No hubbard +U correction for transition metals with anions other than O and F in MPRelaxSet

Materials Project
1

I noticed that in MPRelaxSet (and in many other VASP input settings), there is no hubbard +U values specified for transition metals in compounds with anions other than O and F. Is there physical justification for this setting?

2

There’s a good explanation in the original methodology paper for the Materials Project workflows. The short explanation is that the “optimal” Hubbard U is not the same across different arrangements of the same atoms - see Fig. 1 of this paper

Because of this, we prioritize chemical environments, oxides and fluorides, where including a Hubbard U with the PBE GGA makes meaningful corrections to delocalization errors