Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Quantum XYZ cyclic codes for biased noise

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read This paper proposes quantum XYZ cyclic codes whose code distance grows with code length, giving 50% pure-Pauli thresholds, about 49% under Z-biased noise, and lower physical overhead than XZZX surface codes at matched distance.

desk verdict New cyclic code family with clean partial proofs, but the headline distance-growth claim rests on unproved lower bounds and Monte Carlo upper bounds—the paper's own open questions confirm this. read the letter →

arxiv 2501.16827 v1 pith:WAQZOLQU submitted 2025-01-28 quant-ph

Zhipeng Liang , Fusheng Yang , Zhengzhong Yi , Xuan Wang This is my paper · ORCID
classification quant-ph MSC 81P7094B15 PACS 03.67.Pp
keywords quantumXYZcycliccodesbiasednoiseerrorcorrectioncodedistancerepetitionstructureXZZXsurfaceFDBP-OSDdecodingcode-capacitythreshold
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 proposes a family of stabilizer codes it calls quantum XYZ cyclic codes, generated by cyclic shifts of one weight-six Pauli string, and argues that these are the only quantum cyclic codes known whose code distance grows with code length. If that is right, the family matters for biased-noise hardware: under pure Pauli X, Y, or Z noise the codes behave like repetition codes and show 50% code-capacity thresholds, under Z-biased noise with bias ratio $\eta_Z=1000$ they show about 49%, and under depolarizing noise about 13%. The paper also claims that at matched code distance these codes need fewer physical qubits than XZZX surface codes. A sympathetic reader would care because the construction is a concrete route to reducing qubit overhead in architectures where Z errors dominate.

What carries the argument

The central object is the quantum XYZ cyclic code $C(a,b)$, whose stabilizer generators are all cyclic shifts of the weight-six string $X Z Y Y Z X$, with $b$ identities between $X$ and $Z$, $a$ identities between $Z$ and $Y$, and one identity between the two $Y$'s. The argument runs through circulant matrices over the binary field and their polynomial representatives: $H_x = 1+x^{a+b+2}+x^{a+b+4}+x^{2(a+b)+6}$, $H_z = x^{b+1}+x^{a+b+2}+x^{a+b+4}+x^{2a+b+5}$, and $H_y = H_x + H_z$. Each polynomial factors into pairs such as $(1+x^{a+b+2})(1+x^{a+b+4})$; when the shifted exponents are coprime to $x^N+1$, the rank drops by one and the code behaves as a repetition code under that Pauli type. Lemma 1 then shows that selecting stabilizer generators at regular intervals produces structured products, which is what allows the exact minimum weight for logical X and the upper bounds for logical Y.

What would settle it

For a small code in Table I, such as $C(13,2)$ of length $37$ with claimed distance $7$, exhaustively enumerate all logical operators of weight below $7$, or solve the minimum-distance problem by integer programming; finding any logical operator of weight $6$ or less would show the true distance is smaller, and repeating this at several increasing $a$ values with fixed $b$ would directly test whether the true distance actually grows.

Watch

Extended reading notes

Core claim

The central claim is that for suitable natural numbers $a$ and $b$, the quantum XYZ cyclic code $C(a,b)$ of length $N=2(a+b)+7$ encodes one logical qubit and has code distance that increases with $N$. The paper proves three supporting pieces: under coprimality conditions on $a$, $b$, and $N$, the code has repetition-code structure under each of the three pure Pauli noises; for the parameter choice $a=2l(b+2)+l-1$, the minimum weight of the logical X operator is exactly $2l+1$, which grows linearly with $N$; and the logical Y operator weight is bounded above by $2b+5$, or $2b+3$ when $b=3l-1$. For the logical Z operator, an exact proof is left open and the paper instead reports Monte Carlo estimates that also grow with $a$. Putting these together, the paper concludes that increasing $b$ increases the code distance, giving thresholds of 50% for pure Pauli noise, about 49% for Z-biased noise at $\eta_Z=1000$, and about 13% for depolarizing noise, with lower physical-qubit overhead than XZZX surface codes at matched distance.

Load-bearing premise

The distance-growth conclusion rests on the assumption that the values reported as $d_X$, $d_Y$, and $d_Z$ are the true minimum logical-operator weights; the paper proves an exact minimum only for the $X$ operator in a special parameter sequence, gives only an upper bound for $Y$, and estimates $Z$ with a decoder that is not guaranteed to find the true minimum.

Editorial extensions

If this is right

  • The family would be the first quantum cyclic code family whose code distance grows with code length, a property previously open even for XZZX cyclic codes.
  • Under pure Pauli noise the repetition-code structure makes the logical error probability decay with code length, consistent with the simulated 50% thresholds.
  • At matched code distance, the reported physical-qubit overhead is much lower than that of XZZX surface codes, so the codes could reduce hardware requirements in biased-noise architectures.
  • The parameter table gives concrete "optimal" codes for $b=0$ through $11$ with distances from $5$ to $27$, providing a starting point for implementation or further study.
  • Supplying the missing lower bound for the Z logical operator would turn the numerical distance-growth evidence into a theorem.

Reading between the lines

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

  • Beyond the paper: the distance-growth claim would become fully rigorous only if the open Z-logical lower bound and a matching Y lower bound were proved; until then the family is numerically, not provably, distance-growing.
  • Beyond the paper: the overhead comparison at matched code distance does not by itself guarantee a lower logical error rate at fixed physical overhead; circuit-level simulations with noisy syndrome extraction would test the practical gain.
  • Beyond the paper: the same circulant-polynomial machinery could be applied to other weight-six generator strings to screen for cyclic codes with repetition structure under biased noise.
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 / 5 minor

Summary. The manuscript introduces a family of [[N,1,d]] quantum XYZ cyclic codes C(a,b), with N = 2(a+b)+7 and stabilizer generators obtained by cyclic shifts of a weight-six XZYYZX-type operator. It proves that the stabilizer group is Abelian (Prop. 1), derives a code-dimension criterion (Prop. 2), gives conditions for repetition-code structure under pure X, Y, or Z noise (Prop. 3), derives an exact logical-X weight for a special parameter subsequence (Prop. 4), and derives an upper bound on logical-Y weight (Prop. 5). It then uses FDBP-OSD Monte Carlo simulations to estimate logical-operator weights and reports code-capacity thresholds of about 50% for pure Pauli noise, about 13% for depolarizing noise, and about 49% for Z-biased noise (eta_Z = 1000), together with a physical-overhead comparison against the XZZX surface code. The paper claims that this is the only known family of quantum cyclic codes whose code distance increases with code length.

Significance. If fully substantiated, the code family would be of interest: it combines a cyclic stabilizer construction with repetition-code behavior under biased Pauli noise, and the reported thresholds would be competitive for biased-noise architectures. The proof of repetition structure (Prop. 3) is a genuine strength, and the 50% thresholds follow cleanly from that structure. The 13% and 49% thresholds are simulation results and require full simulation details for evaluation. However, the paper's central novelty claim—that the true code distance grows with code length—is not established by the present arguments. The obstacles are not minor presentation issues but missing lower bounds on logical Y and Z weights, explicitly acknowledged by the authors in Section V.

major comments (3)
  1. [III C; Propositions 4-5; Section V] The central claim that d(C(a,b)) grows with code length is not proved. For a stabilizer code, d = min(dX, dY, dZ). Proposition 4 establishes dX = 2l+1 only for the special subsequence a = 2l(b+2)+l-1. Proposition 5 (Eq. (33)) explicitly provides only an upper bound on dY. No lower bound for dZ is given, and no proof excludes logical-Y operators below the upper bound; Section V lists both as open questions. The sentence in Section III C stating that these results 'indicates the code distance ... can increase with code length by increasing b' is therefore not a logical consequence of the proved statements.
  2. [II D; Algorithm 1; Table I; Fig. 1] The dY and dZ values in Table I and Fig. 1 are upper bounds generated by a heuristic Monte Carlo procedure, not certified minimum distances. Section II D explicitly describes Algorithm 1 as a method for determining an upper bound on code distance, and FDBP-OSD is not a maximum-likelihood decoder. Thus the 'optimal code' definition in Section IV B, which assumes distance equal to 2b+5 or 2b+3, and the overhead comparison in Fig. 2 rest on unproven distance values. If there exists a logical Y or Z operator of weight smaller than the tabled value, the true distance is smaller than reported and the growth and overhead conclusions may fail.
  3. [IV A; Fig. 1] The numerical evidence in Fig. 1 shows that for fixed b the dY estimate saturates at a constant (2b+5 for b=3) while dX and dZ estimates grow with a. Therefore the code distance cannot grow merely by increasing a; the growth claim necessarily relies on increasing b as well. But in that b-growth regime, the missing lower bounds for Y and Z described in the previous comments are exactly the needed ingredients. The numerical upper bounds do not fill that proof gap.
minor comments (5)
  1. [Section I] The reference placeholder '[refs]' in the first paragraph is unresolved and should be completed.
  2. [IV B; Table I] The procedure for selecting the 'optimal' a for each b is not reproducible as stated: the number of Monte Carlo trials T, FDBP-OSD parameters, and stopping criteria are not specified.
  3. [IV B; Fig. 2] The XZZX surface-code data used in the overhead comparison should be specified in detail, including code sizes, distances, and how those distances are computed, so that the comparison is fair.
  4. [III C; IV B; Corollary 1] The text contains several grammatical errors, including 'These results indicates' (Section III C), 'simultaneusly' (Section IV B), and 'don't have' in Corollary 1; these should be corrected.
  5. [Abstract; Section V] The claim that this is the only family of quantum cyclic codes with code distance increasing with code length should either be supported by a systematic literature comparison or be softened, given that the present evidence for distance growth is partly numerical and partly based on upper bounds.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation: the central distance-growth claim rests on unproved lower bounds and heuristic upper-bound estimates, which is a verification gap rather than a circular reduction.

full rationale

The derivation chain is not circular in the load-bearing sense. Proposition 3 proves repetition-code structure from explicit gcd conditions; Proposition 4 gives an exact X-logical weight for a special parameter sequence; and Proposition 5 explicitly labels Eq. (33) as an upper bound on the Y-logical weight. The 50% thresholds follow from the proved repetition structure, while the 13% and 49% thresholds are simulation outputs, not inputs. The main self-referential element is that the dZ and dY entries in Table I and Fig. 1 come from Algorithm 1, which the paper itself describes in Sect. II D as returning the upper bound of the minimum weight of three logical operators; these values are then treated as exact distances when selecting optimal codes and when comparing overhead in Fig. 2. Sect. V explicitly lists as open questions how to strictly prove that the minimum weight of the logical Z operator increases with code length and how to prove that no logical Y operator has weight below dY^up. This is an evidentiary gap rather than a circular construction: no equation is defined in terms of its own output, no fitted parameter is renamed as a prediction, and the self-citations (Refs. [9] and [11]) supply computational tools rather than the theorem being asserted.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claims rest on two unproved background pillars: the rank formula taken from Panteleev and Kalachev, and the quality of the authors' own FDBP-OSD decoder as a maximum-likelihood proxy. In addition, the optimal code parameters are hand-selected by numerical search, and the Monte Carlo distance values are upper bounds by the paper's own definition.

free parameters (2)
  • a and b code parameters = Table I: e.g., b=3, a=20, N=53
    The construction is parameterized by two natural numbers; the 'optimal' codes used for threshold and overhead claims are selected by numerical search to satisfy repetition structure and target distances, so performance numbers are not parameter-free.
  • Monte Carlo trials T and FDBP-OSD settings = not stated in paper
    Algorithm 1 depends on T and decoder hyperparameters; without these, the distance upper bounds and thresholds cannot be reproduced.
assumptions (3)
  • domain assumption Rank formula for circulant stabilizer codes from Proposition 1 of Ref. [20]: k = deg gcd(A(x), B(x), x^N+1)
    Invoked in Proposition 2 without proof; it is a result from prior literature and is load-bearing for the claimed code dimensions.
  • domain assumption FDBP-OSD decoder approximates maximum-likelihood decoding closely enough for Monte Carlo distance and threshold estimates
    Algorithm 1 and all threshold simulations rely on this; the paper does not prove optimality or provide error bars.
  • domain assumption Perfect syndrome measurement
    Stated in Sect. I: simulations assume perfect syndrome measurement, so thresholds are code-capacity bounds, not fault-tolerant thresholds.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Quantum XYZ cyclic codes for biased noise." pith.science (2026). https://pith.science/paper/WAQZOLQU

@misc{pith2026250116827,
  author       = {Pith},
  title        = {Pith review of: Quantum XYZ cyclic codes for biased noise},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WAQZOLQU}},
  note         = {Machine review of arXiv:2501.16827}
}
abstract

In some quantum computing architectures, Pauli noise is highly biased. Tailoring Quantum error-correcting codes to the biased noise may benefit reducing the physical qubit overhead without reducing the logical error rate. In this paper, we propose a family of quantum XYZ cyclic codes, which are the only one family of quantum cyclic codes with code distance increasing with code length to our best knowledge and have good error-correcting performance against biased noise. Our simulation results show that the quantum XYZ cyclic codes have $50\%$ code-capacity thresholds for all three types of pure Pauli noise and around $13\%$ code-capacity threshold for depolarizing noise. In the finite-bias regime, when the noise is biased towards Pauli $Z$ errors with noise bias ratios $\eta_Z=1000$, the corresponding code-capacity threshold is around $49\%$. Besides, we show that to reach the same code distance, the physical qubit overhead of XYZ cyclic code is much less than that of the XZZX surface code.

Figures

Figures reproduced from arXiv: 2501.16827 by the authors.

Figure 1
Figure 1. FIG. 1. The minimum weight of logical (a) [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. The physical qubits overhead of XYZ cyclic codes [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. The error-correcting performance of the quantum XYZ cyclic codes against depolarizing noise. (a) [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: FIG. 4. The error-correcting performance of the quantum XYZ cyclic codes against infinitely biased Pauli noise. (a) [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. The error-correcting performance of the quantum XYZ cyclic codes against Pauli Z biased noise with biased rate [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Romanesco codes: Bias-tailored qLDPC codes from fractal codes

    quant-ph 2025-05 conditional novelty 6.0 of 10

    Romanesco codes are Clifford-deformed bivariate bicycle codes made from two cellular automaton codes that achieve an effective distance scaling linearly with qubit count in the infinite-bias limit.

Reference graph

Works this paper leans on

26 extracted references · 19 canonical work pages · cited by 1 Pith paper

  1. [1]

    C(a, b) should simultaneously has repetition code structure under three types of pure Pauli noise

    work page

  2. [2]

    If b = 3l − 1, k= 3, if (a + 1) mod 3 = 0 1, otherwise

    work page

  3. [3]

    If b = 3l − 2, k= 3, if a mod3 = 0 1, otherwise (17) where l = 1, 2, · · ·. Proof. According to the proof of Proposition 1 in Ref. [20], the rank of the parity-check matrix H = (Hx | Hz) of C(a, b) is 2 (a + b)+7 −deg[gcd(Ax, Bx, x2(a+b)+7 +1], where deg[ f (x)] is the degree of f (x), and A (x) = 1 + xa+b+2 +xa+b+4 +x2(a+b)+6 and B (x) = xb+1 +xa+b+2 + 5...

    work page

  4. [4]

    The code distance of C(a, b) is 2b + 5 (b mod3 ̸= 2) or 2b + 3 (b mod3 = 2). 8 FIG. 3. The error-correcting performance of the quantum XYZ cyclic codes against depolarizing noise. (a) b mod3 = 0, (b) b mod3 = 1, (c) b mod3 = 2. FIG. 4. The error-correcting performance of the quantum XYZ cyclic codes against infinitely biased Pauli noise. (a)b mod3 = 0, in...

    work page

  5. [5]

    Base case When l = 1 and a = 2 ( b + 2), the code length N = 3[2 ( b + 2) + 1], and it can be veri- fied that any operator of weight 1 or 2 must not be logical operator, since these operators must anti- commute with some stabilizers. Thus, when l = 1, I · · ·I| {z } b+2 X I · · ·I| {z } b+2 I · · ·I| {z } b+2 X I · · ·I| {z } b+2 I · · ·I| {z } b+2 X I · ...

    work page

  6. [6]

    Inductive Step When l = m, assume the stabilizer ˆSm = X · · ·X| {z } b+2 I X · · ·X| {z } b+2 | {z } · · ·X · · ·X| {z } b+2 I X · · ·X| {z } b+2 | {z }| {z } 2m+1 contains the most single-qubit Pauli X operators. According to Lemma 1, there exists a binary row vector y m of size 2 (b + 2) + 1 such that V ec ˆSm = (y m, · · ·, y m | {z } 2m+1 )H m = (y m...

    work page

  7. [7]

    P. W. Shor, Scheme for reducing decoherence in quantum computer memory, Phys. Rev. A 52, R2493 (1995)

    1995

  8. [8]

    A. M. Steane, Error correcting codes in quantum theory, Phys. Rev. Lett. 77, 793 (1996)

    1996

Show all 26 references
  1. [9]

    Aliferis, F

    P. Aliferis, F. Brito, D. P. DiVincenzo, J. Preskill, M. Steffen, and B. M. Terhal, Fault-tolerant computing with biased-noise superconducting qubits: a case study, New Journal of Physics 11, 013061 (2009)

    2009

  2. [10]

    Lescanne, M

    R. Lescanne, M. Villiers, T. Peronnin, A. Sarlette, M. Delbecq, B. Huard, T. Kontos, M. Mirrahimi, and Z. Leghtas, Exponential suppression of bit-flips in a qubit encoded in an oscillator, Nature Physics 16, 509 (2020)

    2020

  3. [11]

    Chamberland, K

    C. Chamberland, K. Noh, P. Arrangoiz-Arriola, E. T. Campbell, C. T. Hann, J. Iverson, H. Putterman, T. C. Bohdanowicz, S. T. Flammia, A. Keller, et al., Building a fault-tolerant quantum computer using concatenated cat codes, PRX Quantum 3, 010329 (2022)

    2022

  4. [12]

    M. D. Shulman, O. E. Dial, S. P. Harvey, H. Bluhm, V. Umansky, and A. Yacoby, Demonstration of entan- glement of electrostatically coupled singlet-triplet qubits, science 336, 202 (2012)

    2012

  5. [13]

    D. Nigg, M. Mueller, E. A. Martinez, P. Schindler, M. Hennrich, T. Monz, M. A. Martin-Delgado, and R. Blatt, Quantum computations on a topologically en- coded qubit, Science 345, 302 (2014)

    2014

  6. [14]

    J. P. Bonilla Ataides, D. K. Tuckett, S. D. Bartlett, S. T. Flammia, and B. J. Brown, The xzzx surface code, Na- ture communications 12, 2172 (2021)

    2021

  7. [15]

    Liang, Z

    Z. Liang, Z. Wang, Z. Yi, Y. Wu, C. Qiu, and X. Wang, Determining the upper bound of code distance of quantum stabilizer codes through monte carlo method based on fully decoupled belief propagation, (2024), arXiv:2402.06481 [quant-ph]

    2024 arXiv

  8. [16]

    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, Nature 627, 778 (2024)

    2024

  9. [17]

    Z. Yi, Z. Liang, K. Zhong, Y. Wu, Z. Fang, and X. Wang, Improved belief propagation decoding algorithm based on decoupling representation of pauli operators for quantum ldpc codes, arXiv preprint arXiv:2305.17505 (2023)

    2023 arXiv

  10. [18]

    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

  11. [19]

    Laflamme, C

    R. Laflamme, C. Miquel, J. P. Paz, and W. H. Zurek, Perfect quantum error correcting code, Physical Review Letters 77, 198 (1996)

    1996

  12. [20]

    Q. Xu, N. Mannucci, A. Seif, A. Kubica, S. T. Flam- mia, and L. Jiang, Tailored xzzx codes for biased noise, Physical Review Research 5, 013035 (2023)

    2023

  13. [21]

    A. J. Landahl, J. T. Anderson, and P. R. Rice, Fault- tolerant quantum computing with color codes, arXiv preprint arXiv:1108.5738 (2011)

    2011 arXiv

  14. [22]

    Kapshikar and S

    U. Kapshikar and S. Kundu, On the hardness of the min- imum distance problem of quantum codes, IEEE Trans- actions on Information Theory (2023)

    2023

  15. [23]

    Kapshikar and S

    U. Kapshikar and S. Kundu, Diagonal distance of quan- tum codes and hardness of the minimum distance prob- lem, arXiv preprint arXiv:2203.04262 , 82 (2022)

    2022 arXiv

  16. [24]

    R. Y. Rubinstein and D. P. Kroese, Simulation and the Monte Carlo method(John Wiley & Sons, 2016)

    2016

  17. [25]

    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 Research 2, 043423 (2020)

    2020

  18. [26]

    Panteleev and G

    P. Panteleev and G. Kalachev, Degenerate quantum ldpc codes with good finite length performance, Quantum 5, 585 (2021)

    2021

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.