Qubit reset via adaptive thresholding: a scalable approach for large QPUs

Qubit reset is an essential operation for quantum computing. Compared to passive reset relying on qubit relaxation, measurement-based active reset protocols significantly reduce the wait time. However, the fidelity of an active reset operation is inherently limited by the single-shot readout fidelity. While a Repeat-Until-Success protocol addresses this limitation, its execution time is non-deterministic, potentially causing prolonged wait times on large scale quantum processing units.

In this work, we optimize a novel adaptive thresholding protocol for active reset. We simultaneously prepare an array of qubits in a sequence of measurements, dynamically updating the threshold at each step for each qubit. To perform such updates, our protocol employes a real-time Bayesian estimation using all previous measurement outcomes.

Our results show that this adaptive thresholding protocol achieves high reset fidelity in deterministic time, without the limitation on high single-shot readout fidelity.