CoChat

Quantum error correction of qudits beyond break-even

Open access

The DOI matches its Crossref record; 65 of the 69 references it cites were found in a registry or an index; 14 of the 28 sources its claims cite were checked for support.

Abstract

Hilbert space dimension is a key resource for quantum information processing1,2. Not only is a large overall Hilbert space an essential requirement for quantum error correction, but a large local Hilbert space can also be advantageous for realizing gates and algorithms more efficiently3–7. As a result, there has been considerable experimental effort in recent years to develop quantum computing platforms using qudits (d-dimensional quantum systems with d > 2) as the fundamental unit of quantum information8–19. Just as with qubits, quantum error correction of these qudits will be necessary in the long run, but so far, error correction of logical qudits has not been demonstrated experimentally. Here we report the experimental realization of an error-corrected logical qutrit (d = 3) and ququart (d = 4), which was achieved with the Gottesman–Kitaev–Preskill bosonic code20. Using a reinforcement learning agent21,22, we optimized the Gottesman–Kitaev–Preskill qutrit (ququart) as a ternary (quaternary) quantum memory and achieved beyond break-even error correction with a gain of 1.82 ± 0.03 (1.87 ± 0.03). This work represents a novel way of leveraging the large Hilbert space of a harmonic oscillator to realize hardware-efficient quantum error correction. Quantum error correction of a logical qutrit and ququart were experimentally realized beyond the break-even point with the Gottesman–Kitaev–Preskill bosonic code.

Overview

CoChat Summary

## TL;DR

The paper reports the experimental realization of error-corrected logical qudits beyond the qubit case: a logical qutrit and ququart encoded in a GKP bosonic code. Using reinforcement learning to optimize the error-correction protocol, the authors achieved beyond break-even quantum-memory performance with gains of 1.82 ± 0.03 for the qutrit and 1.87 ± 0.03 for the ququart.

## Key findings

  • The experiment demonstrated experimental error correction of logical qudits with d > 2, specifically a logical qutrit (d = 3) and ququart (d = 4).
  • The optimized logical qutrit outperformed the best physical qutrit in the system by a factor of 1.82 ± 0.03, and the optimized logical ququart outperformed the best physical ququart by 1.87 ± 0.03.
  • The authors report that the QEC gain stayed roughly constant at about 1.8 as the logical GKP qudit dimension increased from 2 to 4.
  • For the ququart, the paper says the QEC gain was “well beyond the break-even point,” and for the qutrit it says the gain was “well beyond the break-even point.”

## Methods

  • Experimental platform: a tantalum transmon dispersively coupled to a three-dimensional superconducting microwave cavity; the cavity hosted the oscillator modes used for the logical GKP states.
  • Code and protocol: the authors used the single-mode square GKP code and adapted the small-big-small (SBS) autonomous stabilization protocol; they optimized protocol parameters with a reinforcement learning agent.
  • Evaluation: they prepared representative eigenstates, applied the optimized QEC protocol for varying numbers of rounds, measured output probabilities in generalized Pauli bases, and fit exponential decay to obtain effective decay rates and QEC gain.

## Limitations

  • The paper states that the logical measurements and generalized controlled-Pauli gates were designed for the ideal code and “incur infidelity” when applied to the finite-energy code.
  • The paper states that the optimized GKP qudits had more energy and thus were expected to suffer increased photon loss and dephasing as d increased; the discussion identifies cavity dephasing as the dominant source of logical error.
  • Evident from the text: the work demonstrates only single-system memory error correction; the paper does not report full-scale multi-qudit computation or entangling-gate demonstrations here.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s41586-025-08899-y

Cites

65 of 65 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Fig. 1 Stabilizing GKP qudits. a, Schematic of the experimental device. b, Geometric structure of the displacement operators that define the single-mode square GKP code. c, Circuit for one round of finite-energy GKP qudit stabilization, generalizing the SBS protocol43. The big ECD gate48 of amplitude ℓd,Δ=√πdcosh(Δ²) is approximately the stabilizer length. The small ECD gates of amplitude εd/2=√πdsinh(Δ²)/2 account for the envelope size Δ. At the end of SBS round j, the cavity phase is updated by ϕj (Methods). d, Measured real part of the characteristic function of the maximally mixed GKP qudit state for d = 1 to 4 with Δ = 0.3, prepared by performing 300 SBS rounds starting from the cavity in its vacuum state |0⟩.
Fig. 1 Stabilizing GKP qudits. a, Schematic of the experimental device. b, Geometric structure of the displacement operators that define the single-mode square GKP code. c, Circuit for one round of finite-energy GKP qudit stabilization, generalizing the SBS protocol43. The big ECD gate48 of amplitude ℓd,Δ=√πdcosh(Δ²) is approximately the stabilizer length. The small ECD gates of amplitude εd/2=√πdsinh(Δ²)/2 account for the envelope size Δ. At the end of SBS round j, the cavity phase is updated by ϕj (Methods). d, Measured real part of the characteristic function of the maximally mixed GKP qudit state for d = 1 to 4 with Δ = 0.3, prepared by performing 300 SBS rounds starting from the cavity in its vacuum state |0⟩.CC BY-NC-ND · PubMed Central
Fig. 2 Realization of a logical GKP qutrit. a, State preparation of qutrit Pauli eigenstates |P0⟩3 with Δ = 0.32. b, Circuit for measuring a qutrit in the basis of Pauli operator P3 using an ancilla qubit, where θ0=2arctan(1/√2). The first measurement distinguishes between the state |P0⟩3 and the subspace {|P1⟩3, |P2⟩3}, whereas the second distinguishes between |P1⟩3 and {|P0⟩3, |P2⟩3}. The Bloch spheres depict the trajectories taken by the ancilla when the qutrit is in each Pauli eigenstate. c, Backaction of the qutrit Pauli measurement in the Z3 basis, applied to the maximally mixed qutrit state. d, Decay of qutrit Pauli eigenstates |P0⟩3 under the optimized QEC protocol. The dashed black lines indicate a probability of 1/3. The solid grey lines are exponential fits. From left to right, we found γX0⁻¹=1,153±13μs, γZ0⁻¹=1,120±15μs, γXZ0⁻¹=743±10μs and γX²Z0⁻¹=727±11μs.
Fig. 2 Realization of a logical GKP qutrit. a, State preparation of qutrit Pauli eigenstates |P0⟩3 with Δ = 0.32. b, Circuit for measuring a qutrit in the basis of Pauli operator P3 using an ancilla qubit, where θ0=2arctan(1/√2). The first measurement distinguishes between the state |P0⟩3 and the subspace {|P1⟩3, |P2⟩3}, whereas the second distinguishes between |P1⟩3 and {|P0⟩3, |P2⟩3}. The Bloch spheres depict the trajectories taken by the ancilla when the qutrit is in each Pauli eigenstate. c, Backaction of the qutrit Pauli measurement in the Z3 basis, applied to the maximally mixed qutrit state. d, Decay of qutrit Pauli eigenstates |P0⟩3 under the optimized QEC protocol. The dashed black lines indicate a probability of 1/3. The solid grey lines are exponential fits. From left to right, we found γX0⁻¹=1,153±13μs, γZ0⁻¹=1,120±15μs, γXZ0⁻¹=743±10μs and γX²Z0⁻¹=727±11μs.CC BY-NC-ND · PubMed Central
Fig. 3 Realization of a logical GKP ququart. a, State preparation of ququart Pauli eigenstates |P0⟩4 and parity state |+, 0⟩4 with Δ = 0.32. b, Circuit for measuring a ququart in the basis of Pauli operator P4 using an ancilla qubit. The first measurement distinguishes between the even and odd states |Peven/odd⟩4, and the second measurement distinguishes between the remaining two states. The Bloch spheres depict the trajectories taken by the ancilla when the ququart is in each Pauli eigenstate. c, Backaction of the GKP ququart Pauli measurement in the Z4 basis applied to the maximally mixed ququart state. d, Circuit for measuring a ququart in the parity basis {|±, m⟩4: m = 0, 1}, where X4²∣±,m⟩4=±∣±,m⟩4 and Z4²∣±,m⟩4=(−1)^m∣±,m⟩4. The first measurement determines the eigenvalue of X4², and the second determines that of Z4². e, Backaction of the GKP ququart parity measurement applied to the maximally mixed ququart state. f, Decay of ququart Pauli eigenstates |P0⟩4 and parity state |+, 0⟩4 under the optimized QEC protocol. The dashed black lines indicate a probability of 1/4. The solid grey lines are exponential fits. From left to right, we found γX0⁻¹=840±8μs, γZ0⁻¹=836±9μs, γXZ0⁻¹=519±6μs, γX²Z0⁻¹=507±9μs, γX³Z0⁻¹=571±7μs, γXZ0²⁻¹=562±9μs and γ+,0⁻¹=607±8μs.
Fig. 3 Realization of a logical GKP ququart. a, State preparation of ququart Pauli eigenstates |P0⟩4 and parity state |+, 0⟩4 with Δ = 0.32. b, Circuit for measuring a ququart in the basis of Pauli operator P4 using an ancilla qubit. The first measurement distinguishes between the even and odd states |Peven/odd⟩4, and the second measurement distinguishes between the remaining two states. The Bloch spheres depict the trajectories taken by the ancilla when the ququart is in each Pauli eigenstate. c, Backaction of the GKP ququart Pauli measurement in the Z4 basis applied to the maximally mixed ququart state. d, Circuit for measuring a ququart in the parity basis {|±, m⟩4: m = 0, 1}, where X4²∣±,m⟩4=±∣±,m⟩4 and Z4²∣±,m⟩4=(−1)^m∣±,m⟩4. The first measurement determines the eigenvalue of X4², and the second determines that of Z4². e, Backaction of the GKP ququart parity measurement applied to the maximally mixed ququart state. f, Decay of ququart Pauli eigenstates |P0⟩4 and parity state |+, 0⟩4 under the optimized QEC protocol. The dashed black lines indicate a probability of 1/4. The solid grey lines are exponential fits. From left to right, we found γX0⁻¹=840±8μs, γZ0⁻¹=836±9μs, γXZ0⁻¹=519±6μs, γX²Z0⁻¹=507±9μs, γX³Z0⁻¹=571±7μs, γXZ0²⁻¹=562±9μs and γ+,0⁻¹=607±8μs.CC BY-NC-ND · PubMed Central
Fig. 4 Comparing GKP qudits. a, Effective lifetime of the physical cavity Fock qudit and logical GKP qudit for d ∈ {2, 3, 4}. The arrows indicate the QEC gain. b, Effective envelope size Δeff of the optimized GKP qudit for d ∈ {2, 3, 4}. c, Mean number of photons in the cavity for the optimized GKP qudit for d ∈ {2, 3, 4}.
Fig. 4 Comparing GKP qudits. a, Effective lifetime of the physical cavity Fock qudit and logical GKP qudit for d ∈ {2, 3, 4}. The arrows indicate the QEC gain. b, Effective envelope size Δeff of the optimized GKP qudit for d ∈ {2, 3, 4}. c, Mean number of photons in the cavity for the optimized GKP qudit for d ∈ {2, 3, 4}.CC BY-NC-ND · PubMed Central

Verification

Claims the paper makes

9 Valid · 0 Not valid · 5 Not checked · 26 Own results and interpretations

15 more claims · 2 with their cited sources checked

Create a free account to see all 40 claims and to verify the 6 references still open.

See all claims — free account
  • It may state two separate claims as one. 13
  • Probably not a substantial claim. 10
  • The quote only partly states this claim. 2

Create a free account to see these claims.

Citations

Cites

Show all 65 references

Cited by

See all 64

Metadata

Authors
Brock, B. L., Singh, S., Eickbusch, A., Sivak, V. V., Ding, A. Z., Frunzio, L., Girvin, S. M., & Devoret, M. H.
Affiliations
  • Yale University (US)
  • Google (United States) (US)
  • University of California, Santa Barbara (US)
Corresponding author
  • Brock, Benjamin L.
  • Devoret, Michel H.
Venue
Nature
Published
2025-05-14
Volume
641
Issue
8063
Pages
612-618
ISSN
  • 0028-0836 (ISSN-L)
  • 1476-4687
OpenAlex
W4410362117
Cited by
64
References
69
Citations by year
2026: 38 · 2025: 26
FWCI
25.47
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature (publishedVersion, cc-by-nc-nd)
  • PubMed Central (publishedVersion, cc-by-nc-nd)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • arXiv (Cornell University)
Funders
  • U.S. Department of Energy (DE‐SC0012704, SC0012704.)
  • Yale University
  • Office of Science (DE-SC-0012704, SC0012704)
  • Air Force Office of Scientific Research (FA9550-, FA9550-19-1-0399)
  • Army Research Office (W911-NF-23-1-0051, W911NF)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Quantum Computing Algorithms and Architecture
  • Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Quantum Information and Cryptography
  • Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Quantum Mechanics and Applications
Keywords
quantum error correction, qudits, qutrits, ququarts, Gottesman-Kitaev-Preskill code, bosonic codes, harmonic oscillator, reinforcement learning optimization, quantum memory

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Quantum error correction of qudits beyond break-even. https://app.cochat.ai/papers/10.1038/s41586-025-08899-y
The paper
Brock, B. L., Singh, S., Eickbusch, A., Sivak, V. V., Ding, A. Z., Frunzio, L., Girvin, S. M., & Devoret, M. H. (2025). Quantum error correction of qudits beyond break-even. Nature. https://doi.org/10.1038/s41586-025-08899-y

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.