Google researchers address the challenge of evaluating quantum computer performance in the noisy intermediate-scale quantum (NISQ) era, where quantum processors are highly susceptible to noise. The fundamental problem is determining whether quantum systems, despite their noise limitations, can outperform classical supercomputers in specific computational tasks. The research focuses on understanding how quantum computers behave under noise and whether they can still demonstrate quantum advantage—a key milestone in quantum computing.
Random circuit sampling (RCS) has emerged as a leading method to evaluate quantum processors and was introduced in 2019. RCS tasks are computationally hard for classical computers due to the exponential growth of information as quantum circuits scale. The key problem is that classical computers struggle to simulate or sample from a quantum circuit’s output distribution as circuit volume increases. RCS measures quantum circuit volume, a key indicator of performance, which helps identify when quantum systems can surpass classical supercomputers, even in the presence of noise. Google’s research showed a twofold increase in circuit volume while maintaining the same fidelity as previous benchmarks. These advancements suggest that noisy quantum systems can still offer practical value by performing tasks beyond classical capabilities.
The proposed method involves benchmarking quantum devices using RCS to estimate fidelity, measuring how closely the noisy quantum processor mimics an ideal, noise-free system. Researchers introduced patch cross-entropy benchmarking (XEB), a technique to verify fidelity by dividing the full quantum processor into smaller patches. XEB calculations for these patches provide a feasible way to estimate fidelity for larger circuits. The study confirms that despite the noise, current quantum processors like Sycamore are capable of achieving beyond-classical results, doubling the circuit volume compared to earlier experiments while maintaining fidelity. It also identifies phase transitions in RCS behavior based on noise strength and circuit depth, further validating the reliability of RCS for assessing quantum computers.
Along with the impact of noise on quantum processors, Google researchers discovered two distinct noise-induced phase transitions. In low-noise conditions, quantum computers can achieve full computational power. However, high noise levels can create uncorrelated subsystems, making it easier for classical computers to simulate their results. This phase transition helps determine if quantum computers are truly outperforming classical computers. The Sycamore processor operates in a low-noise regime, confirming its quantum advantage.
In conclusion, Google researchers provide a significant step towards fault-tolerant quantum computing by demonstrating how random circuit sampling can effectively measure quantum performance in the presence of noise. The discovery of noise-induced phase transitions offers a new way to understand the behavior of quantum processors under different conditions.