Learning about quantum noise at scale

Abstract

The main challenge to the realization of quantum computers is a complex set of noise processes that inevitably corrupt calculations. The ability to learn faithful representations of noise affecting a quantum system can reveal the underlying physical processes that generate the noise, thus allowing one to suppress, mitigate, or even correct its effect. Naively learning the full set of noise processes is challenging in a large-scale quantum device due to the exponential number of parameters. Furthermore, protocols that attempt to learn the noise are themselves also affected by noise and in general their efficacy and faithfulness in learning are weakened. Here, we show both theoretical and experimental results that demonstrate robust enhancement even in the presence of imperfections. Our results highlight the utility of noisy quantum systems in learning and characterizing physical processes.