Random circuit sampling experiments have demonstrated quantum advantage by solving finite-fidelity computational tasks at around 0.1% fidelity, which remains exponentially harder for classical computers than trivial baselines. The key insight is a phase transition in the weak-noise regime where cross-entropy benchmarking (XEB)
Table of contents
What’s the issue?The noisy sampling taskQuantum advantage of finite-fidelity RCSGetting the scaling right: weak noise and low depthFidelity versus XEB: a phase transitionWhere are the experiments?The phase transition matters!But we can’t compute the XEB!That’s exactly how experiments work!What are the counter-arguments?Sort: