← The feed Many Minded · The brief
KNOWLEDGE · forward · impact 2/5 · 2026-11-17

Qubit protected from magnetic noise without any mid-run measurement

A USTC and Zhejiang team kept a logical qubit alive almost 12 times longer than its best bare qubit under added noise, with no measurements or feedback.

Researchers at the University of Science and Technology of China and Zhejiang University have demonstrated a quantum error correction method that needs no repeated error checks and no real-time feedback. The work appeared in Nature Physics in 2026 under the title "Beating the break-even point with autonomous quantum error correction", with a preprint on arXiv (2504.16746). Yi Li is one of the lead authors.

One logical qubit was stored across four energy levels of a single trapped calcium ion. The protocol targets phase errors caused by fluctuations in the surrounding magnetic field. An entropy-conversion pulse shifts the disturbance from the error into the shared vibrational motion of two trapped ions. A second, "coolant" ion then removes that entropy by laser cooling, which resets the system for the next correction cycle.

The team compared the logical qubit's lifetime with the best-performing physical qubit in the setup. With magnetic noise deliberately added, the protected qubit held information for almost 12 times as long. Under normal lab conditions, the estimated lifetime was 42.7 ms against 29.4 ms. None of this used intermediate measurements, real-time feedback or post-selection.

This matters for abundance because error correction is the main overhead standing between today's quantum devices and machines useful for science and materials work. A scheme that corrects errors without measuring them could cut the hardware and control burden. If that holds at scale, it would make research computing cheaper and more available, which feeds knowledge and, over time, health work such as materials and molecular research.

The caveats are real. The method corrects only certain phase errors from magnetic field fluctuations, not every error type, and the normal-conditions lifetime is an estimate. Multiple logical qubits, entangling gates between encoded ions and combination with larger error-correcting codes are still future goals. Benefits for quantum memories and networks remain potential, not demonstrated. A MIT and Caltech paper by Kyle DeBry and colleagues, published a few months earlier in Nature Physics, also used multiple levels of a trapped ion for autonomous error correction and reached break-even.

Source: Phys.org