Google's Quantum AI team has demonstrated a new 'magic state distillation' method on its Willow processor. This technique addresses a major resource bottleneck in quantum computing, where generating high-quality quantum states previously consumed up to 95% of total system capacity.
The new distillation process acts as a filter, injecting noisy quantum states into a specific area of the chip. Through coordinated measurements, the system identifies and removes errors, resulting in high-purity, low-error quantum states that can be scaled for practical applications.
During the development phase, researcher Emma Rosenfeld identified that electrical drift and temperature fluctuations were interfering with benchmark results. To resolve this, the team implemented additional control measures and developed a new benchmarking tool that uses the distillation technology itself for verification.
The experimental results, published in the journal Science, show that the Willow processor achieved a 40-fold reduction in error compared to non-distilled states. Additionally, the system reached a record-breaking 99.99% logical fidelity for T-states.
The team also successfully validated the compatibility of this theory with their existing architecture. They demonstrated the ability to switch between color codes and surface codes, confirming that purified quantum states can be stably maintained and transferred within the hardware.
