Quantum error correction faces a fundamental question about whether to reset qubits after measurement. While textbook approaches recommend resetting, practical implementation challenges have led many experiments to adopt a “no-reset” approach. This research explores the comparative performance of three reset strategies: unconditional reset (the traditional method), conditional reset, and no reset.
For quantum memory experiments, this research confirms through numerical simulations that the choice of reset strategy has minimal impact on performance, supporting previous claims that no-reset approaches are viable alternatives. However, a significant difference emerges during logical operations, particularly in fault-tolerant quantum computation scenarios like lattice surgery.
The key finding is that unconditional reset can potentially double the number of measurement errors that can be tolerated during logical operations. This translates to reducing the duration of fault-tolerant logical operations by up to a factor of two compared to no-reset approaches. The advantage stems from how measurement errors propagate differently through the system depending on the reset strategy.
However, this theoretical advantage depends heavily on reset speed and fidelity. Numerical simulations reveal that no-reset approaches perform better when reset operations are slow (exceeding approximately 100ns) or when physical error probabilities are relatively high (above 0.003). This insight provides practical guidance for near-term quantum computing experiments where error rates remain significant.
To address the time overhead of no-reset approaches, the researchers propose two novel syndrome extraction circuits. The first uses an additional two-qubit gate per measurement round to transform classification errors into more manageable forms, nearly matching unconditional reset performance. The second circuit employs two auxiliary qubits per stabilizer, effectively compressing two measurement rounds into one. While requiring approximately 50% more qubits, this approach demonstrates substantially better performance in the tested noise regimes.
These findings offer valuable guidance to experimentalists designing quantum error correction systems. The optimal reset strategy depends on specific hardware capabilities and error characteristics. For superconducting quantum computing platforms with fast, high-fidelity reset operations, unconditional reset offers significant advantages for logical operations. However, for current experimental systems with higher error rates or slower reset operations, no-reset approaches with optimized circuit designs may be preferable.
Gehér, G.P., Jastrzebski, M., Campbell, E.T. et al. To reset, or not to reset—that is the question. npj Quantum Inf 11, 39 (2025). doi:10.1038/s41534-025-00998-y
