Should we enforce Unitful coordinates?

[Gregstrq]

This is the place to discuss whether we want to enforce specific types for the coordinates and, by extension, for the velocities.

Right now, the proposed way is to assume that the coordinates must have a unitful type with dimension of the length, while the velocities must have a unitful type with the corresponding dimension of the velocity. As a consequence, all of the realizations of the abstract interface need to enforce this assumption.

What is the opinion of the community members about this restriction?

My take on it is that this is an unnecessary restriction. Although I agree that the majority of use cases pertain to the particles in real space, I can imagine doing dynamical simulations with the particles living in some abstract space. For example, in momentum space. As a result, I propose to relax the restriction and allow the coordinates and velocities to have arbitrary types. Instead, I propose that we should enforce the "homogeneity" of the used types: one should not be able to mix the unitless and unitful coordinates while describing the same system (one should not be able to use unitful quantities with different dimensions as well).