David L. Mobley

All for pinning versions; having the tests frequently start failing because of a change in somethign we don't control is frustrating.

> We should also generalize NonbondedForce to permit other forms (e.g. Buckingham exp-6, Halgren buffered 14-7, etc.). Yes. I didn't mention that here, as `HarmonicAngleForce` and `HarmonicBondForce` are tags in...

On versioning - I commented on #42.

That's a good point. But it should also be possible to look at things in a slightly more fine-grained way I would imagine, at least in this test - i.e....

#78 implements tracking of score versus time. It's worth noting that there is no guarantee (as in MD simulations) that the final score corresponds to the best score, and in...

@bannanc - I don't actually think this is a bug, more of an issue of "confusing output". Specifically, imagine numbering your carbons 1-2-3-4. You have three carbon-carbon bonds, sure, except...

(Update/note to self: This applies also to the main `ForceField` class which now replaces forcefield_labeler, so I should clarify the info printed there.)

See also #167

Regarding the formal charge, we can use `OEAssignFormalCharges` ( http://docs.eyesopen.com/toolkits/python/oechemtk/OEChemFunctions/OEAssignFormalCharges.html ) with this additional info: > "Sets the formal charge property of an atom, based upon a simplistic valence model....

So, an interesting question is whether we would allow fractional bond orders only for SMIRKS matching `~` bond types or whether we would allow them for any. Presumably it would...