- decide what to do about basis ordering
- compare performance to more type-stable version
- Add automated benchmarks (for learning purposes)?
- Think about automatic meanfield functions
- Register package?
- ╔ (Optimization) Matrix construction
- ║ Generalize matrix construction to directly construct dense matrices, build up sparse matrices
- ║ Include possibility to directly construct Matrix in symmetrized subspace -> SpinSymmetry.jl
- ║ Code of building: setindex!(arr,x,y,v)
- ║ For dense -> set directly, for sparse -> have a COO and just append to I,J,V vectors
- ║ For symmetry sector: put x,y through the projection matrix and apply factor to v, then setindex!
- ╚
Explain basis ordering! (after deciding)- MORE Usage + examples
(after implementing the geometry stuff) - benchmarks vs. kron?
Mention SpinSymmetry.jl