MOCharges API Question?

[ryanmrichard]

Right now (or maybe it's still a WIP?) MOCharges returns a number of centers by number of MOs matrix. The centers used are the ones in from space of the MOs. This could conceivably not be what the caller wants because: e.g., the same center was added to the AO basis set multiple times, there are ghost atoms, and/or there are mid-bond functions. My proposal is:

• we introduce a class PointSet which is a container of Point objects (this would live in LibChemist)
• Refactor the Atom class to inherit from PointCharge (related to this issue)
• Ensure AOBasisSet and Molecule are implicitly convertible to PointSet
• Have MOCharges take a PointSet instance and a DerivedSpace instance

This allows the caller of MOCharges to specify the centers they care about.

Thoughts? Takers?

Edit: Also I propose the name MOPopulations since AFAIK this is simply counting the number of electrons at a point (the electronic charge would -2 times the matrix returned by this PT for restricted spatial orbitals).

[jwaldrop107]

The structure of the return tensor for these modules is the intended form, but the name MOCharges is probably not the most appropriate one. I think all of this makes sense. I can change the name on the property type whenever. I can also work on the other things here, but probably not at this moment unless it is particularly pressing.

[ryanmrichard]

It's not immediately pressing because I use hard-coded Mulliken populations for testing the sparse map formation.