OFF-EP 0008 — Periodicity-dependent vdW methods and "no-cutoff" option

[mattwthompson]

Resolves #51 #7

[jchodera]

Looks great, thanks!

As a reminder, OpenMM's CutoffPeriodic still uses a modified reaction field method unless the external solvent dielectric is set to 1, so implementations might need to be aware of this. OpenMM also requires cutoffs to be equal between LJ and electrostatics, but this is not something we should be worried about at the spec level.

[davidlmobley]

This looks good to me as well. Thank you! If Jeff approves we could actually get this resolved without meeting to discuss this one.

[lilyminium]

Thanks for putting this together @mattwthompson, I think this'll be an important improvement to the spec!

[j-wags]

Could this be updated to explicitly say that PME is an allowed periodic_method value? People are already talking about a LJPME FF refit so it'd be good to have that explicitly stated so we can implement support.

[mattwthompson]

I'd prefer that be a separate proposal; I haven't thought through exactly what it would mean to support LJPME (#11 #50 #15 #18 openforcefield/openff-toolkit#989) and I worry doing it wrong the first time would be worse than not doing it at all. I think it warrants more care than tacking it on the end of this proposal.

[j-wags]

Makes sense. Thanks for pushing this forward, LGTM!

[mattwthompson]

Thanks all for the prompt feedback! IIUC these are four approvals from the four committee members, and approvals on GitHub bypass the need for a meeting per se.