Probing the classical complexity of quantum dynamics experiments

For decades, a simple idea has guided the race to build quantum computers: that simulating the behavior of large quantum systems is so difficult for classical machines that it is practically impossible. This difficulty is the very reason scientists believe quantum computers will eventually outperform their classical counterparts, solving problems in chemistry, materials science, and cryptography that are currently out of reach. The standard way to measure this difficulty has been to look at how much information is shared between different parts of a system. If the parts are deeply linked in a complex, non-local way, the system is considered hard to simulate. However, a growing body of recent work has suggested that this traditional view might be incomplete. It appears that many quantum experiments, even those that look incredibly complex, can actually be predicted by classical computers if those computers focus only on local information—how nearby particles influence one another—while deliberately ignoring the more distant, complicated connections.

This raises a critical question for the field: if classical computers can sneakily simulate so many quantum experiments, how do we know which ones are truly hard? And more importantly, how can experimentalists tell if their own quantum devices are doing something that a classical computer cannot? To answer this, researchers Thomas Schuster and Andreas Elben have introduced a new way to measure the complexity of a quantum experiment. They call it "reactivity." Unlike previous measures that look at a quantum state in isolation, reactivity looks at the entire experiment as a whole. It asks a simple question: how much does the outcome of the experiment change if you poke it with a small, local disturbance? If the experiment is sensitive to these small, local nudges, it is considered simple and easy for a classical computer to simulate. If it is largely unaffected by local changes, it is likely complex and potentially beyond the reach of classical simulation.

The researchers developed a method to measure this reactivity in the lab, a technique they call "Pauli path spectroscopy." Instead of trying to calculate the answer on a supercomputer, which is often impossible for large systems, the experimentalist simply runs the quantum experiment multiple times. In some runs, they intentionally insert a small amount of noise or a random local change at a specific moment. By comparing the results of the noisy runs to the clean runs, they can mathematically reconstruct how much the experiment relies on local versus non-local information. This process is like listening to how a bell rings after being struck; the sound reveals the structure of the metal. Here, the "sound" is the change in the measurement outcome, which reveals the "structure" of the quantum complexity.

The team tested this idea using numerical simulations of two different types of quantum systems. In one case, a system that is known to be easy for classical computers to simulate, they found that the reactivity was concentrated on small, local changes. The experiment was very sensitive to local noise, meaning a classical computer could easily track the outcome by focusing on nearby interactions. In the second case, a system known to be much harder to simulate, the reactivity shifted. The experiment became less sensitive to small, local changes and more reliant on complex, non-local connections that span the entire system. This shift meant that classical algorithms, which rely on tracking local information, would struggle to keep up. The researchers found a strong correlation between this measured reactivity and the actual computational resources required to simulate the system: as the reactivity moved toward larger, more complex scales, the memory needed for a classical simulation grew exponentially.

Beyond just measuring complexity, the paper shows that this new metric has practical implications for what quantum computers can do. The researchers proved that if an experiment has low reactivity, it can be "learned" efficiently. This means that by running the experiment with random settings and measuring the outcomes, a classical computer can build a model that predicts the result for any new setting without needing to simulate the full quantum evolution. Similarly, they showed that low-reactivity experiments can be "fast-forwarded," allowing the prediction of long-term behavior by combining measurements of shorter segments. These findings suggest that the reactivity is not just a theoretical curiosity but a practical tool for identifying which quantum experiments are truly advancing the field.

The work also highlights a boundary in our current understanding. The researchers note that while reactivity is a powerful new tool, it is not a perfect shield against all classical tricks. Just as a complex-looking circuit could be designed to fool a specific test, a cleverly constructed quantum experiment could appear complex by this measure while still being simple in other ways. However, the ability to measure reactivity directly in an experiment, even when the experiment itself is too large for a classical computer to simulate, is a significant step forward. It allows scientists to probe the "classical complexity" of their devices in real-time, determining whether they are truly operating in a regime where quantum advantage exists. By providing a way to distinguish between experiments that are merely complex and those that are fundamentally hard to simulate, this research offers a new compass for navigating the future of quantum technology.