Regarding non empirical U parameter calculation in vasp

[subrata]

Dear All,

I was trying to reproduce the U parameter for NiO using PBE+U from non empirical - linear response theory method as done in PHYSICAL REVIEW B 96, 100101(R) (2017) (DOI: 10.1103/PhysRevB.96.100101. I followed the example given in vaspwiki : https://www.vasp.at/wiki/index.php/Calc ... for_LSDA+U and was able to reproduce the result. However the obtained U value from this example (5.58 eV) heavily mismatches with the reported U value of 3.35 eV in the aforementioned paper. though the article specifies 64-atom supercell, which is different from 32 atom supercell of example, I found in both cases the U value is 5.74 eV. I had also increased the KPOINTS from 1x1x1 to 4x4x4 and the result do not change and I am doing PBE calculation only. I have attached all my calculations , script file and a python script to generate the U value as done in example. Kindly help me understand the discrepancy.

Also do mention how this process should (if any) modified for Meta GGA rung of functionals.

Thanks

[merzuk.kaltak]

Dear Subrata,
Your calculated U value of approximately 5.74 eV appears reasonable based on your input parameters and methodology, including the increase in k-points to 4x4x4.
However, due to limited human resources and the nature of this platform, we are unable to provide specific research assistance or troubleshoot the complex discrepancies between your results and the values reported in the academic paper (PHYSICAL REVIEW B 96, 100101(R) (2017)) [1]. The discrepancy could stem from specific computational nuances, pseudopotential differences, or subtle variations in the exact method parameters used in the published research that are not explicitly detailed in the general VASP example.
For detailed clarification on their specific methodology and results, we suggest you contact the authors of the paper directly.

Regarding your question about adapting this process for meta-GGA functionals, the general linear-response theory approach remains similar in principle; you would perform the initial self-consistent field (SCF) calculation using the chosen meta-GGA functional instead of PBE. However, specific implementation details within VASP might require careful configuration of the INCAR tags relevant to the meta-GGA functional, e.g., METAGGA = SCAN, etc. .