Andrea Rizzi

> From this forum post I was lead to believe that even if a particle-particle interaction is excluded, the reciprocal space interaction is still present as long as the charges...

> turn on a CustomNonbonded force that defines regular Coulomb interactions but applies an interaction group for just alchemical-alchemical This is something that we can implement in `AbsoluteAlchemicalFactory` to enable...

I've changed the title of this to remind us to implement decoupled electrostatics with exact PME and direct-space PME preserving the reciprocal space during the alchemical transformation.

> Does MCMCSampler need to make a deep copy? I don't think that's necessary for now. It's just a precautionary thing in case the `thermodynamic_state` becomes private in the future....

> What do you think our approach should be here? The `Timer` object is essentially a dict that compute elapsed times automatically, so I think one way can be converted...

> I'll just do the minimal amount of refactoring Ok. I think it's all internal anyway, so we can easily postpone it to later modifications if you have higher priority...

Hmm. That'd be a good feature to have, but I'd avoid making changes to the API of `MCMCMoves` just to accomodate debugging features if possible. What about just use the...

Otherwise we can have a centralized and more sophisticated timing utility, but I'd still try to keep it outside the API of the single objects.

Looking at the error, I believe that the problem is that `CompoundIntegrator.getIntegrator()` does not preserve the Python class. I think Peter fixed it for `Context.getIntegrator()`, but apparently those changes did...

Or maybe even better, removing `_measure_heat` completely and just implement `get/set_heat` (and other variables related to shadow work etc.) with something like ```python def get_heat(self): try: return self.getGlobalVariableByName('heat') except: raise...