量子物理学：「時間を逆転させる」ことで量子ダイナミクスを探る

Manipulating quantum circuits in a way that reverses the scrambling of information could potentially probe the properties of quantum computers and improve their performance, a study in Nature suggests. Google Quantum AI and Collaborators report the measurement of a quantity called out-of-time-order correlators (OTOCs) in a superconducting quantum processor. OTOCs may serve as a tool for understanding quantum computers and building verifiable demonstrations of beyond-classical performance.

Building quantum computers that reach high enough performance to achieve quantum advantage, where they excel over classical computer at some specific and ideally useful task, is a long-standing goal of quantum computing. A number of issues need to be overcome to achieve this goal by reducing noise and imperfections; one such issue is probing the quantum dynamics of the many components of such a system to distinguish genuine quantum effects from classical noise. These systems can be difficult to study because the behaviour of interacting elements becomes unpredictable and hard to track over time, especially when measuring only some of their elements at a given instant in time. A potential solution plays with the concept of time reversal: disturbing a system, letting the disturbance ripple out and then running the system backwards to attempt to reverse the scrambling of information and in this way acquire information over the system as a whole.  

Researchers at Google Quantum AI and Collaborators, led by Hartmut Neven, use time-reversal protocols to measure high-order OTOCs, a tool for studying how quantum information spreads in many-particle quantum systems, in a superconducting quantum processor. They find that the experimental observables remain sensitive to genuine quantum effects for sufficiently long timescales to sample a large part of the processor over the spread-and-reverse dynamics. In this way, measurement of the OTOCs reveals microscopic properties of quantum systems that remain inaccessible to classical computation, the authors add, which they suggest raises the potential to use these sorts of many-particle measurements as an ingredient of potential future robust quantum advantage demonstration like nuclear magnetic resonance.

The authors note that the circuits used in this demonstration are a “toy model”, but they indicate that the scheme could be applied to real physical systems.

• Article
• Open access
• Published: 22 October 2025

Google Quantum AI and Collaborators. Observation of constructive interference at the edge of quantum ergodicity. Nature 646, 825–830 (2025). https://doi.org/10.1038/s41586-025-09526-6