Short answer: Qubit reset returns a qubit to its ground state so computation or error-correction cycles can start cleanly. Published superconducting-qubit experiments report reset in tens to a few hundred nanoseconds without waiting for the qubit to decay on its own, using flux modulation, a microwave-induced interaction with the readout resonator, or an engineered cold bath. All are laboratory results on small devices.
Why reset matters
A 2018 Physical Review Letters paper calls active qubit reset a key operation in many quantum algorithms, and in particular in quantum error correction. A 2021 Nature Communications paper says qubit initialisation is a critical task, and that earlier approaches were either measurement-based and needed fast feedback, suffered from crosstalk, or needed sophisticated calibration.
Microwave-induced reset, 2018
The 2018 paper demonstrates an all-microwave reset for a three-level transmon coupled to a large-bandwidth resonator. The protocol uses a microwave-induced interaction between the second excited state of the transmon with no photons and the ground state with one photon in the resonator. The resonator then lets the photon leak out. The authors characterise the process and demonstrate reinitialisation of the transmon-resonator system to its ground state.
Flux-modulated swap, 2021
The 2021 paper reports a fast, high-fidelity reset that needs no additional chip architecture. By modulating the flux through a transmon, the authors realise a swap between the qubit and its readout resonator that suppresses the excited-state population to 0.08 +/- 0.08% within 34 ns, or 284 ns when photon depletion of the resonator is needed. They also report effective second-excited-state depletion, negligible effects on neighbouring qubits, and a way to entangle the qubit with an itinerant single photon.
An engineered dissipative bath
A paper in Physical Review Applied demonstrates unconditional reset of a frequency-tunable transmon using a metamaterial waveguide engineered to provide a cold bath over a wide spectral range, while protecting the qubit against Purcell decay. It resets several excited states at once. The authors report reset error below 0.13% for the first excited state and 0.16% for the second, within 88 ns.
Reset with a Purcell-protected resonator, 2022
A 2022 arXiv paper (2202.06202) couples a qubit to a resonator with an intrinsic Purcell filter, which suppresses resonator-mediated qubit decay by more than two orders of magnitude over a 600 MHz bandwidth. With it the authors demonstrate a 100 ns reset with residual excitation below 1.7%, alongside a 40 ns readout with 99.1% fidelity.
Caveats
The schemes use different devices and report different metrics, such as residual population or error, so the figures are not a ranking. Each paper reports results on a small number of qubits.
Choosing between schemes
The papers show different trade-offs. The flux-swap scheme needs no extra chip architecture but modulates the qubit frequency. The all-microwave scheme needs a three-level transmon and a resonator with large bandwidth. The bath-engineering scheme needs a dedicated metamaterial waveguide and resets several levels at once. The Purcell-filter design builds reset and readout into one resonator. None of the papers says which is best in general, and each reports its own measure of success, so a designer would weigh them against the target device and gate times.
FAQ
What is unconditional reset?
Reset that does not depend on first measuring the qubit state and applying feedback.
How fast was the fastest reset reported here?
The 2021 flux-swap scheme reports 34 ns, or 284 ns when resonator photons must be depleted.
Does reset also remove higher-level leakage?
Several of the papers report depleting the second excited state as well.
Why use a Purcell filter?
It suppresses qubit decay through the resonator, which otherwise limits how fast the resonator can be made.
Is this in use on commercial machines?
These papers report laboratory experiments and make no such claim.
All field notes