemrxiv.org/doi/abs/10.26434/chemrxiv.15003939/v1>
Am 08.06.2026 um 11:51 schrieb Michael LaCount <lacou... at gmail.com>:
I looked into this earlier this year.
Tblite does seem to support the spin polarized version (spGFNn) see https://onlinelibrary.wiley.com/doi/pdfdirect/10.1002/jcc.27185 "All (sp)GFNn-xTB calculations were performed with the tblite v. 0.3.0 development version either as a standalone or in the xtb v. 6.5.1 development version." A bit ambiguous but it seems to me to read like tblite supported spGFN-xTB.
But I'll note that I found using LSD with TBLITE GFN1 doesn't work properly (at least in v2026.1) (see https://github.com/cp2k/cp2k/issues/4818 and https://github.com/cp2k/cp2k/pull/4804). Also TBLITE with GFN2 has (or at least had) some odd behavior assigning electrons for nitrogen.
In short, at least as of CP2K 2026.1 there is no way (to my knowledge) to get a proper spGFNn calculation done with CP2K. The best you can do is triplet GFN0/1 using the internal CP2K implementation of XTB.
Best,
Michael
On Monday, June 8, 2026 at 7:12:08 AM UTC-7 Jürg Hutter wrote:
Hi
This extension is not included in the CP2K native gfn0/1 xTB Hamiltonians.
I don't know if it is available through tblite, but most likely it has to be triggered by a
keyword and defaults to the original spinless Hamiltonian.
Probably this keyword is not yet accessible through the CP2K input?
best regards
JH
________________________________________
From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of Dmitry Ryndyk <dmitry... at tu-dresden.de>
Sent: Monday, June 8, 2026 3:36 PM
To: cp2k
Subject: Re: [CP2K:22267] Spin polarized calculation with xTB
Dear all,
What is going on at present?
The xTB site<https://xtb-docs.readthedocs.io/en/latest/spgfn.html#spgfnn-xtb> states that spin-polarized calculations with tblite are possible, and spin-dependent Hamiltonians are described in the 2023 paper<https://onlinelibrary.wiley.com/doi/10.1002/jcc.27185>.
Moreover, at the CP2K site<https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/QS/XTB/TBLITE.html>, it is written:
Spin-polarized CP2K/tblite calculations are selected with the regular DFT/UKS keyword or its LSD/SPIN_POLARIZED aliases; DFT/MULTIPLICITY sets the number of unpaired electrons. UKS uses tblite’s spin-polarized Hamiltonian.
At the moment, however, it seems to be blocked. One always gets restricted orbitals.
Any ideas?
Best wishes,
Dmitry
hut... at chem.uzh.ch schrieb am Freitag, 5. März 2021 um 19:09:23 UTC+1:
Hi
you can calculate a triplet using UKS. For xTB the results should
show equal orbital energies for alpha and beta spin, even though
the occupations will be different.
regards
Juerg
--------------------------------------------------------------
Juerg Hutter Phone : ++41 44 635 4491<tel:+41%2044%20635%2044%2091><tel:+41%2044%20635%2044%2091>
Institut für Chemie C FAX : ++41 44 635 6838<tel:+41%2044%20635%2068%2038><tel:+41%2044%20635%2068%2038>
Universität Zürich E-mail: hut... at chem.uzh.ch
Winterthurerstrasse 190
CH-8057 Zürich, Switzerland
---------------------------------------------------------------
-----cp... at googlegroups.com wrote: -----
To: cp... at googlegroups.com
From: "Pierre-André Cazade"
Sent by: cp... at googlegroups.com
Date: 03/05/2021 05:16PM
Subject: Re: [CP2K:14885] Spin polarized calculation with xTB
Dear Juerg,
It's a very interesting topic. Provided what you described in your reply, is it possible to use xTB for a triplet state? If yes, is it necessary to use UKS, since the Hamiltonian doesn't have spin-dependence?
Regards,
Pierre
On Fri, Mar 5, 2021 at 3:42 PM <hut... at chem.uzh.ch> wrote:
Hi
you can use unrestricted calculations also with xTB. However,
the xTB Hamiltonian doesn't have a spin dependence. So what you
get is the same as using a restricted open shell calculation.
Unrestricted is double the work, but can use OT methods.
Restricted open shell (ROKS will not work btw) will need
diagonalization and smearing. What is more efficient in the end
might depend on your system.
best regards
Juerg Hutter
--------------------------------------------------------------
Juerg Hutter Phone : ++41 44 635 4491<tel:+41%2044%20635%2044%2091><tel:+41%2044%20635%2044%2091>
Institut für Chemie C FAX : ++41 44 635 6838<tel:+41%2044%20635%2068%2038><tel:+41%2044%20635%2068%2038>
Universität Zürich E-mail: hut... at chem.uzh.ch
Winterthurerstrasse 190
CH-8057 Zürich, Switzerland
---------------------------------------------------------------
-----cp... at googlegroups.com wrote: -----
To: "cp2k" <cp... at googlegroups.com>
From: "Mauro Sgroi"
Sent by: cp... at googlegroups.com
Date: 03/05/2021 09:33AM
Subject: [CP2K:14881] Spin polarized calculation with xTB
Dear all,
I'm trying to understand if it is possible to use xTB to perform spin polarized calculations.
If yes, which is the right way to proceed?
Is there an howto to follow?
Thannks a lot and best regards,
Mauro Sgroi
Centro Ricerche FIAT
Italy
--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/f6049f47-2a0d-4d21-bdd7-8a1fa320dfb3n%40googlegroups.com.
--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/OFDF4C8D2B.FF7BE2D3-ONC125868F.0056498D-C125868F.0056498E%40lotus.uzh.ch.
--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/CAFnf9gqvD3jnN_wRAgbYZGaN%2BC1%2BGANJHs8-Nqf-ruYMGjKHHQ%40mail.gmail.com.
--
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 view this discussion visit https://groups.google.com/d/msgid/cp2k/05c0af2b-f6cb-4bdb-817c-521bcf5acfb9n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/05c0af2b-f6cb-4bdb-817c-521bcf5acfb9n%40googlegroups.com?utm_medium=email&utm_source=footer>.
--
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.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/25710ec6-ea83-4356-80e3-69a6dfae4d27n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/25710ec6-ea83-4356-80e3-69a6dfae4d27n%40googlegroups.com?utm_medium=email&utm_source=footer>.
--
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+unsubscribe at googlegroups.com<mailto:cp2k+unsubscribe at googlegroups.com>.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/bb989d03-22dd-4a73-9264-1a24d83d98a3n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/bb989d03-22dd-4a73-9264-1a24d83d98a3n%40googlegroups.com?utm_medium=email&utm_source=footer>.
--
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+unsubscribe at googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/ZR0P278MB075977FC5B751174D323274C9F1D2%40ZR0P278MB0759.CHEP278.PROD.OUTLOOK.COM.
More information about the CP2K-user
mailing list