More dimensions

This is a follow-up to #14, but more speculatory.

At the moment, I'm working with atoms with spherical symmetry, which allows me to reduce my differential equations to one-dimensional radial problems, with the angular coordinates treated exactly using angular momentum algebra. In the future, I will probably want/need to go to molecules which possess less/other symmetries. There are many different ways of treating molecules, but the dominating by far is linear combinations of (multi-centred) 3d Gaussians (actually polynomials multiplied by Gaussians). It would be interesting to see if we could fit these type of basis functions within the ContinuumArrays framework. If we get this to work, a Hartree–Fock code could conceivably be used for both atoms and molecules without modifications (some tensor magic would need to happen).

I am by no means well-versed on the topic, and would need to read up quite a bit (although I have a lot of good references). @jarvist could possibly be interested in this as well.

Start by reading issue #14 and the existing ContinuumArrays framework, then review the linked Gaussian orbital references and the discussion here. The goal is to determine whether multicentred 3D Gaussian basis functions can fit the framework well enough for Hartree–Fock code to support atoms and molecules without modification; no implementation files or tests are identified.