Pith. sign in

REVIEW 3 major objections 4 minor 74 references

A Highly Accurate Fast Decoding Framework for QLDPC codes Accelerated by Noise Perturbation and Ensemble Decoding

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Noise-perturbed Tanner forests make QLDPC decoding fast and accurate.

desk verdict A promising QLDPC decoder framework with a real sign inconsistency in the inference algorithm that has to be resolved before the exactness and performance claims can be trusted. read the letter →

arxiv 2608.09583 v1 pith:OD73L2LV submitted 2026-08-10 quant-ph

Mainak Bhattacharyya , Ankur Raina This is my paper · ORCID
classification quant-ph PACS 03.67.Pp
keywords quantumerrorcorrectionQLDPCcodesTannerforestdynamicprogrammingdecodingensemblecircuit-levelnoisebeliefpropagationorderedstatistics
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper introduces a decoding framework, NAED, that replaces iterative message passing with exact inference over an ensemble of tree-like subgraphs (Tanner forests) of a quantum code's detector error model. The forest construction is guided by a syndrome- and channel-derived column score, and controlled noise perturbations generate diverse forests so that at least one forest spans the observed syndrome. On circuit-level simulations, NAED matches or beats the accuracy of BP+OSD0 for surface codes with only 11 ensemble instances, and for bivariate bicycle codes reaches OSD-level logical error rates when followed by a cheap post-BP stage. Per-round decoding times are reported to be orders of magnitude lower than BP+OSD0, because each forest is decoded by a single linear-time dynamic programming pass.

What carries the argument

The load-bearing object is the noise-perturbed column score $n_b^q = w_q + \tau_b \varepsilon_q$ (Eq. (10)), with base score $w_q = -\mathrm{llr}_q + \alpha \sum_{c \in N(q)} (2s_c - 1)$, which equals the single-flip objective change $f(x=0)-f(x')$ in Eq. (9). This score orders the columns processed by a Kruskal-style forest constructor; for BB codes the score is upgraded to $n_b^q = \kappa w_q + \tau \varepsilon_q + \beta G_q(r)$ (Eq. (15)), where $G_q(r) = 2a_q(r) - |N(q)|$ is the dynamic gain in residual syndrome weight. Inference on each forest is performed by a two-stage dynamic program: an upward pass of two-entry messages with a parity-constrained cheapest-flip rule (Algorithm 1, CheckUpdate) and a downward traceback (Algorithm 2, ETFI), costing $\Theta(|E_F|)$ time per forest. The ensemble pools member results by most-likely-error cost or first-valid output.

What would settle it

The paper's Table V shows that for BB [[108,8,10]] and [[144,12,12]] codes at p=0.002, between 0.16% and 1.55% of sampled syndromes are not resolved by the full-DEM forest stage. A decisive test: take those unresolved syndromes, run NAED with the residual-aware dynamic construction while increasing the ensemble size B well beyond 100, and check whether any member ever produces a syndrome-spanning forest. If the failure set never shrinks, the noise-perturbation diversity claim is falsified for those codes.

Watch

Extended reading notes

Core claim

The central claim is that a decoder does not need BP's iterative soft information: a single-column flip-cost heuristic computed from the syndrome and channel LLRs, numerically equal to the objective change $f(x=0)-f(x')$ in Eq. (9), is enough to prioritize decoding-matrix columns for building a Tanner forest, and adding Gaussian noise to these scores creates an ensemble of forests whose exact inference recovers the error. For graph-like codes such as surface codes, any forest built from the sparse detector error model spans the syndrome, so the ensemble's best solution is guaranteed to resolve it; for BB codes, a residual-syndrome-aware dynamic gain reorders columns online and, together with a low-cost post-BP stage, closes the gap to OSD accuracy. The paper thus targets the degenerate maximum-likelihood decoding problem rather than the standard QMLD formulation, arguing that the ensemble of forest inferences approximates DQMLD better than BP-based approximations.

Load-bearing premise

The load-bearing premise is that the heuristic score of Eq. (7) ranks decoding-matrix columns so well that at least one noise-perturbed Tanner forest in the ensemble spans the observed syndrome or a logically equivalent correction; if no forest spans the syndrome, exact inference on the forests cannot recover the error.

Editorial extensions

If this is right

  • Circuit-level decoding of surface codes can run without BP iterations, replacing the heuristic iteration cap with a single linear-time exact pass per forest.
  • Ensemble parallelism means wall-clock latency is set by the fastest syndrome-valid member, not by the slowest, making the decoder a candidate for FPGA/GPU real-time pipelines.
  • For BB codes, NAED's forest stage resolves over 98% of syndromes; only a cheap post-BP round is needed to reach OSD-level logical error rates.
  • The reported orders-of-magnitude speedup over BP+OSD0 grows with code distance, since the comparison decoder's per-round cost rises faster than the forest construction and inference.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same synthetic-score mechanism could be tuned per code family: the residual-aware dynamic gain $G_q(r)$ is a drop-in replacement for the static score, so codes with higher column weights may inherit the BB-code treatment.
  • If the speedup survives hardware implementation, the per-round latency budget freed by NAED could be spent on denser measurement schedules or higher-distance codes, shifting the real-time bottleneck from decoding to readout.
  • A natural test is to combine NAED with decoder switching: use NAED as the always-on fast decoder and invoke a heavier decoder only on the few percent of syndromes that no forest spans.
  • The framework's dependence on syndrome-spanning forests suggests that code design criteria should include forest-spanning probability, not just distance and rate; codes engineered to maximize that probability would make the post-BP stage unnecessary.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper introduces Noise Assisted Ensemble Decoding (NAED), a decoder for QLDPC codes that builds an ensemble of Tanner forests from the detector error model and performs exact inference on each forest by a linear-time dynamic program. The forest construction is driven by a syndrome- and channel-derived column score (Eq. 7), perturbed by ensemble-specific Gaussian noise (Eq. 10), with a residual-syndrome-aware variant (Eq. 15) for non-graph-like codes such as bivariate bicycle codes. Circuit-level simulation results are reported for rotated surface codes and BB codes, with claims of logical error rates comparable to or better than BP+OSD0 and of large per-round decoding speedups.

Significance. If correct, NAED would be a genuinely interesting decoder: exact inference on trees is linear-time, the ensemble is naturally parallel, and the Appendix B spanning argument for graph-like surface-code DEMs is standard and sound. The dynamic-programming derivation in Section III.D follows known tree-inference principles and is largely plausible. The paper is also honest about the main limitation, namely that syndrome-spanning forests cannot be guaranteed for general QLDPC codes. However, the central algorithm as written has a sign inconsistency that changes the optimized objective, and the empirical claims cannot be independently checked because no code or data are provided. These issues block acceptance in the current form.

major comments (3)
  1. [Algorithm 2; Sections III.B and III.D] Section III.B defines λ_q = -w_q and explicitly states that λ_q > 0 means flipping q increases the objective, so x_q=0 should be preferred, while λ_q < 0 means x_q=1 should be preferred. Consequently the per-column cost in Eq. (16) must be C_q = λ_q. Algorithm 2, lines 7 and 14, instead set the upward and root column costs to -λ_noise_q x_q, i.e. C_q = -λ_noise_q. Concretely, on a tree consisting of one check c with syndrome s_c=1 and two columns q1,q2 with λ_q1=1 and λ_q2=10, the objective derived from Section III.B chooses {q1=1,q2=0} at cost 1, whereas Algorithm 2 as written returns {q1=0,q2=1}, whose cost under the stated convention is 10. If the intended convention is that the variable λ_noise in Algorithm 2 is actually the perturbed weight n_b rather than the LLR -n_b, that contradicts the sentence after Eq. (10) and the call ETFI(F_b, s, n_b) in Algorithm 3. In either reading, the pseudocode does not implement the objective derived in Eqs. (7)-(9). Since every LER and timing claim is produced by this exact-inference routine, the numerical results must be re-derived under a consistent sign convention.
  2. [Section IV and Section VI] The central reproducibility concern: Section VI states that the NAED code will be released only 'in the near future' and that numerical data are available only on request. The main claims—LER competitiveness in Figs. 3, 5, 7, 8 and the orders-of-magnitude timing improvement in Figs. 4 and 6—are therefore not independently verifiable from the manuscript. The timing comparison also omits baseline implementation details, such as whether BP+OSD0 is single-threaded, which OSD library or variant is used, and whether the NAED time in Fig. 4 includes construction of all ten forests or stops at the first valid result. These details are essential because the advertised speed advantage is a central contribution.
  3. [Algorithm 1 and Algorithm 3] Algorithm 1 (CheckUpdate) does not define a result when a detector node c has no child column nodes but the required parity w = s_c ⊕ x_p equals 1: the loop over ch(c) is empty, the condition on line 8 is true, and line 9 takes the arg min over an empty set. This case is reachable for non-syndrome-spanning forests, which the paper explicitly allows for BB codes and which Algorithm 3 processes before the span check of line 7. The routine should propagate an infinite-cost message for infeasible parity; as written, the behavior is undefined for those inputs.
minor comments (4)
  1. [Tables I-IV] The 'Success percentage' in Tables I-IV does not state whether the denominator is all 10,000 sampled syndromes or only those not resolved by the previous decoding stage; Table V is clearer because it gives counts. Please state the denominator explicitly for each table.
  2. [Section III.B, Fig. 1] The text around Fig. 1 refers to a 'target detector pair' {s139, s185} without explaining how this pair is selected. The ensemble-member examples would be easier to follow if the selection criterion were given.
  3. [Algorithm 3, line 11] The condition 'If DEM matrix H is sparse' is ambiguous because the algorithm has just been run on both full and sparse DEMs. Please clarify what 'sparse' refers to and what the lifting operation does when the current DEM is the sparse one.
  4. [Section III.C and Fig. 5] The scalar parameters κ=0.5, τ=0.75, β=2.0 are stated for the BB-code experiments, but no sensitivity analysis or selection criterion is provided. Given the heuristic nature of Eq. (15), a brief sensitivity study would strengthen the empirical claims.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: NAED is benchmarked against external decoders, and its soft-weight formula is an explicit single-flip cost difference, not a fitted prediction.

full rationale

The derivation chain is self-contained. Equation (7)'s base weight w_q = -llr_q + αΣ(2s_c-1) is exactly the single-flip objective change f(x=0)-f(x') computed from the explicitly stated objective in Eq. (8), so the paper's 'rigorous interpretation' is a direct algebraic identity, not an assumed conclusion. The forest inference in Algorithm 2 and Eq. (16) is a standard tree dynamic program whose exactness does not depend on the authors' prior work. The only self-citation, Ref. [31], appears in a survey list of message-passing-dynamics papers and is not load-bearing. The free scalars (τ, κ, β, α) are empirical hyperparameters; the paper does not rename them as predictions against its own fitted data. Appendix B's syndrome-spanning claim for surface codes is an original path-spanning argument (a rejected weight-2 column's syndrome lies in the span of the retained forest path), not a self-imported uniqueness theorem, and the paper explicitly concedes that forests need not span syndromes for BB codes. One non-circular internal concern remains: Section IIIB's LLR convention says positive λ_q should discourage x_q=1, but Algorithm 2 line 7 uses -λ_noise_u x_u, opposite in sign to the cost +λ x_u; this is a correctness or consistency bug, not a circular reduction, because it makes the implemented optimization disagree with the stated objective rather than making any claimed result identical to its input by construction.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The ledger shows the central claim depends on hand-set scalars and on assumptions that certain heuristics rank columns well. No new physical entities are introduced. The main risk is not circularity but fragile parameter choice and unverified empirical performance.

free parameters (5)
  • alpha in Eq. (7) = not stated in main text
    Weight of the syndrome term in the base score; chosen by hand, no sensitivity analysis reported.
  • tau_b noise scaling range = {0, 0.5} uniformly across 11 ensemble members for surface codes
    Controls ensemble diversity; decoding accuracy depends on this range.
  • kappa, tau, beta in Eq. (15) = kappa=0.5, tau=0.75, beta=2.0
    Set in the Fig. 5 caption for BB code column weights; no justification or sensitivity study.
  • ensemble size B = 11 (surface), 100 (BB), 10 (timing benchmark)
    Larger ensembles improve syndrome-spanning chances but increase cost; chosen without scaling analysis.
  • min-sum BP normalization factor for BP stages = 1.0 and 0.625
    Reported in Tables I-IV; changes how many syndromes BP resolves and affects the multi-stage success numbers.
assumptions (4)
  • standard math Exact inference on a Tanner forest is achievable by one upward pass and traceback.
    Classic min-sum/two-pass property on trees; the paper invokes it in Section III D and Algorithm 2.
  • domain assumption Sparse DEM sparsification from Ref. [28] preserves the error mechanisms needed for decoding.
    NAED decodes on full and sparsified DEMs; Appendix B shows non-graph-like BB columns can be lost, so this is a load-bearing assumption.
  • ad hoc to paper The synthetic score in Eq. (7) ranks columns by usefulness for forest construction.
    Motivated by a one-step objective change, but no optimality proof; the decoder's accuracy depends on it.
  • domain assumption Surface-code sparse DEM columns are graph-like and weight at most 2.
    Used in Appendix B to guarantee syndrome-spanning forests; relies on the external sparsification routine preserving this property.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Highly Accurate Fast Decoding Framework for QLDPC codes Accelerated by Noise Perturbation and Ensemble Decoding." pith.science (2026). https://pith.science/paper/OD73L2LV

@misc{pith2026260809583,
  author       = {Pith},
  title        = {Pith review of: A Highly Accurate Fast Decoding Framework for QLDPC codes Accelerated by Noise Perturbation and Ensemble Decoding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OD73L2LV}},
  note         = {Machine review of arXiv:2608.09583}
}
abstract

A well-balanced decoder has been central to the development of modern fault-tolerant quantum computing. However, the inherent topologies of quantum error correcting codes can limit the performance of many well-studied decoding algorithms. In this work, we introduce Noise Assisted Ensemble Decoding (NAED), a highly accurate decoding framework with a significant advantage in real-time speed. NAED constructs an ensemble of Tanner forests, obtained as acyclic subgraphs of the original Tanner graph, and performs exact inference on each Tanner forest using a lightweight dynamic programming algorithm. The forest construction is guided by synthetic soft information derived jointly from the measured syndrome and channel statistics, with controlled noise perturbations generating diverse yet informative decoding matrix column orderings for the Tanner forest construction across the ensemble. Our benchmark results show that the proposed synthetic soft information-driven construction and inference on the Tanner forests can achieve improved or comparable decoding performances to the state-of-the-art decoding solutions, such as BP+OSD$0$, while also providing orders-of-magnitude improvements in per-round decoding speed under circuit-level noise.

Figures

Figures reproduced from arXiv: 2608.09583 by the authors.

Figure 1
Figure 1. FIG. 1. We show how different scaling factors for the applied noise can help an ensemble explore different Tanner paths [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. In this figure, we show how the dynamic gain term in Eq. ( [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. We show the circuit-level logical error rates of the rotated planar surface codes obtained under both the NAED [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: FIG. 4. We compare the mean per error correction round de [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. We show the circuit-level LERs of BB codes obtained under various decoding methods. We compare the standard [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. We compare the mean per error correction round [PITH_FULL_IMAGE:figures/full_fig_p016_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7. A circuit-level LER comparison between the two-stage NAED + post-BP and BP+OTF decoder with decimation [PITH_FULL_IMAGE:figures/full_fig_p019_7.png]
Figure 8
Figure 8. Figure 8: FIG. 8. A circuit-level LER comparison between the [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

74 extracted references · 49 canonical work pages

  1. [1]

    On the other hand,λnoise q =−n b q is used as the synthetic soft information for the infer- ence stage of the decoding

    For the ordered forest construction, we usenb q as the sorted priority weights. On the other hand,λnoise q =−n b q is used as the synthetic soft information for the infer- ence stage of the decoding. Less positive (more negative) λnoise q encourages prior selection of the corresponding col- umn (data qubit) nodeqcompared to the columns (data qubits) with ...

    work page

  2. [2]

    We now argue that wq isagoodalternativetotheBP-basedsoftinformation

    The second term serves as a heuristic reliability score based on the syndrome-driven single-column (data qubit) flip gain for column (data qubit)q. We now argue that wq isagoodalternativetotheBP-basedsoftinformation. Consider the decoding problem as an optimization of the cost objective: f(x) =α m∑ c=1 rc(x) + n∑ q=1 llrqxq,(8) whererc(x)istheresidualsynd...

    work page

  3. [3]

    Performing the upward pass on all the connected components of the forest opti- mizes Eq

    The Upward Message Passing The upward pass algorithm on a connected tree com- ponent essentially solves a smaller sub-tree optimization on the fly and later assists in the execution of a down- ward traceback of the optimal inference solution based on the decision at the root column (data qubit) node of the connected tree component. Performing the upward p...

    work page

  4. [4]

    The DP program essentially tracks the minimum cost child column (data qubit) assignments satisfy- ing a required parity

    work page

  5. [5]

    These parity assignment costs are up- dated on the fly whenever a new child column (data qubit) node assumes values from{0,1}

    We associate a dynamic parity cost with each pos- sible parity due to the child column (data qubit) assignments. These parity assignment costs are up- dated on the fly whenever a new child column (data qubit) node assumes values from{0,1}. Suppose, for a particular detector (check) nodec, the dy- namic parity cost after adding theith child column (data qu...

    work page

  6. [6]

    Therefore, the constraint on the parity of the child columns (data qubits) iss c⊕xq, wherex q is the bit-value of the parent column (data qubit) node of detector (check)c

    The target parity of all the neighbor columns (data qubits) of a particular detector (check) nodecis its syndrome values c. Therefore, the constraint on the parity of the child columns (data qubits) iss c⊕xq, wherex q is the bit-value of the parent column (data qubit) node of detector (check)c

    work page

  7. [7]

    The message from childcto parentqis the mini- mum cost of child column (data qubit) assignments that satisfy the required parity, i.e. Mc→q(xq) =D|ch(c)|[sc⊕xq], which is already the minimum cost for the re- quired parity assignment of the child columns (data qubits) ensured by the dynamic least cost assign- ment of Eq. (20) and Eq. (21)

    work page

  8. [8]

    This parity dynamic program process each child column (data qubit) node once and performs a constant number ofΘ(|N(c)|)operations per detector (check) node

    While we dynamically minimize the cost of child column (data qubit) node assignments based on the syndrome bit constraint, the column (data qubit) node assignments should also be maintained on the fly, which gives the minimum cost for the dynamic parity at each step. This parity dynamic program process each child column (data qubit) node once and performs...

    work page

Show all 74 references
  1. [9]

    The Downward Traceback procedure The upward message pass is followed by a downward traceback to complete the exact inference algorithm. It 10 Algorithm 1:Dynamic Parity Program at detector (check) nodec 1FunctionCheckUpdate(): Input :Detector (check) nodec, syndrome bits c, an...

  2. [10]

    The key intuition of this technique is a parallel implementation of multiple decoding instances and net pool a decoding decision from this ensemble

    Ensemble decoding Ensembling is very efficient for strategizing accurate decoding outcomes when individual decoding attempts 11 might have shortcomings [57]. The key intuition of this technique is a parallel implementation of multiple decoding instances and net pool a decoding...

  3. [11]

    D. J. MacKay, G. Mitchison, and P. L. McFadden, Sparse-graph codes for quantum error correction, IEEE Transactions on Information Theory50, 2315 (2004)

    2004

  4. [12]

    The LER comparisons for both decoders have been shown in Fig. 7. Note that the BP+OTF decoder here uses a decimation strategy in the final OTF+BP round. We assign a decimation parameter of10−9 similar to the Ref. [28]. The decimation strategy assigns a very small probability t...

  5. [13]

    N. P. Breuckmann and J. N. Eberhardt, Quantum low- density parity-check codes, PRX quantum2, 040101 (2021)

    2021

  6. [14]

    Vasic, V

    B. Vasic, V. Savin, M. Pacenti, S. Borah, N. Raveen- dran,et al., Quantum low-density parity-check codes, IEEE BITS the Information Theory Maga- zine 10.1109/MBITS.2026.3713965 (2026)

    2026

  7. [15]

    Panteleev and G

    P. Panteleev and G. Kalachev, Quantum ldpc codes with almost linear minimum distance, IEEE Transactions on Information Theory68, 213 (2021)

    2021

  8. [16]

    Panteleev and G

    P. Panteleev and G. Kalachev, Asymptotically good quantum and locally testable classical ldpc codes, inPro- ceedings of the 54th annual ACM SIGACT symposium on theory of computing(2022) pp. 375–388

    2022

  9. [17]

    Leverrier and G

    A. Leverrier and G. Zémor, Quantum tanner codes, in 2022 IEEE 63rd Annual Symposium on Foundations of Computer Science (FOCS)(IEEE, 2022) pp. 872–883

    2022

  10. [18]

    Gottesman, Fault-tolerant quantum computation with constant overhead, Quantum Info

    D. Gottesman, Fault-tolerant quantum computation with constant overhead, Quantum Info. Comput.14, 1338–1372 (2014)

    2014

  11. [19]

    Bravyi, A

    S. Bravyi, A. W. Cross, J. M. Gambetta, D. Maslov, P. Rall, and T. J. Yoder, High-threshold and low- overhead fault-tolerant quantum memory, Nature627, 778 (2024)

    2024

  12. [20]

    Q. Xu, J. P. Bonilla Ataides, C. A. Pattison, N. Raveen- dran, D. Bluvstein, J. Wurtz, B. Vasić, M. D. Lukin, L. Jiang, and H. Zhou, Constant-overhead fault-tolerant quantum computation with reconfigurable atom arrays, Nature Physics20, 1084 (2024)

    2024

  13. [21]

    Strikis and L

    A. Strikis and L. Berent, Quantum low-density parity- check codes for modular architectures, PRX Quantum4, 020321 (2023)

    2023

  14. [22]

    Berthusen, D

    N. Berthusen, D. Devulapalli, E. Schoute, A. M. Childs, M. J. Gullans, A. V. Gorshkov, and D. Gottesman, To- ward a 2d local implementation of quantum low-density parity-check codes, PRX Quantum6, 010306 (2025)

    2025

  15. [23]

    Panteleev and G

    P. Panteleev and G. Kalachev, Degenerate quantum ldpc codes with good finite length performance, Quantum5, 585 (2021)

    2021

  16. [24]

    A. Y. Kitaev, Fault-tolerant quantum computation by anyons, Annals of physics303, 2 (2003)

    2003

  17. [25]

    Dennis, A

    E. Dennis, A. Kitaev, A. Landahl, and J. Preskill, Topological quantum memory, Journal of Mathematical Physics43, 4452 (2002)

    2002

  18. [26]

    A. G. Fowler, A. M. Stephens, and P. Groszkowski, High- threshold universal quantum computation on the surface 21 code, Physical Review A—Atomic, Molecular, and Opti- cal Physics80, 052312 (2009)

    2009

  19. [27]

    Bravyi, D

    S. Bravyi, D. Poulin, and B. Terhal, Tradeoffs for reliable quantum information storage in 2d systems, Physical re- view letters104, 050503 (2010)

    2010

  20. [28]

    We use the same sparsification procedures from Ref

    the authors propose a sparsification routine, which constructs a sparse DEM matrix with bounded column weights. We use the same sparsification procedures from Ref. [28] for the surface and BB codes. Essentially, the sparsification tries to capture the sparse nature of the unde...

    2024

  21. [29]

    Tillich and G

    J.-P. Tillich and G. Zémor, Quantum ldpc codes with positive rate and minimum distance proportional to the square root of the blocklength, IEEE Transactions on Information Theory60, 1193 (2013)

    2013

  22. [30]

    Mandelbaum, J

    R. Mandelbaum, J. Gambetta, J. Chow, T. Mittal, T. J. Yoder, A. Cross, and M. Steffen, How ibm will build the world’s first large-scale, fault-tolerant quantum com- puter, IBM Research Blog (2025)

    2025

  23. [31]

    T.RichardsonandS.Kudekar,Designoflow-densitypar- ity check codes for 5g new radio, IEEE Communications Magazine56, 28 (2018)

    2018

  24. [32]

    Gallager, Low-density parity-check codes, IRE Trans- actions on information theory8, 21 (1962)

    R. Gallager, Low-density parity-check codes, IRE Trans- actions on information theory8, 21 (1962)

    1962

  25. [33]

    D. J. MacKay and R. M. Neal, Near shannon limit per- formance of low density parity check codes, Electronics letters32, 1645 (1996)

    1996

  26. [34]

    F. R. Kschischang, B. J. Frey, and H.-A. Loeliger, Factor graphs and the sum-product algorithm, IEEE Transac- tions on information theory47, 498 (2001)

    2001

  27. [35]

    Roffe, D

    J. Roffe, D. R. White, S. Burton, and E. Campbell, De- coding across the quantum low-density parity-check code landscape, Physical Review Research2, 043423 (2020)

    2020

  28. [36]

    Poulin and Y

    D. Poulin and Y. Chung, On the iterative decoding of sparse quantum codes, arXiv preprint arXiv:0801.1241 10.48550/arXiv.0801.1241 (2008)

    2008 doi arXiv

  29. [37]

    Fuentes, J

    P. Fuentes, J. E. Martinez, P. M. Crespo, and J. Garcia- Frías, Degeneracy and its impact on the decoding of sparse quantum codes, IEEE Access9, 89093 (2021)

    2021

  30. [38]

    Raveendran and B

    N. Raveendran and B. Vasić, Trapping sets of quantum ldpc codes, Quantum5, 562 (2021)

    2021

  31. [39]

    Hillmann, L

    T. Hillmann, L. Berent, A. O. Quintavalle, J. Eisert, R. Wille, and J. Roffe, Localized statistics decoding: A parallel decoding algorithm for quantum low-density parity-check codes, arXiv preprint arXiv:2406.1865510, 10.1038/s41467-025-63214-7 (2024)

    2024

  32. [40]

    deMarti iOlius, I

    A. deMarti iOlius, I. Etxezarreta Martinez, J. Roffe, and J. Etxezarreta Martinez, An almost-linear time de- coding algorithm for quantum ldpc codes under circuit- level noise, npj Quantum Information 10.1038/s41534- 026-01292-1 (2026)

    2026 doi

  33. [41]

    Wolanski and B

    S. Wolanski and B. Barber, Ambiguity clustering: an accurate and efficient decoder for qldpc codes, arXiv preprint arXiv:2406.14527 10.48550/arXiv.2406.14527 (2024)

    2024 doi arXiv

  34. [42]

    Du Crest, M

    J. Du Crest, M. Mhalla, and V. Savin, Stabilizer inactiva- tionformessage-passingdecodingofquantumldpccodes, in2022 IEEE Information Theory Workshop (ITW) (IEEE, 2022) pp. 488–493

    2022

  35. [43]

    Bhattacharyya and A

    M. Bhattacharyya and A. Raina, Decoding quantum ldpc codes using collaborative check node removal, arXiv preprint arXiv:2501.08036 10.48550/arXiv.2501.08036 (2025)

    2025 doi arXiv

  36. [44]

    K. Yin, X. Fang, J. Ruan, H. Zhang, D. Tullsen, A. Sorn- borger, C. Liu, A. Li, T. Humble, and Y. Ding, Sym- break: mitigating quantum degeneracy issues in qldpc code decoders by breaking symmetry, arXiv preprint arXiv:2412.02885 10.48550/arXiv.2412.02885 (2024)

    2024 doi arXiv

  37. [45]

    Chytas, N

    D. Chytas, N. Raveendran, and B. Vasic, Enhanced min-sum decoding of quantum codes using previous iteration dynamics, arXiv preprint arXiv:2501.05021 10.48550/arXiv.2501.05021 (2025)

    2025 doi arXiv

  38. [46]

    Bhatnagar, M

    S. Bhatnagar, M. Pacenti, N. Raveendran, D. Declercq, and B. Vasić, Impulse decoding of quantum ldpc codes: Equivalence of degeneracy and code-shortening, arXiv preprint arXiv:2606.18240 10.48550/arXiv.2606.18240 (2026)

    2026 doi arXiv

  39. [47]

    Chytas, N

    D. Chytas, N. Raveendran, and B. Vasić, Collec- tive bit flipping-based decoding of quantum ldpc codes, IEEE Transactions on Communications 10.1109/TCOMM.2025.3535897 (2025)

    2025

  40. [48]

    A. K. Pradhan, N. Raveendran, N. Rengaswamy, X. Xiao, and B. Vasić, Learning to decode trapping sets in qldpc codes, in2023 12th International Symposium on Topics in Coding (ISTC)(IEEE, 2023) pp. 1–5

    2023

  41. [49]

    Ninkovic, O

    V. Ninkovic, O. Kundacina, D. Vukobratovic, C. Häger, and A. G. i Amat, Decoding quantum ldpc codes us- ing graph neural networks, inGLOBECOM 2024-2024 IEEE Global Communications Conference(IEEE, 2024) pp. 3479–3484

    2024

  42. [50]

    A. S. Maan and A. Paler, Machine learning message- passing for the scalable decoding of qldpc codes, npj Quantum Information11, 78 (2025)

    2025

  43. [51]

    J. Blue, H. Avlani, Z. He, L. Ziyin, and I. L. Chuang, Machine learning decoding of circuit-level noise for bi- variate bicycle codes, arXiv preprint arXiv:2504.13043 10.22331/q-2026-06-30-2149 (2025)

    2025 arXiv

  44. [52]

    Gottesman,Stabilizer codes and quantum error cor- rection(California Institute of Technology, 1997)

    D. Gottesman,Stabilizer codes and quantum error cor- rection(California Institute of Technology, 1997)

    1997

  45. [53]

    A. R. Calderbank and P. W. Shor, Good quantum error-correcting codes exist, Physical Review A54, 1098 (1996)

    1996

  46. [54]

    Series A: Mathematical, Physical and Engineering Sciences452, 2551 (1996)

    A.Steane,Multiple-particleinterferenceandquantumer- ror correction, Proceedings of the Royal Society of Lon- don. Series A: Mathematical, Physical and Engineering Sciences452, 2551 (1996)

    1996

  47. [55]

    Poulin, Optimal and efficient decoding of concate- natedquantumblockcodes,PhysicalReviewA—Atomic, Molecular, and Optical Physics74, 052333 (2006)

    D. Poulin, Optimal and efficient decoding of concate- natedquantumblockcodes,PhysicalReviewA—Atomic, Molecular, and Optical Physics74, 052333 (2006)

    2006

  48. [56]

    P.IyerandD.Poulin,Hardnessofdecodingquantumsta- bilizer codes, IEEE Transactions on Information Theory 61, 5209 (2015)

    2015

  49. [57]

    A.deMartiiOlius, P.Fuentes, R.Orús, P.M.Crespo,and J. E. Martinez, Decoding algorithms for surface codes, Quantum8, 1498 (2024)

    2024

  50. [58]

    A. Gong, S. Cammerer, and J. M. Renes, Toward low-latency iterative decoding of qldpc codes un- der circuit-level noise, arXiv preprint arXiv:2403.18901 10.48550/arXiv.2403.18901 (2024)

    2024 doi arXiv

  51. [59]

    G. D. Forney, The viterbi algorithm, Proceedings of the IEEE61, 268 (2005)

    2005

  52. [60]

    Ollivier and J.-P

    H. Ollivier and J.-P. Tillich, Trellises for stabilizer codes: Definition and uses, Physical Review A—Atomic, Molec- ular, and Optical Physics74, 032304 (2006)

    2006

  53. [61]

    H.-L.Lou,Implementingtheviterbialgorithm,IEEESig- nal processing magazine12, 42 (2002)

    2002

  54. [62]

    F. R. Kschischang and B. J. Frey, Iterative decoding of compound codes by probability propagation in graphical models, IEEE Journal on Selected Areas in Communica- tions16, 219 (2002). 22

    2002

  55. [63]

    Pearl,Probabilistic reasoning in intelligent systems: networks of plausible inference(Elsevier, 2014)

    J. Pearl,Probabilistic reasoning in intelligent systems: networks of plausible inference(Elsevier, 2014)

    2014

  56. [64]

    M. J. Wainwright, T. S. Jaakkola, and A. S. Willsky, Tree-based reparameterization framework for analysis of sum-product and related algorithms, IEEE Transactions on information theory49, 1120 (2003)

    2003

  57. [65]

    Du Crest, F

    J. Du Crest, F. Garcia-Herrero, M. Mhalla, V. Savin, and J. Valls, Check-agnosia based post-processor for message- passing decoding of quantum ldpc codes, Quantum8, 1334 (2024)

    2024

  58. [66]

    A. D. Kumar and A. Dukkipati, A two stage selective av- eraging ldpc decoding, in2012 IEEE International Sym- posium on Information Theory Proceedings(IEEE, 2012) pp. 2866–2870

    2012

  59. [67]

    J. B. Kruskal, On the shortest spanning subtree of a graph and the traveling salesman problem, Proceedings of the American Mathematical society7, 48 (1956)

    1956

  60. [68]

    Tarjan, Depth-first search and linear graph algo- rithms, SIAM journal on computing1, 146 (1972)

    R. Tarjan, Depth-first search and linear graph algo- rithms, SIAM journal on computing1, 146 (1972)

    1972

  61. [69]

    Shutty, M

    N. Shutty, M. Newman, and B. Villalonga, Efficient near- optimal decoding of the surface code through ensembling, Physical Review Letters136, 070603 (2026)

    2026

  62. [70]

    C.Gidney,Stim: afaststabilizercircuitsimulator,Quan- tum5, 497 (2021)

    2021

  63. [71]

    P.-J. H. Derks, A. Townsend-Teague, A. G. Burchards, and J. Eisert, Designing fault-tolerant circuits using de- tector error models, Quantum9, 1905 (2025)

    2025

  64. [72]

    Toshio, K

    R. Toshio, K. Kishi, J. Fujisaki, H. Oshima, S. Sato, and K. Fujii, Decoder switching: Break- ing the speed-accuracy tradeoff in real-time quan- tum error correction, arXiv preprint arXiv:2510.25222 10.48550/arXiv.2510.25222 (2025)

    2025 doi

  65. [73]

    G. P. Gehér, C. McLauchlan, E. T. Campbell, A. E. Moylett, and O. Crawford, Error-corrected hadamard gate simulated at the circuit level, Quantum8, 1394 (2024)

    2024

  66. [74]

    Higgott and C

    O. Higgott and C. Gidney, Sparse blossom: correcting a million errors per core second with minimum-weight matching, Quantum9, 1600 (2025)

    2025

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.