A wave packet in a chaotic billiard scrambles into apparent randomness, yet averaged over long times it remains at least twice as likely to be found in its original condition as in any comparable one.

Chaos is supposed to be forgetful. That is close to its definition: in a chaotic system, trajectories starting from nearly identical conditions diverge exponentially, and after enough time the system’s state tells you nothing about where it began. Stir cream into coffee and the swirl pattern loses all trace of the first drop.

A study in Physical Review X from Tampere University and Harvard reports that quantum systems do not manage this. Even in full chaos, they retain what the authors call a quantum birthmark — a permanent bias toward their own starting condition that does not fade.

“In the everyday world, chaos wipes the slate clean,” said Joonas Keski-Rahkonen of Tampere University, one of the study’s lead authors. “What we found is that quantum systems can’t hide their origin, even in the middle of chaos.”

The stadium, and what happens in it

The system studied is a stadium-shaped billiard — a rectangle capped with semicircular ends, which is the standard laboratory of quantum chaos. Classically, a ball bouncing inside it follows chaotic trajectories; the shape was chosen decades ago precisely because its classical dynamics are provably chaotic, making it the cleanest place to ask what quantum mechanics does under the same conditions.

The researchers launched a quantum wave packet — a localised lump of probability, the nearest quantum analogue of a ball at a definite place — and watched it evolve. It scrambled quickly, spreading into a pattern that looks random, exactly as chaos predicts.

The finding is in what survives that scrambling. Averaged over long periods, the system remained at least twice as likely to be found in its original condition as in any other comparable condition. The bias persists for as long as the system stays quantum.

A factor of two is modest in magnitude and categorical in meaning. The expectation from chaos is not a small preference for the initial state but none at all — a system that has genuinely forgotten should be equally likely to be found in any of its accessible states. Being twice as likely to sit in one particular state, indefinitely, is a qualitative failure of that expectation.

Ergodicity, and why this breaks it

The property at stake has a name: ergodicity. An ergodic system, given time, visits all of its accessible states with equal probability, so that a long-time average over one trajectory equals an average over all possible states. Ergodicity is the assumption that makes statistical mechanics work — it is why a gas in a box can be described by temperature and pressure rather than by tracking every molecule.

The paper’s title names what it reports: ergodicity breaking beyond scarring. A system with a permanent twofold preference for its starting state is not ergodic, and the breaking is not a transient effect that disappears at long times. It is structural.

A forty-year-old anomaly, generalised

The result’s lineage runs through quantum scarring, discovered in 1984 by Eric Heller — now a Harvard professor and a co-author on this paper. Scars are concentrations of quantum probability along the paths of unstable classical periodic orbits: in a chaotic billiard, certain wavefunctions are unexpectedly intense along particular closed trajectories, as though the classical orbit had left a mark on the quantum state. Scarring has been a well-studied curiosity for four decades, generally regarded as a special feature tied to those specific orbits.

The new work reframes it. Scarring, on this account, is a special case of a universal effect: quantum systems retain an imprint of their initial condition generally, and scars are what that imprint looks like when the initial condition happens to sit on a periodic orbit. That is a reinterpretation of a known phenomenon rather than a new phenomenon, and if it holds it is the more significant kind of result — it says the anomaly was never anomalous.

Where it might matter

Keski-Rahkonen makes two claims for the implications, of rather different weight.

The foundational one is the stronger. “That matters beyond textbook physics,” he said. “It speaks to how quantum systems settle into equilibrium, and how our familiar classical world emerges from the quantum-mechanical rules.” How a quantum system thermalises — reaches equilibrium — and how classical behaviour arises from quantum mechanics are both open problems, and both rest on assumptions about quantum systems exploring their available states. A permanent memory of initial conditions is a constraint those accounts have to accommodate.

The practical claim is more speculative. “Moreover, quantum simulators and nanoscale electronics are nowadays small enough that these effects count,” he said. “For instance, it matters that a system which quietly remembers its starting point behaves differently from one that forgets.”

The logic is sound as far as it goes: devices have shrunk to scales where quantum coherence is maintained long enough for such effects to be relevant, and a quantum simulator intended to model a thermalising system would give misleading answers if it retained a bias toward however it was initialised. The release’s further suggestion that birthmarks and scars could eventually be harnessed for next-generation nanoelectronics is a longer reach — harnessing a subtle statistical bias as a device mechanism would require control that nobody has demonstrated.

What to hold lightly

This is a theoretical and computational result in a model system. The stadium billiard is the right model system, with four decades of literature behind it, but it is still an idealisation — a clean two-dimensional cavity with hard walls and no environment. Real systems couple to their surroundings, and that coupling destroys quantum coherence. The qualifier in the finding is explicit: the bias persists for as long as the system stays quantum, which in any laboratory device is a finite and often short time.

The release gives no information on how the effect scales with system size or energy, no indication of whether it has been confirmed in other geometries, and no experimental test. The factor of two is the only number provided.

The release names only Keski-Rahkonen and Heller, describing the former as one of the study’s lead authors, and does not identify a corresponding author, so the endnote below carries the paper title alone. No funding sources and no conflict-of-interest statement are listed. The paper was published on 10 September 2026, with the release following on 5 October.

The phrase the authors chose is apt. A birthmark is permanent, specific to the individual, and tells you nothing useful about them except where they came from. On this account every quantum system carries one.

---

Notes

1. “Quantum Birthmarks: Ergodicity Breaking Beyond Scarring.” Physical Review X, 10 September 2026. DOI: 10.1103/dhzb-28rb.
2. Tampere University. “Quantum systems never quite forget where they came from.” EurekAlert!, 5 October 2026. https://www.eurekalert.org/news-releases/1146482.