[WIP] Handlers for DampedExp6810, AmoebaMultipoleForce, AxilrodTeller potentials

[aehogan]

Description

This is staging PR for in progress work on three new handlers, DampedExp6810 from https://doi.org/10.1021/acs.jctc.0c00837, Multipole from AmoebaMultipoleForce, and AxilrodTeller 3 body potentials.

Todos

• Create handlers in handlers/nonbonded.py
• Create collections in collections/nonbonded.py
• Add tests in tests/handlers
• Final checks

Questions

• Can we get charges from AM1BCCHandler in a more elegant way?

Status

• Ready to go

[codecov-commenter]

Codecov Report

Patch coverage: 91.16% and project coverage change: +8.54% 🎉

Comparison is base (3df2d30) 75.20% compared to head (9d16428) 83.74%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #56      +/-   ##
==========================================
+ Coverage   75.20%   83.74%   +8.54%     
==========================================
  Files           4        4              
  Lines         242      523     +281     
==========================================
+ Hits          182      438     +256     
- Misses         60       85      +25     

Flag	Coverage Δ
unittests	83.74% <91.16%> (+8.54%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files Changed	Coverage Δ
smirnoff_plugins/collections/nonbonded.py	91.36% <89.79%> (-4.33%)	⬇️
smirnoff_plugins/handlers/nonbonded.py	92.40% <100.00%> (+7.03%)	⬆️

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[mattwthompson]

ToolkitAM1BCCHandler is in a strange place now that its main purpose (assigning partial charges) no longer lives in the toolkit. Ultimately the best way is just to access Molecule.assign_partial_charges. Make sure you use "am1bccelf10" if it's available, following the SMIRNOFF spec. You can see my implementation here (a bit obfuscated, but you can get the general idea): https://github.com/openforcefield/openff-interchange/blob/v0.3.0/openff/interchange/smirnoff/_nonbonded.py#L531-L549

[mattwthompson]

@aehogan let me know when you'd like a review of this

[aehogan]

@mattwthompson if you could look at how I deal with unique molecules / force.addParticle that would be great, I would like to use virtual charge sites in the future.

smirnoff-plugins/smirnoff_plugins/collections/nonbonded.py

Line 611 in 7dea090

for _ in range(topology.n_atoms):

smirnoff-plugins/smirnoff_plugins/collections/nonbonded.py

Line 643 in 7dea090

for unique_mol_index, mol_map in topology.identical_molecule_groups.items():

[mattwthompson]

@aehogan I think everything from my first pass is resolved. Are there any other specific issues you could use feedback on, and is there more work you plan to do here? No rush from my end, but when it's ready to go I can give the code another light pass. In that case I'd trust that these plugins produce the results you expect it to produce, and I'd mostly be looking for things like reliability and potentially problematic corner cases.

[aehogan]

@mattwthompson I have one more thing I want to add, checking for a custom bonded force that has a coloumb's law expression and zero'ing out those charges as well. OpenFF interchange creates such a force to handle scaled 1-4 charge interactions and so it would be a safety feature.

Besides that I have checked a bunch of things such as high temperature, high pressure noble gas density isotherms with all combinations of potentials as well as running >1000 openff-evaluator simulations and haven't encountered any problems so far.

[aehogan]

@mattwthompson I'm good to merge this whenever you get a chance to look over it again, right now the buckingham water example simulation is crashing but my tests are working on my end.

[mattwthompson]

Great, I'll see if I can fix upstream

[mattwthompson]

@aehogan could you please

1. Merge upstream back in once more (:tm:)
2. Add the following to .git-blame-ignore-revs:

# Versioneer auto-generated
94a0ae93bc06fa2e1cef137eec3f7fdefc0f9c01

[mattwthompson]

I'm going to trust my past self, your report, and the automation (no warnings or non-typing errors in logs) here