Closed richardjgowers closed 1 year ago
@mikemhenry : I'm going to have to defer to you here, since I don't understand how the new OpenFF unit system works.
Looking into this, I think that unless we have a compelling reason to support both the old and new toolkit, I'd rather just update to support only the new one for perses.
This should be solved with #1128 . Closing.
The atom mapping scoring seems to have been broken by off tk 0.11, see here:
https://github.com/choderalab/perses/blob/main/perses/rjmc/atom_mapping.py#L1142
gives:
I suspect it's that the line
old_positions = atom_mapping.old_mol.conformers[0] / unit.angstroms
doesn't remove units any more (unit
is openmm units, whereas now the conformer arrays are openff units and these units don't cancel), so you end up with a weird "double united" Quantity array. I suspect you'll want to use @mikemhenry newensure
function or similar to cancel out units