Benchmarking Qubit Reset Techniques that leverage fast advanced control flows and doubly-thresholded readout

Initializing qubits to a known state, is a requirement for quantum algorithms. This reset is required after intentional operations that leave qubits in an unknown state as well as unintentional processes that result in "hot" qubits in a statistical mixture. Traditional methods rely on relaxing the qubit to a thermal steady state by either slow passive qubit decay or faster by pumping of the excited state population to a short-lived (readout) resonator. The quality of reset in both methods is limited to a thermal equilibrium. Alternatively, fast active reset with conditional pi pulses has been explored extensively to overcome this limit.

In this talk, we present two novel reset techniques that capitalize on the advanced features of the Qblox Cluster: (1) active reset using two thresholds, and (2) feedback-driven measurements with conditional scheduling, that enable qubit characterization even when qubits are too “hot” to calibrate with simple active reset. Finally, we reduce experiment time by heralding, without the need for postprocessing on the host PC.

Our results show these methods to improve qubit reset fidelity, reduce experiment time, and lower overhead, making qubit reset more reliable and efficient.