Breakthrough FPGA workflow stabilizes quantum gates with 74,000 recalibrations in hours

Scientists have developed a new on-FPGA workflow to tackle parameter drift in superconducting qubits, enabling millisecond-scale calibration and benchmarking of quantum gates. The system sustained over 74,000 consecutive quantum recalibrations while maintaining high gate fidelity.

The workflow integrates pulse generation, data acquisition, analysis, and feed-forward directly on an FPGA chip, removing delays caused by traditional CPU-based processing. As a result, Clifford randomized quantum gate benchmarking was completed in just 107 milliseconds.

Rapid qubit fluctuations posed a major challenge, with T1 coherence times shifting within two-second intervals. Despite this, the system achieved continuous closed-loop optimization for quantum gates for 24 hours, recalibrating over 74,000 times. Environmental noise and hardware variability created signal model drift, but the workflow suppressed quantum gate error coupling while preserving qubit coherence.

The research also identified optimal parameter regimes for estimating qubit and pulse settings under sparse sampling conditions, allowing for faster readout calibration, spectroscopy, and pulse-amplitude optimization. However, the reliance on specific signal models remains a limitation, as real quantum systems often behave differently than predicted.

Compared to initial calibration, the approach delivered improved quantum gate performance. It also enabled real-time coherence estimation and quantum gate benchmarking without manual intervention.

The on-FPGA workflow marks a step forward in managing qubit instability through rapid, automated quantum recalibration. By sustaining over 74,000 cycles in six hours, it demonstrates potential for long-term quantum system stability. Future work may focus on reducing dependence on fixed signal models to handle more complex quantum behaviors.