Open mattwthompson opened 5 years ago
I think the most reasonable approach would be to not have a clearly-defined element for non-physical things and instead just keep that information directly in the Site
. Which is to say the mass, since I don't think the other atomic information is relevant for a UA/CG bead.
For non-atomistic systems, the
ParmEd
/MDTraj
approach is to defined a custom "Element" that is an instance of someElement
class but does not correspond to anything physically - instead it just tracks a name and a mass (which is all we think we need to care about).My idea was to have a special Element-like class that inherits from our
Element
class but only includes the information we care about.Alex suggested that, because physically a UA/CG bead doesn't really include the physical concept of an element, maybe we shouldn't have them as elements? A
Site
can include the mass of the bead and other things can probably just be tied to thatSite
(or, in some situations, instead ofAtomType
). I like this approach - it makes sense and is consistent with the "be able to specify things but also fall back if needed" sort of design we use elsewhere.