Nearly 5,000 Resets Steered 100 Qubits Out of Chaos

Researchers running an IBM quantum processor measured and reset individual qubits nearly 5,000 times during a single computation, pulling a chain of up to 100 qubits out of chaotic behaviour and into a state they had chosen in advance. The experiment, published in Nature Physics on 12 October 2026, is among the largest demonstrations of real-time quantum feedback control reported so far, according to the authors.

The team, led from Rutgers University and IBM, used a 156-qubit IBM Quantum Heron processor and selected a connected chain of as many as 100 qubits. The circuit alternated two competing operations. One was a scrambler that mixed information among neighbouring qubits, in the way that repeated shuffling destroys the order of a deck of cards. The other read an individual qubit mid-circuit and reset it when needed. Across the run the group performed nearly 5,000 operations linking pairs of qubits and nearly 5,000 checks and resets.

A sharp line between chaos and control

Which operation ran more often decided the outcome. When scrambling happened more than half the time, the system stayed chaotic. When measurement and reset happened more often, the researchers could drive the qubits towards the orderly state they wanted. At roughly 50-50, a small shift in the balance flipped the whole system at once, a phase transition of the kind that separates water from ice. The paper reports that the transition became more sharply defined as the system grew, which is what the group's theory predicted.

Size is the point. Classical simulation of this kind of circuit runs out at roughly two dozen qubits, so the 100-qubit chain took the theory into territory where it could not otherwise be checked. The hardware had to keep working while being interrupted.

What the result is not

It is not error correction. The group demonstrated the machinery error correction needs, repeated mid-circuit measurement with action taken on the result, but the state they steered to was one they picked rather than a protected logical qubit, and the paper does not claim a longer-lived one. No fault-tolerant quantum computer exists. Jedediah Pixley of Rutgers, a senior author and director of its Center for Materials Theory, set out the size of the remaining gap.

The honest reading is that this is a physics finding in engineering clothes, and the two halves will be useful to different people. Hardware engineers will take the operational claim: that measurement and reset can be coordinated thousands of times across 100 qubits while the processor keeps computing. Physicists get the confirmation that a chaos-to-control transition known in ordinary physics survives in quantum systems, which Justin Wilson of Louisiana State University, a co-author, said was not obvious beforehand.

The theory behind it started with a bet at a birthday party in Highland Park, New Jersey, in 2021, where Pixley's longtime collaborator Sriram Ganeshan wagered that a puzzling quantum effect had a counterpart in ordinary physics. Pixley lost and paid in two scoops of blueberry ice cream. Beyond Rutgers and IBM, the author list runs to the City College of New York, the CUNY Graduate Center, Iowa State University, Ames National Laboratory, Pennsylvania State University and Louisiana State University.