My understanding is that the Materials Project uses the older POT_PAW_GGA_PBE PAW potentials (from around 2000 - 2002). However looking through the files that have been uploaded to NOMAD, it appears that about 9% of the calculations use at least one PAW potential that is not in this set. The calculations I have found so far were run in September and October of 2019, and the results of these calculations appear to be currently mixed in the legacy database. Other than going through NOMAD, is there some way to use the legacy API to identify and filter out these calculations?
I looked into this a little more, and it appears that some of these calculations are included in the current (next-gen) database as well. For example, this is the NOMAD entry for the static calculation for mp-15788. The OUTCAR file says the following POTCARS were used:
POTCAR: PAW_PBE Li_sv 10Sep2004
POTCAR: PAW_PBE Y_sv 06Sep2000
POTCAR: PAW_PBE S 17Jan2003
The OUTCAR file gives a final energy of -24.04593188 eV, which matches the value for the energy from the legacy API exactly. The energy per atom is -6.01148297 eV, which matches the uncorrected energy per atom from the new API exactly.
I believe that to be consistent with the rest of the database, the Li POTCAR should have been:
POTCAR: PAW_PBE Li_sv 23Jan2001