This PR is competing with #10 as a proposal for a coefficient container type for QuArrays.
There are few key advantages:
Conjugation/transposition is performed eagerly. This is closer to standard Julia Array behavior than implementing a lazy transpose, and should mean less work for us moving forward, since more functions will be able to be simply delegated to the arr field without having to worry about moving around indices.
Now that everything is eagerly performed anyway, we have no need for disambiguating a "dual" vector from the conjugate transposed vector. This is much cleaner than having a dual function and a ctranspose function, IMO.
Note that transposition for an N=1 instance of QuCoeffs is simply a copy with the transpose flag flipped. This allows us to represent row vectors as single dimensional.
Let me know what you think. Personally, I greatly prefer this approach to that of #10.
This PR is competing with #10 as a proposal for a coefficient container type for QuArrays.
There are few key advantages:
arr
field without having to worry about moving around indices.dual
function and actranspose
function, IMO.Let me know what you think. Personally, I greatly prefer this approach to that of #10.