Michael F. Herbst

Getting a basic version working should not be too hard, I agree. However, to get it up to speed and get it to fully we still need a bit of...

What you're thinking is a struct giving like a (partial) representation of the density matrix? That sounds reasonable to me.

- Cartesian coordinates and fractional - Read some standard input file formats (by relying on ASE / pymatgen)

See also the comments about special-casing 1D: https://github.com/JuliaMolSim/DFTK.jl/blob/a96f551ed78a47f6cbd5280fedf08f6a7443d31e/src/postprocess/band_structure.jl#L19-L25

#555 implements this using Brillouin and a user-specified space group (which for 2D lattices spglib cannot yet determine automatically)

I'll try to get some numbers for the current version and then we can see. For now we do in fact suck for cases like Caffeine where there are few...

Yeah. Many factors to consider in fact.

Oh wow, that looks pretty nice

Interesting could also be https://github.com/eth-cscs/SpFFT. Moving there might also be overcomplicating the code due to the tricks they need to get a good parallelisation. Perhaps a non-variational ansatz is better...

One thing that also came up in other projects was that setting the number of electrons explicitly instead of just saying something like "one extra electron" is a bit unhandy....