Update energy tests #11
Conversation
This reverts commit 8bd6f10.
# Conflicts: # deforcefields/tests/test_deforcefields.py # devtools/conda-envs/test_env.yaml
| ref_energies = [1005.0846252441406, 44.696786403656006, 10.453390896320343] | ||
|
|
||
| for found, ref in zip(found_energies, ref_energies): | ||
| assert found.m_as(kj_mol) == pytest.approx(ref, rel=1e-3) |
There was a problem hiding this comment.
Note this is a fairly loose tolerance; I was getting errors like these if I used the defaults. How exactly these energies have drifted slightly from their artisan hand-crafted values a couple of years back, I'm not sure
$ python -m pytest -v deforcefields/tests/test_deforcefields.py::test_energy_sites | grep "E "
E assert 1005.0861476907567 == 1005.0846252441406 ± 1.0e-03
E
E comparison failed
E Obtained: 1005.0861476907567
E Expected: 1005.0846252441406 ± 1.0e-03
E assert 44.69819545912324 == 44.696786403656006 ± 4.5e-05
E
E comparison failed
E Obtained: 44.69819545912324
E Expected: 44.696786403656006 ± 4.5e-05
E assert 10.454626091527247 == 10.453390896320343 ± 1.0e-05
E
E comparison failed
E Obtained: 10.454626091527247
E Expected: 10.453390896320343 ± 1.0e-05
There was a problem hiding this comment.
I think the original values were calculated using the toolkit pre-interchange but maybe its time to do them by hand to double check, for now though this is a great match!
There was a problem hiding this comment.
Fantastic work @mattwthompson, hopefully, this should be the end oV-site-site issues!
|
Hopefully! If everything's up to date, could we get a release (maybe 1.0.2?) to get this rolled out to the "new" infrastructure? I'm happy to do the feedstock PR, but I'm not going to tag a release on one of your projects 😅 |
This is a follow-on to #4 with just a couple of substantive changes
smirnoff-pluginsmakes incorrect assumptions about atom/particle ordering, so I just ported it over to here with the right order.