Expose Numpy simulator as Backend

What is the expected enhancement?

Qiskit includes a statevector simulator that uses Numpy. This is very nice because it means users can simulate circuits without installing Aer. However, I think the interface could be improved.

Currently the simulator is called using the Statevector class:

from qiskit.quantum_info import Statevector
state = Statevector(circuit)

It would be useful to expand this functionality so it is used in the same way as other backends. That way it can be incorporated into code workflows that use backends.

[kevinsung]

@Cryoris explained to me offline that the BasicAer backend exists and is exactly what I am asking for. So I’ll close this issue, though I will say that the existence of BasicAer should probably be mentioned more prominently in the documentation.

[jakelishman]

Sorry about the delay re-opening, these issues had fallen off the front page of my notifications between them updating and me looking at my queue.

I’ve re-opened #7634, but left #7633 closed on the basis that “use quantum_info” is a duplicate of #7670, and the “support all matrices” is something that we can’t officially fulfill within the Backend spec - if BasicAer is converted to use quantum_info underneath, then BasicAer.run() will support all instructions with a matrix, but we can’t make BasicAer appear that way to things consuming the Backend API (more detail in #7633).

I’ll close this on the basis that #7670 is the direction we want to take to solve the underlying issue here, rather than making a new backend. If I’ve missed a further point in one of these issues, please let me know.