I spotted significant differences between the two database versions that can be accessed through materialsproject.org (and the API available through pymatgen) and next-gen.materialsproject.org. In this example of YMnO3, the space group of the lowest-energy polymorph changed (see pictures below). This gives a problem for results that are based on data that has previously been obtained using the API available in pymatgen, which seems to be the legacy API.
Is there any register of which data has been updated or the database version differences?
Hi @springerm, I frequently work with the YMnO3 structures and noticed this issue once as well. I think one of the particular database releases (2020_09_08, or whatever you can access through the legacy API) had a problem with building entries for certain materials.
mp-19385 should ideally be the ground-state structure. The next-gen website looks correct to me. I would recommend only using the new API (and newest database release) if possible – a lot of these issues should be fixed.
If you want to confirm what database you’re using you can always check with one of the MPRester methods:
from mp_api import MPRester
with MPRester() as mpr:
Thank you very much for your post. There is not such list with problematic entries, is it?
I don’t think so, unfortunately