Page 1 of 1
SCF divergence when enabling SOC
Posted: Mon Mar 23, 2026 12:32 pm
by alok_shukla1
Dear Experts,
In my 2D system, I am facing a divergence issue in the SCF run with SOC. Relaxation, SCF, and band structure calculations are working fine without SOC. In the relaxed structure (obtained without SOC), when I switch on SOC for SCF run, the calculation starts diverging. I am willing to calculate magnetocrystalline anisotropy energy (MAE), but at this stage, the calculation itself is not converging. The relevant files are attached. Could you please suggest if there is anything wrong in my input files?
https://drive.google.com/file/d/1eganxa ... sp=sharing
Re: SCF divergence when enabling SOC
Posted: Mon Mar 23, 2026 12:43 pm
by henrique_miranda
Could you share the OSZICAR too?
Can you try running the calculation without LDA+U and VDW first and see if it converges?
If it does can you run with LDA+U first and see if it still converges?
If it does than you know that the problem is the VDW corrections.
Does it work if you try the LDA+U and VDW calculation starting from a converged WAVECAR with PBE?
Can you also try using reciprocal space projectors (i.e. without LREAL=A)?
Re: SCF divergence when enabling SOC
Posted: Mon Mar 23, 2026 4:10 pm
by alok_shukla1
Dear Expert,
Thank you so much for your response.
I have rechecked my shared files, and I found that the OSZICAR file is also there (in the shared folder).
I'll try all the suggestions and update here if I get the converged results.