SparsePauliOp.apply_layout should do nothing if given None

[nonhermitian]

What should we add?

Simulators give None for the layout when a circuit goes through the transpiler. Currently this raises in SparsePauliOp.apply_layout. Rather, it would be nice if None is a null op in apply_layout so that I can run the same code for a sim and a real backend.

[SoranaAurelia]

From my understanding, the only change will be in the apply_layout function by adding the case for a null layout (in which case the function should do nothing). If this is good for a first issue I would like to help with this one :)

[nonhermitian]

It is a good first issue, at least in my mind.

[SoranaAurelia]

Great! Then I would like to work on it :)

[mtreinish]

I've assigned this issue to you @SoranaAurelia

[rishabhRsinghvi]

@mtreinish I think it's solved is it?

[jakelishman]

What behaviour would we expect here if num_qubits is given and is not equal to the number of qubits in the initial operator? My first thought is that that must be an error, but I'm not certain.

[SoranaAurelia]

Maybe we should treat it the same way we treat the case in which the layout is specified and the num_qubits is larger then the num qubits from the layout provided? That is, expanding the operator to the requested number of qubits. Raising an error might also solve the problem, but I think it can be in some cases contradictory to the intention of this feature request: the same code that works for a transpiled circuit for a certain backend might pass while the one transpiled for a simulator will raise due to this. Thanks for catching this!

[nonhermitian]

I would also expand the operator.

[jakelishman]

Personally I'm not a fan of implicitly assuming that the given qubits are the "low" indexed ones - that's the kind of thing a layout is supposed to tell you - but looking at the code, it seems like we've already made that assumption, so it's most consistent to extend it to this case as well. Thanks both.