A superconducting qubit left alone relaxes to |0⟩ by spontaneous emission — the excited-state population decays exponentially with time constant T₁:
p1(t) = p1(0) · e^(−t/T1)
Waiting for this passively to reach 99.9% ground-state fidelity takes about 7×T1 — tens to hundreds of microseconds, dead time the qubit can't be used for a new circuit.
Active feedback reset shortens this by closing a real quantum-classical control loop: a dispersive readout pulse measures the qubit via its coupled resonator, a classical FPGA decides the outcome, and — if the qubit read out as |1⟩ — fires a conditional π-pulse to flip it back down. The measurement itself is imperfect, so the round is modelled with a symmetric readout-fidelity F:
P(measure 1) = p1·F + (1−p1)·(1−F)
each round ≈ readout + pulse time (here 0.4 μs)
loop repeats until 2 consecutive "0" reads confirm the ground state
Each conditional π-pulse also has a small residual error probability — it can occasionally fail to flip the state, which is why the loop re-checks rather than trusting a single shot.
- Passive / Active — pick which reset strategy runs on Prepare |1⟩ & Reset.
- T₁ slider — the qubit's natural relaxation time; longer T1 makes passive reset proportionally slower but barely touches active reset.
- Readout fidelity — how reliably the resonator measurement tells |0⟩ from |1⟩; lower fidelity forces more feedback rounds.
- π-pulse error — chance a conditional flip pulse silently fails, occasionally re-arming the loop.
Real-world relevance: this exact measure→decide→correct loop is the basic building block of active qubit reset on IBM/Google/Rigetti hardware, and the same primitive — mid-circuit measurement feeding a classical decision back into the quantum circuit — underlies quantum error correction and real-time feedback control more broadly.