Consultation on LMAXMIX=6

[Yicheng]

Dear VASP developers,

I am performing simulations on oxides containing d-elements (e.g., Fe3+) and f-elements (e.g., Dy3+) using PAW pseudopotentials (without Hubbard U correction). I have set LMAXMIX=6 for systems containing Dy3+, but I would like to confirm whether it is necessary to set LMAXMIX=6 when using the Dy_3 pseudopotential (i.e., 4f-in-core).

Additionally, I forgot to change LMAXMIX=6 to LMAXMIX=4 in simulations for systems containing d-elements without f-elements. I am wondering if using LMAXMIX=6 instead of 4 will affect the results for these systems. For the sake of energy comparison, should I redo these simulations with LMAXMIX=4?

Thank you for your time and I am looking forward to your reply.
Sincerely,
Yicheng

[henrique_miranda]

Dear Yicheng,

The LMAXMIX tag determines the max quantum number of the channels that are passed on to the density mixer.
In principle, this should not affect the final results of the SCF calculations if the electronic convergence was reached.
In practice, however, it can happen that the convergence of the total energy with SCF iteration is very slow, and the iteration is stopped too soon.
This is unusual, but it can happen. You can check if that is the case in your calculation.
If you want to be 100% sure, I would suggest running the calculations again with LMAXMIX=6.

If you don't have any f orbitals in the valence, then LMAXMIX=4 should be sufficient.

For the case where you used LMAXMIX=6 instead of LMAXMIX=4 that should not be a problem either.

Here you can find some additional information about this tag
https://www.vasp.at/wiki/index.php/LMAXMIX

[Yicheng]

Dear Dr. Miranda,

Thank you very much for your reply! I feel relieved about my simulations now.

Gratefully,
Yicheng