Just making sure that there is currently no way to read forces from a table that are a function of position and not of just distance between particles. These forces would be obtained from a 3D PMF calculations of a molecule interacting with heterogeneous channel surfaces.
No, there is not, and such a table would have to be rather large,
unless you specify some spatial symmetries to reduce dimensionality!
It would be simple enough to write a fix to implement this. And you
should look into aton variables and the addforce fix
(http://lammps.sandia.gov/doc/fix_addforce.html). It's arguments can
be atom style variables that let you compute the forces dynamically
(as a function of position for example).