[CP2K-user] [CP2K:21275] Re: Energy Instabilities in SCF Convergence of High Impact Energy Collisions

[Quentin Pessemesse]

Hi Nick, 
I cannot really provide input on the QM/MM part as I have no expertise in 
this. For your SCF convergence issues, you might want to look into 
disabling safe DIIS, which uses steepest descent if the DIIS algorithm does 
not decrease the energy 
(https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/SCF/OT.html#CP2K_INPUT.FORCE_EVAL.DFT.SCF.OT.SAFE_DIIS). 
I've found it sometimes causes issues.
On a more general note, for light atoms like those you are using, you might 
get similar results and better convergence using the Gaussian-Plane-Wave 
method (GPW) instead of GAPW, and using the GTH effective core potentials 
optimized for PBE0 and associated TZVP or TZV2P MOLOPT basis sets (which 
are available here 
https://github.com/juerghutter/GTH/blob/master/PBE0/POTENTIAL for the ECPs 
and here for the basis sets 
https://github.com/juerghutter/BASIS/blob/master/MOLOPT/PBE0/MOLOPT_PBE0). 
This would dramatically reduce the number of electrons.
Take care, 
Q.
Le lundi 10 mars 2025 à 03:31:39 UTC+1, Nicholas Laws a écrit :

> Hi Quentin,
>
>
> Thank you for your thoughtful suggestions. I have implemented your 
> recommendations, all of which have certainly improved the computational 
> efficiency. However, I am still encountering persistent SCF convergence 
> failures at the moment of collision with the external potential, regardless 
> of whether diffuse functions are included in the basis set (I tested 
> aug-cc-pVTZ, cc-pVTZ, def2-TZVPPD, and def2-TZVPP).
>
>
> To circumvent the challenges posed by the external harmonic potential, I 
> created a QM/MM workflow that models the Au surface classically. 
> Unfortunately, similar SCF convergence issues arise when the EMI-BF4 
> neutral collides with the Au surface in the QM/MM framework. I was curious 
> if there are any gaps in my current QM/MM framework that may cause this 
> lack of SCF convergence? Additionally, I was curious if you or the 
> community could recommend accurate sources or standard references for 
> Lennard-Jones (LJ) pair potentials between the relevant QM atoms (N, C, H, 
> B, and F) and Au atoms? Currently, I am using Lorentz-Berthelot combing 
> rules from the Efficient 'universal' shifted Lennard-Jones model for all 
> KIM API supported species developed by Elliott and Akerson (2015) v003 
> <https://openkim.org/id/MO_959249795837_003>.
>
>
> I have attached the relevant input file for your review in .zip format. 
> Thank you again for your valuable advice, and I look forward to any 
> insights you or others in the CP2K community might have.
>
>
> All my best,
>
> Nick
>

-- 
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/dacf6240-89f5-45dc-a405-6321f4bcf3f3n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20250310/4d98ff06/attachment.htm>