BlochMatrix type with internalized matrix about quantica.jl HOT 1 CLOSED

How would this handle a situation in which you use several threads to operate on different harmonics? You would need to have one BlochMatrix per thread, which would have non-aliased matrix fields, even if they share the same Hamiltonian, which would be read-only. To avoid computing matrix (merge of Harmonics) once per thread, we should implement Base.copy(b::BlochMatrix) = BlochMatrix(copy(b.matrix), b.h)

Also, I would rename BlochMatrix to Bloch, since it is created with bloch

If we implement #180, we would also have a simplification of the proposal, without the ugly bloch(h, flatten) method. Since flatten would be lazy, we would do

m = lattice |> hamiltonian(model) |> parametric(modifiers) |> flatten |> bloch

from quantica.jl.