Hefei National Laboratory Team Develops Elea-Cafe Workflow for High-Precision CZ Gate Calibration

Huili Zhang and colleagues at Beijing Academy of Quantum Information Sciences and Hefei National Laboratory present a new calibration workflow that sharply enhances CZ gate fidelity. They achieved a CZ gate fidelity exceeding 99.9% on an 84-qubit processor, suppressing coherent errors to just 0.007%, and demonstrated a median fidelity of 99.25% across 72 parallel CZ gates. The workflow provides an efficient and automated method for quantum computation using superconducting quantum systems, representing a key advance in the field.

Automated calibration achieves record fidelity and stability in 84-qubit superconducting processor

Error rates for CZ gates dropped to 0.007%, a substantial improvement over previous methods. Comparable fidelity on large processors had previously proved difficult to achieve. Dr. Yunseong Nam and colleagues at the Institute of Quantum Technology utilised a closed-loop workflow, employing diagnostic circuits named ELEA and CAFE, to suppress population leakage and refine gate parameters with unprecedented precision. This process also enhanced the stability of the CZ gate over extended monitoring periods, lasting nine hours, establishing an efficient route to quantum computation with superconducting quantum systems.

This breakthrough exceeds 99.9% CZ gate fidelity on an 84-qubit processor, overcoming limitations imposed by increased incoherent errors and demanding calibration requirements as systems scale up. A median fidelity of 99.25% was achieved across 72 concurrent CZ gates, demonstrating the scalability of the automated calibration workflow. This stability was maintained during nine hours of continuous monitoring. Despite these strong advances, maintaining such high fidelity as qubit counts increase further remains a challenge, as does demonstrate error correction beyond these initial two-qubit operations.

Scaling automated calibration to enable universal quantum computation

CZ gate fidelity exceeding 99.9% represents a vital step towards scalable, fault-tolerant quantum computation. However, the current calibration workflow addresses only this specific two-qubit gate. While parallel calibration of 72 CZ gates was successfully demonstrated, the adaptability of this automated process to more complex gate arrangements necessary for universal quantum computation remains a significant question. Extending these gains beyond CZ gates, and to multi-qubit interactions, will be essential to unlock the full potential of this technology and build truly flexible quantum processors.

The Hefei National Laboratory team has established an automated calibration workflow that sharply improves the precision of CZ gates on a superconducting processor. By employing diagnostic circuits, echoed leakage error amplification and repurposed context-aware fidelity estimation, they successfully suppressed unwanted leakage of quantum information. This resulted in a CZ gate fidelity exceeding 99.9% with coherent errors reduced to 0.007% and a maximum fidelity of 99.92% on their 84-qubit Shenglian processor. This demonstration, sustained over nine hours, validates the workflow’s stability and scalability for parallel gate calibration, confirming the Institute of Quantum Technology’s findings regarding long-term performance.

Superconducting qubits are currently a leading platform for realising quantum computation, but their susceptibility to noise presents a significant hurdle. As the number of qubits in these processors increases, so too does the accumulation of errors. These errors arise from both incoherent processes, such as energy relaxation and dephasing, and coherent errors, stemming from imperfections in the control pulses used to manipulate the qubits. The CZ (controlled-Z) gate, a fundamental two-qubit gate, is particularly crucial for implementing many quantum algorithms. Achieving high fidelity in CZ gate operations is therefore paramount for building practical quantum computers. The challenge lies in precisely calibrating the control parameters of these gates while minimising the impact of both coherent and incoherent error sources.

The researchers addressed this challenge by developing a closed-loop calibration workflow. This workflow leverages two key diagnostic circuits: echoed leakage error amplification (ELEA) and repurposed context-aware fidelity estimation (CAFE). ELEA is designed to amplify the effects of leakage errors, which occur when quantum information escapes from the computational subspace. By carefully analysing the signal obtained from the ELEA circuit, the team could identify and correct for parameters that contribute to leakage. CAFE, on the other hand, provides a robust estimate of the gate fidelity, taking into account the specific conof the gate operation. This allows for a more accurate assessment of the gate performance and facilitates finer adjustments to the control parameters. The combination of these two circuits within a closed-loop system enables automated and iterative refinement of the CZ gate calibration.

The implementation of this workflow on the 84-qubit Shenglian processor yielded remarkable results. The achieved CZ gate fidelity of 99.9% represents a significant improvement over previously reported values for processors of comparable size. Critically, the coherent error contribution was suppressed to just 0.007%. This level of precision is essential for enabling fault-tolerant quantum computation, where errors must be actively detected and corrected. Furthermore, the team demonstrated the scalability of the workflow by calibrating 72 CZ gates in parallel, achieving a median fidelity of 99.25% across all gates. The stability of the calibration was also confirmed by maintaining this performance over a continuous monitoring period of nine hours, demonstrating the robustness of the automated process. The long-term stability is particularly important for complex quantum algorithms that require sustained, high-fidelity gate operations.

While this work represents a substantial advance, several challenges remain. The current calibration workflow is specifically tailored to the CZ gate. Expanding this automated process to encompass other essential quantum gates, such as single-qubit rotations and multi-qubit interactions, is crucial for achieving universal quantum computation. Moreover, demonstrating effective quantum error correction beyond these initial two-qubit operations is a necessary step towards building truly fault-tolerant quantum computers. Future research will focus on addressing these challenges and exploring the potential of this automated calibration workflow to enable the construction of larger and more powerful quantum processors. The ability to reliably calibrate and control a large number of qubits with high precision is a cornerstone of realising the full potential of quantum computing, and this work provides a promising pathway towards that goal.

The researchers demonstrated a CZ gate fidelity exceeding 99.9% on an 84-qubit processor, with coherent error suppressed to 0.007%. This precision is important because it reduces errors in gate operations, a critical step towards building reliable quantum computers. They also calibrated 72 CZ gates in parallel, achieving a median fidelity of 99.25%, and maintained this performance for nine hours, showing the process is stable. The authors intend to expand this automated calibration workflow to other quantum gates and explore its use in larger processors.