mapping_fxn has no access to HOOMD box

[RainierBarrett]

In tensorflowcompute.py, the mapping_fxn argument is presently unable to access information about the simulation box, only particle positions. In order to properly treat periodic boundary conditions, the box dimensions are needed.

[RainierBarrett]

This could be addressed in a couple of ways. Easiest is probably to change the documentation and call to mapping_fxn to expect it to take in the simulation box as well as positions.

[whitead]

Is the box not present in a snapshot?

[RainierBarrett]

Sure it is, but I'm talking about the mapping_fxn call expected signature, not the signature of enable_mapped_nlist. Inside of enable_mapped_nlist we say the expected signature of the mapping function is only

:param mapping_fxn: a function whose signature is ``f(positions)`` where positions is an
                            ``Nx4`` array of fine-grained positions and
                            whose return value is an ``Mx4`` array
                            of coarse-grained positions.
:type mapping_fxn: python callable

so when we call it within enable_mapped_nlist we don't presently feed in box information:

g_pos = mapping_fxn(
            snap.particles.position.astype(self.model.dtype))

I should be able to open a PR with a small change either today or tomorrow on this.

[whitead]

Got it, thanks

[whitead]

Closed by #337