[CP2K-user] [CP2K:22258] [charged calculations on metal oxides]

[Miguel RP]

Hi everyone, 

I’m trying to run some charged CP2K calculations for an optimal-tuning test 
of a CAM-like range-separated hybrid functional on isolated TiO2 
nanoclusters.

The neutral CHARGE 0 calculations converge fine, but the charged systems 
(CHARGE +1  (-1), UKS .TRUE., MULTIPLICITY 2) are very unstable and do not 
converge. The SCF energy keeps jumping by several Hartree even after many 
iterations.

The setup is a 100% long-range CAM-like functional, using for example:

SCALE_COULOMB 0.19
SCALE_LONGRANGE 0.81
OMEGA 0.15

I also generated a neutral UKS singlet restart file (CHARGE 0, UKS .TRUE., 
MULTIPLICITY 1) and used it to restart the charged calculations, which 
solved the previous nspin mismatch issue. However, the SCF itself remains 
unstable.

Has anyone dealt with charged open-shell hybrid calculations in CP2K for 
this type of systems? Any suggestions for a more stable SCF protocol, or 
whether this may indicate that the extra electron/hole is not well 
localized/bound in this setup?

So far, I have tried / am trying:
- restarting charged UKS doublets from neutral UKS singlet .wfn
- diagonalization + Broyden mixing
- lowering ALPHA in the mixing

Thanks!

-- 
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/4bf406f3-2bb4-4e73-845f-6c8e83c73aa6n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20260528/e049f50e/attachment.htm>