14 May 2025

Logical qutrit and ququart memories exceed break-even in bosonic error correction

Quantum technologies
AI illustration · Illustrative scene, not a photograph of the specific device or experiment.

What was demonstrated

In an experiment reported in Nature, a logical qutrit (d=3) and a logical ququart (d=4), each encoded in a Gottesman-Kitaev-Preskill (GKP) bosonic code and used as a quantum memory, were error-corrected beyond break-even. The authors report gains of 1.82 ± 0.03 for the qutrit and 1.87 ± 0.03 for the ququart. A reinforcement learning agent was used to optimise the GKP qutrit and ququart as ternary and quaternary memories.

Why it matters

The experiment extends the error-corrected memory demonstration to logical qutrits (d=3) and ququarts (d=4), not only two-level logical units. The work is a memory demonstration only.

Caveats

  • The result is about storing error-corrected logical states. It does not show logical gates, universal fault-tolerant computation or application-scale computing.
  • The gain figures were checked against the preprint abstract (arXiv:2409.15065). The Nature version was not read in full, because the publisher page redirects to a login page. Its full error budget, methods and comparison baselines were not checked for this record.
  • The result depends on the GKP code and the reinforcement-learning-optimised recovery used in this experiment.
  • The preprint abstract describes a harmonic oscillator and does not name the hardware platform, so the link to superconducting-quantum-computers should be checked against the full text.

Technology profiles

Sources

  1. peer reviewed paper
    Quantum error correction of qudits beyond break-even
    Benjamin L. Brock, Shraddha Singh, Alec Eickbusch et al.. Nature, 14 May 2025. DOI: 10.1038/s41586-025-08899-y.
  2. preprint
    Quantum Error Correction of Qudits Beyond Break-even
    Benjamin L. Brock, Shraddha Singh, Alec Eickbusch et al.. arXiv, 23 Sept 2024.

Last reviewed 22 Oct 2026.