The Google Quantum AI team has successfully demonstrated magic state cultivation on its Willow superconducting quantum processor. The research, published in the journal Science, confirms that this process, previously confined to theoretical models, is now viable on actual quantum hardware.
Magic states are essential for universal quantum computing, specifically for executing T-gates. Traditional methods for preparing these states are resource-intensive, potentially consuming up to 95% of a quantum computer's capacity. The new cultivation method involves injecting noisy quantum states into a small area of the chip, followed by systematic measurements to filter out errors.
Experimental results show that the logic error rate dropped from approximately 0.44% to a range of 0.01% to 0.02%, representing a 40-fold improvement. The logical fidelity reached 99.99%. However, the process retains only about 8% of the experimental results, indicating that the error reduction comes with a significant trade-off in success rates.
The team also demonstrated the ability to convert and transfer a distance-3 color code logical state into a distance-5 surface code. While the Willow chip features 105 physical qubits, these results are limited to specific experimental circuits rather than large-scale, fault-tolerant programs. Researchers also addressed hardware control issues caused by temperature fluctuations and electronic drift.
