Inquiry about SCF Convergence Issues with the Diagonalization Method in CP2K

[yis...@163.com]

Dear CP2K Developers and Experienced Users,

Hello everyone,

I am a beginner with CP2K and am encountering an issue with Self-Consistent Field (SCF) convergence when using the diagonalization method, even after confirming the correct spin multiplicity for my system.

To identify the ground state, I first performed a spin multiplicity scan using the Orbital Transformation (OT) method. The results confirmed that a spin multiplicity of 17 yields the lowest energy for my system.

Subsequently, I switched to the diagonalization method with k-points to perform a geometry optimization. However, I found the SCF procedure to be very difficult to converge under this setup.

I am writing to seek your advice and have attached my input and output files for reference. Specifically, I would like to ask:

Are there any issues with my input file, and is there room for further optimization?

Why does the diagonalization method often struggle with convergence compared to the OT method?

Under what conditions can the convergence speed of the diagonalization method be comparable to that of the OT method?

Any insights, suggestions, or guidance you could provide would be greatly appreciated. Thank you for your time and help.

Best regards,

Yisi chi

[Que Zhu]

I reviewed your output file and found that Convergence has been oscillating.

yis...@163.com <yis...@163.com> 于2025年10月24日周五 21:12写道：

--
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/0d79829a-8a88-4363-adee-a7cb741d9e16n%40googlegroups.com.

[Johann Pototschnig]

You can try to increase the cutoffs:

    &MGRID
      CUTOFF  400
      REL_CUTOFF  55
    &END MGRID

I t can be  helpful to decrease

 ALPHA

or change the buffer

NBROYDEN