arXiv:2610.11477
Quantitative Rules for Parallel Calibration and Drift-Resilient Maintenance of Large-Scale Superconducting Quantum Processors
Zhihao Wu, Kechengqi Zhu, Shichuan Xue, Xinfang Zhang, Zhaohua Yang, Zhida Zhang, Shun Hu, Cong Li, Dingdong Liu, Kangding Zhao, Peng Luo, Lingling Lao, Xiang Fu, Mingtang Deng, Junjie Wu, Kai Lu
Abstract
As superconducting quantum processors scale toward hundreds of qubits, manual calibration becomes increasingly costly and difficult to sustain. Here we address two system-level factors governing automated calibration, namely, crosstalk induced by parallel operations and temporal drift of control parameters. On a 66-qubit superconducting processor, cross-entropy benchmarking shows that single-qubit calibration tolerates full-chip concurrency with minimal fidelity loss, whereas CZ calibration requires a minimum coupler-graph distance of four to avoid elevated control errors. Continuous monitoring of representative qubit and coupler parameters further yields quantitative refresh rules that assign refresh priority by error-budget consumption. These results turn two conventionally heuristic choices, the concurrency density of calibration tasks and the refresh priority of each control parameter, into measurable operating rules. Implemented within an automated calibration framework, the rules support a six-qubit Greenberger-Horne-Zeilinger state with a zero-noise-extrapolated fidelity of 89.64%, distance-dependent logical-error suppression in a single-cycle repetition-code benchmark, and a reduction of the mean single-qubit error from 0.023 to 0.011 on a 337-qubit processor, providing an experimentally grounded reference for the automated calibration and maintenance of current and future large-scale superconducting quantum systems.
AI Summary
The paper investigates how calibration concurrency and parameter drift constrain automated operation of superconducting processors. Experiments use a 66-qubit, 109-coupler device to select parallel-calibration densities and prioritize parameter refreshes within a dependency-ordered calibration workflow. Single-qubit XEB error distributions remain nearly unchanged from isolated operation to full-chip concurrency, with measured errors below $0.4\%$. For CZ calibration, minimum coupler-graph distance $d_{\mathrm C}=2$ increases mean control error to $0.91\%$, compared with $0.61\%$, $0.59\%$, and $0.54\%$ at distances 4, 6, and isolated operation. Control error is estimated by subtracting speckle-purity benchmarking error from XEB error. Distance 4 is therefore the densest tested configuration preserving approximately isolated performance; timing fits indicate saturation from classical-control overhead. Monitoring 26 qubits and 36 couplers over 44–48 hours yields refresh priorities using one-hour-equivalent parameter changes, calibration uncertainty, and modeled gate-error-budget consumption. CZ coupling, qubit frequency, and CZ dynamic phase receive highest priority. The method produces rankings, not numerical refresh intervals; sequential updates confound underlying drift sources, and the hierarchy is specific to the device and observation window. Circuit validation gives six-qubit GHZ fidelities of $71.92$–$82.10\%$ before zero-noise extrapolation and a maximum extrapolated value of $89.64\%$. Single-cycle repetition-code majority-vote errors decrease by over three orders of magnitude from distance 3 to 9, but are dominated by preparation/readout errors and are not comparable to surface-code errors per cycle. Separately, automated recalibration reduces mean single-qubit XEB error from $0.023$ to $0.011$ across 337 active qubits. The concurrency rule remains specific to the tested layout, pulses, and control resources.
Central claim may not hold: The claimed minimum CZ separation of four exceeds the evidence, which establishes only that four is the densest tested acceptable separation. The drift-priority method neither removes measurement noise from squared increments nor justifies its one-hour temporal scaling, so its detectability threshold does not establish resolved physical drift, particularly for the high-priority dynamic-phase entry. The validation experiments lack a comparison that isolates the benefit of the proposed concurrency and refresh-priority rules from ordinary calibration and maintenance. The microwave-crosstalk appendix presents an unjustified ordinary-integral exponential as exact evolution and uses a non-Hermitian propagator that does not describe unconditional relaxation and dephasing. The flux-crosstalk appendix misidentifies normalized squared trace overlap as average gate fidelity, although its final conditional-phase infidelity coefficient is correct for actual average gate fidelity.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.