What circuits the circuits method in RB experiments should return

[itoko]

What is the expected behavior?

What circuits should the circuits method in RB experiments return? Especially, should it return

• circuits with wrapped or unwrapped Clifford operations? (or don't care?)
• circuits with basis-transformed or not? (or don't care?)

Currently, the circuits method in StandardRB returns circuits with wrapped Clifford operations without basis transformation like this:

from qiskit_experiments.library.randomized_benchmarking import StandardRB
exp = StandardRB(qubits=(0,), lengths=[1, 2, 3], seed=12, backend=FakeManila())
print(exp.circuits()[1].draw())

         ░ ┌─────────────────┐ ░ ┌────────────────┐ ░ ┌──────────────────────────────────────────────────────┐ ░ ┌─┐
     q: ─░─┤ Clifford-1Q(14) ├─░─┤ Clifford-1Q(6) ├─░─┤ Clifford: Stabilizer = ['-Y'], Destabilizer = ['-X'] ├─░─┤M├
         ░ └─────────────────┘ ░ └────────────────┘ ░ └──────────────────────────────────────────────────────┘ ░ └╥┘
meas: 1/══════════════════════════════════════════════════════════════════════════════════════════════════════════╩═
                                                                                                                  0 

You can see the decomposed circuits with one more method call:

print(exp.circuits()[1].decompose("Clifford*").draw())

         ░ ┌──────┐┌───┐ ░ ┌───┐ ░ ┌───┐┌───┐┌───┐┌───┐ ░ ┌─┐
     q: ─░─┤ √Xdg ├┤ Y ├─░─┤ X ├─░─┤ Y ├┤ S ├┤ H ├┤ S ├─░─┤M├
         ░ └──────┘└───┘ ░ └───┘ ░ └───┘└───┘└───┘└───┘ ░ └╥┘
meas: 1/═══════════════════════════════════════════════════╩═
                                                           0 

FYI: You can see the transpiled circuits (basis-transformed and applied mapping to physical qubits) by

print(exp._transpiled_circuits()[1].draw(idle_wires=False))

global phase: 5π/4
          ░ ┌─────────┐┌─────────┐┌────┐┌─────────┐┌─────────┐┌─────────┐┌────┐┌────────┐┌────┐┌──────────┐ ░ ┌───┐ ░ »
q_0 -> 0 ─░─┤ Rz(π/2) ├┤ Rz(π/2) ├┤ √X ├┤ Rz(π/2) ├┤ Rz(π/2) ├┤ Rz(π/2) ├┤ √X ├┤ Rz(2π) ├┤ √X ├┤ Rz(7π/2) ├─░─┤ X ├─░─»
          ░ └─────────┘└─────────┘└────┘└─────────┘└─────────┘└─────────┘└────┘└────────┘└────┘└──────────┘ ░ └───┘ ░ »
 meas: 1/═════════════════════════════════════════════════════════════════════════════════════════════════════════════»
                                                                                                                      »
«         ┌─────────┐┌────┐┌────────┐┌────┐┌──────────┐┌─────────┐┌─────────┐┌────┐┌─────────┐┌─────────┐ ░ ┌─┐
«q_0 -> 0 ┤ Rz(π/2) ├┤ √X ├┤ Rz(2π) ├┤ √X ├┤ Rz(7π/2) ├┤ Rz(π/2) ├┤ Rz(π/2) ├┤ √X ├┤ Rz(π/2) ├┤ Rz(π/2) ├─░─┤M├
«         └─────────┘└────┘└────────┘└────┘└──────────┘└─────────┘└─────────┘└────┘└─────────┘└─────────┘ ░ └╥┘
« meas: 1/═══════════════════════════════════════════════════════════════════════════════════════════════════╩═
«                                                                                                            0 

I would like to hear from many users about what the circuits should returns.

1. Circuits with wrapped Cliffords without basis transformation (Current spec)
2. Circuits with wrapped Cliffords with basis transformation
3. Circuits with unwrapped Cliffords without basis transformation
4. Circuits with unwrapped Cliffords with basis transformation
5. Other?

Which one do you expect?

[itoko]

I prefer 2 or 4.

[itoko]

Settle down to (2) Circuits with wrapped Cliffords with basis transformation once at #982