Limitations of the current neighborlists

[bodduv]

Currently the neighbor lists won't pick up isolated molecules, therefore, their intra-molecular contributions are not considered. For example, a system of just one molecule.

No immediate/apparent solution comes to my mind.

[davydden]

Indeed, but is not a deal-breaker for us. I would add an assert that we have at least two

[bodduv]

[WIP] comment The above limitation is addressed by:

• Adding [(cell_I, cell_I), (cell_molecule_I, cell_molecule_I)] entries in order to account for intra-molecular contributions for isolated molecules. Addressed in ___.

A related following issues also poped up recently.

• Currently, intra-molecular interactions are computed multiple times. This is a bug in compute_local() qc.cc file. This is caused because currently molecule_I's intra-molecular interactions are computed whenever molecule_I is mentioned in neighbor_lists. This is fixed in ___.

• Currently, the interaction between atoms are declared through their atom_type. However, we would want a more flexible way to define interactions, for example, we might want to define a bond between two atom types (a core and a shell of an atom) within a molecule but the same interaction shouldnot be declared between a core of one atom and a shell of another atom. This is fixed in ___.