UZH vs non-UZH basis sets, SCF convergence

[Martin Konôpka]

Dear CPMD experts,

Recently I started to use the UZH-labeled basis sets and pseudopotentials in some of my calculations. In case of a single oxygen atom I found very difficult to achieve SCF convergence compared to the non-UZH basis set and PP. I guess that the basis set is the source of my difficulties rather then the PP. I was doing the singlet-state calculation (which is not the ground state but that should not matter). I attach my input and output files (see especially cp2k.out_UZH) and would be grateful for any advice or notification if I am doing something wrong.  Are there some general recommendations regarding the use of the UZH-labeled basis sets and PPs?

Best regards

Martin Konôpka.

[Martin Konôpka]

PS Dear CP2K experts, sorry for the inappropriately addressing you. I am still using also the CPMD package so I did the mistake.

Dátum: streda 8. januára 2025, čas: 13:25:24 UTC+1, odosielateľ: Martin Konôpka

[Thomas Kühne]

Dear Martin Konôpka, 

please set EPS_DEFAULT to at least 1.0E-12 to reach a SCF threshold of 1.0E-7. 

Also, reduce the STEPSIZE parameter within the OT section to e.g. 0.1 when using 

OT in combination with the DIIS minimizer. A good diagnostics is that whenever 

DIIS switches to SD action is needed. Either reduce STEPSIZE, otherwise switch 

from DIIS to CG, followed by changing the preconditioned to FULL_ALL and if even 

that doesn’t work switch to TD instead of OT. 

Greetings, 

Thomas Kühne

--
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 visit https://groups.google.com/d/msgid/cp2k/54eb792a-bb03-42d4-8270-da6d0a696df7n%40googlegroups.com.
<cp2k.inp_UZH><cp2k.out_UZH><cp2k.out_non-UZH><cp2k.inp_non-UZH>