appropriate LMAXMIX value for DFT+U calculations

[sophie_weber]

Hello,

I am investigating the magnetic properties of a slab with an insulating, antiferromagnetic transition-metal oxide, Cr2O3, interfaced with the heavy metal Pt. I am using LDA+U calculations within the Dudarev method using a U of 4 eV on the Cr d states. I have been setting LMAXMIX to 4 with the rational that I am using a U on just d states, and the VASP wiki advises to set LMAXMIX to 4 for systems with d states in DFT+U. However, I am struggling with convergence, and am wondering whether, even though I do not have a finite U value for the Pt, I should still set LMAXMIX to 6 because the Cr2O3-Pt heterostructure contains f electrons. The wording of the wiki is a bit unclear.

If someone can clarify this I would greatly appreciate it. Thank you in advance!

[pedro_melo]

Dear sophie_weber,

Would it be possible to have a look at your input files? This could help figuring out the issues with your calculation.

Kind regards,
Pedro Melo

[sophie_weber]

Hello,

I am so sorry, I haven't checked the forum in a long time and didn't notice that you responded. Attached are example input files. There are convergence issues with and without implementing constrained magnetism.

Thank you in advance for your help,
Sophie

[pedro_melo]

Dear Sophie,

My apologies, but I was not able to run your calculation. Do you happen to have a smaller system (with fewer atoms) where this problem shows up?

Kind regards,
Pedro