PAW pseudopotentials choice and Bi_d

I have some questions about PAW potential when I’m going to use MP database.
first, I want to ensure which version of PAW-POT is used, PAW-2003 or PAW-2012?
second, when I reading the Calculations Guide of Documentation, I found the choice of Bi was still a question mark. Can you give us a definite option now?
Thanks and Forgive my poor English.

For the current data on MP, PAW-2003 is used. The PSP for Bi is the standard Bi one.

sorry to bother you again!
I wanna ensure the pseudopotentials used in current MP database. The documents indicates that the 2012 versions of the VASP pseudopotentials was used in The Materials Project.

I did a test on NaCl( mp-22862), the calculated total energies are -3.3881 ev/atom (2003 VASP pseudopotentials) and -3.3891 eV/atom (2012 VASP pseudopotentials), respectively. The result indicate that the calculated total energy with 2003 VASP pseudopotentials is more consistent with MP (-3.3881 eV/atom).
So, :dizzy_face: :dizzy_face:, which one should I use?

Yes, it’s the older pseudopotentials, or more specifically the pseudopotentials distributed with VASP that are not the _52 or _54 versions. The pymatgen documentation is in error and has been corrected.

New versions of pymatgen actually check the hashes of your pseudopotentials when using the VASP input sets to ensure that you’re using the correct version, to prevent accidental use of incompatible pseudopotentials. If you follow the pymatgen installation instructions, you should be using the correct pseudopotentials by default.