Transmon readout in the large detuning limit – PART 2 / 2: Results and their interpretation

High-fidelity non-demolition measurement of qubits is crucial for quantum error correction. Superconducting transmon qubits are measured via the dispersive interaction between the transmon and a readout resonator. Fast dispersive readout relies on measuring the response of the resonator when it is subjected to a strong probe tone, but the stronger the probe, the more probable are undesired transitions to highly excited transmon states. These transitions limit readout fidelity and lead to uncorrectable errors. In this work, we suppress this leakage to highly excited transmon states by detuning our transmon qubit far from the readout resonator. This reduces the number and strength of the resonant multiphoton processes responsible for leakage, while conserving speed and increasing fidelity. We then elucidate the properties of dispersive readout in the far-detuned regime by deriving a simple model that accounts for the dressing of the transmon states by the strong probe tone.