Engineers at Google Quantum AI have made significant strides by cultivating "magic states" on their Willow processor with an impressive accuracy of 99.99%. This advancement is crucial for performing complex operations while maintaining error protection in quantum computing.
The innovative research, led by physicist Emma Rosenfeld and theorist Craig Gidney, was published in a prominent scientific journal and involved nearly three hundred collaborators. The team's method of "magic state cultivation" drastically reduced errors by a factor of 40 when preparing one of the essential resources for quantum computers.
In quantum computing, qubits, the basic units of information, can exist in states of 0 and 1 simultaneously, which allows for different computational capabilities than traditional processors. However, these states are extremely fragile and can be disrupted by minor environmental interferences. To safeguard against errors, multiple physical qubits are combined to form one "logical" qubit. While simple operations can be executed easily under this error correction, the execution of more complex operations poses challenges, making the use of "magic states" essential. These pre-prepared quantum states are integrated into calculations to facilitate difficult operations without compromising error protection.
Previously, the primary method for generating magic states was distillation, which involved selecting high-quality states from a large batch of imprecise preparations, a process that required extensive equipment. In contrast, Gidney and his colleagues proposed a new approach theoretically in 2024, focusing on gradually enhancing the size and reliability of a single state without reprocessing an entire batch. At each step, the state is verified, and unsuccessful attempts are discarded, akin to a gardener nurturing a single plant through careful tending.
In their experiments, the team initially cultivated the magic state within one error correction scheme before transitioning it to a surface code, the foundation for Google's quantum processors. The resulting states exhibited a remarkable accuracy, reflecting no more than one error per 10,000 attempts. However, achieving this level of precision came at a cost, as only about 8% of attempts in the most stringent trials were successful, with the remaining attempts discarded. Notably, this accuracy is comparable to results achieved in ion trap systems, but Google's superconducting qubits operate at significantly higher speeds.
The most challenging aspect was accurately measuring the results, as temperature fluctuations in laboratory equipment distorted readings. To assess the reliability of the states without these interferences, the team developed a new measurement technique called kickback tomography.
This breakthrough not only showcases Google's leadership in quantum computing but also sets a high benchmark for competitors in the field, potentially reshaping the landscape of quantum technologies.
Informational material. 18+.
