Pith. sign in

REVIEW 1 major objections 5 minor 27 references

Designer Codes from GALA: Compact, Self-Dual, and Rate-1/2 QEC on Reconfigurable Atom Arrays

T0 review · 1 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 claims that GALA codes — built by lifting over a product of a small non-abelian group and a large cyclic group — produce certified rate-1/2 codes with distances 12, 14, and 16 and millisecond QEC cycles on atom arrays.

desk verdict A strong co-design construction with proven bounds; the exact-distance certification needs to be reproducible before the barrier-breaking claims are accepted. read the letter →

arxiv 2608.07431 v1 pith:4ENEJTDZ submitted 2026-08-07 quant-ph

Willers Yang , Casey Duckering , Arpit Dua This is my paper · ORCID
classification quant-ph PACS 03.67.Pp
keywords quantumLDPCcodesreconfigurableneutral-atomarraysactiveorthogonalitygroup-actionliftsacousto-opticdeflectorscodeautomorphismsZX-dualityfold-transversalCliffordgates
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 claims that the conflict between high encoding rate, large distance, and physical implementability in quantum LDPC codes can be resolved by factoring the code construction itself. The GALA construction lifts a block-circulant pair of check matrices over a product group $G=H_k\times C_m$ whose two factors have separate jobs: a small non-abelian factor (such as $S_3$) enforces active orthogonality, which keeps the stabilizer rows light while turning the discarded latent rows into logical operators whose distance is no longer bounded by the check weight, and a large cyclic factor $C_m$ supplies the symmetries that become explicit parallel row-and-column atom moves and code automorphisms. The concrete payoff is a self-dual $[[132,30,12]]$ code on 132 atoms with a 3.1 ms syndrome-extraction cycle, a girth-6 rate-1/2 $[[672,336,12]]$ code, and rate-1/2 $[[1752,880,14]]$ and $[[2232,1120,16]]$ codes whose certified distances exceed their check weight, all shorter than previously known hardware-compatible rate-1/2 codes. If the claims hold, near-term atom-array processors become a credible path to low-overhead fault-tolerant quantum memory.

What carries the argument

The load-bearing object is the product-group lift with active orthogonality. Generators $F=\sum_i z_i\otimes F_i\otimes f_i$ and $G=\sum_i z_i\otimes G_i\otimes g_i$ live in the group algebra $\mathbb{F}_2[\mathbb{Z}_{L/2}\times H_k\times C_m]$; their commutators satisfy $[F_i,G_j]=0$ iff the small non-abelian factors commute, so the choice of the active set $\Gamma$ — the $J\times J$ band of block rows that will remain stabilizers — is a finite pattern inside $H_k$ (e.g. $S_3$), enumerable independently of $m$. The abelian factor then does the hardware work: the diagonal copy $I_{kL}\times C_m$ is a group of code automorphisms that organizes the logical space into hypercubes, and the syndrome-extraction schedule reads off as rigid row and column translations, one pair of AOD sweeps per round. Closed-form bounds tie design parameters to outcomes: rate $\ge 1-2J/L$; for monomial lifts the barrier relation $d\le w=L$ holds whenever $J>L/4$ or $L<12$, which is why the compact instances use polynomial lifts (sums of group elements per entry) to decouple the weight from $L$; and quotient inflation gives $d\le m\,d_{Q_H}$ and $d\le k\,d_{Q_C}$. ZX-dual variants impose a fold symmetry $F_i=G_{r(i)}^T t$ on the generators, yielding fold-transversal Clifford gates $H$ and $S$ at the cost of one parallel rearrangement plus one gate layer.

What would settle it

Run an independent exclusion search on the published group-ring generators of $[[1752,880,14]]$ and $[[2232,1120,16]]$: enumerate every vector in $\langle H_Z\rangle^\perp\setminus\langle H_X\rangle$ and in $\langle H_X\rangle^\perp\setminus\langle H_Z\rangle$ of weight below the claimed distance (even weights through 12 and 14, since all logicals are even). Finding any such vector refutes the certified distance, and a verifier that reproduces both exclusions plus the weight-$d$ witnesses would confirm it.

Watch

Extended reading notes

Core claim

The paper's central claim is that the augmentation barrier — the rule that deleting stabilizer rows to raise the encoding rate caps the distance at the stabilizer weight — can be broken by construction rather than by post-hoc search. The mechanism is active orthogonality: in a two-block CSS code lifted over the group ring $\mathbb{F}_2[H_k\times C_m]$, the condition $H_XH_Z^T=0$ is enforced only on the $J$ active block rows that are kept as stabilizers, while the remaining latent rows are deliberately made non-orthogonal and removed; those removed rows become logical degrees of freedom whose weight is not limited by the bounded check weight. Because commutativity of the lifted generators $F_i,G_j$ is decided entirely inside the small factor $H_k$, the orthogonality pattern is a finite, enumerable combinatorial condition independent of the large factor $C_m$, which is then free to supply automorphisms and AOD schedules. The paper reports that exhaustive enumeration certifies every reported distance exactly — no logical operator below weight 12, 14, or 16 exists for the flagship instances — and that circuit-level simulation extrapolates logical error rates below $10^{-10}$ per cycle at $10^{-3}$ physical error for the rate-1/2 $[[672,336,12]]$.

Load-bearing premise

The reported distances 14 and 16 rest on a custom exhaustive enumeration that the paper does not ship and describes only in outline; if that enumeration is incomplete or contains a bug, the barrier-breaking parameter claims collapse.

Editorial extensions

If this is right

  • Rate-1/2 codes with check weight 12 and certified distances up to 16 are realizable in the few-hundred-to-two-thousand-qubit regime of present-day atom-array devices, with syndrome-extraction cycles of roughly 3–9 ms depending on the number of crossed AODs.
  • The claimed $[[1752,880,14]]$ and $[[2232,1120,16]]$ codes are shorter than every previously known hardware-compatible rate-1/2 code at those distances, breaking the $d\le w$ barrier with exact certification.
  • Hardware compatibility and logical operations become inputs rather than post-hoc checks: every GALA code inherits an explicit AOD move schedule and shift automorphisms, and ZX-dual members carry fold-transversal Clifford gates costing one rearrangement and one gate layer — the same primitives as one QEC round.
  • The GALA family reproduces the previously known rate-1/2 affine-permutation codes as special cases, so the new parameter bounds, logical bases, and ZX-dual variants apply to that whole family.

Reading between the lines

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

  • The paper's reported trade — about one-and-a-half decades worse logical error rate for an eightfold reduction in qubit overhead — suggests a scaling law worth probing: the abelian factor controls cycle time and automorphisms while the non-abelian top sets the distance ceiling, so scanning tops beyond $S_3$ and $S_4$ at fixed $L,J$ is the natural next search dimension.
  • Because the certified distances rest on an unshipped exhaustive enumerator, the decisive reproducibility test is an independent re-implementation of the exclusion search on the published generators; the paper enables this by tabulating explicit group-ring generators for every reported code.
  • The constructive quotient logicals can be up to $k$ times heavier than the minimum weight, so the move-based gates are not yet a fault-tolerant logical library; composing the shift automorphisms with the homomorphic CNOT gadgets along quotient chains is the implied route to a complete Clifford set, with fault tolerance still to be proven.
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

1 major / 5 minor

Summary. The paper introduces GALA codes, a family of CSS qLDPC codes constructed by lifting a block-circulant proto-matrix over a product or semidirect product of a small non-abelian group (typically S_3 or S_4) and a cyclic group. The factorization gives active orthogonality from the small non-abelian factor and symmetries/AOD move schedules from the abelian factor. The authors prove closed-form bounds on rate, distance, and girth, give explicit logical operators and automorphisms, construct ZX-dual variants with fold-transversal Clifford gates, and report a search producing compact rate-1/2 codes including [[132,30,12]], [[672,336,12]], and the headline barrier-breaking instances [[1752,880,14]] and [[2232,1120,16]], all claimed to have exactly certified distances at check weight 12. The paper also gives syndrome-extraction cycle-time estimates and circuit-level logical error rate simulations.

Significance. If the claims are correct, the paper makes a strong co-design contribution: it provides a transparent algebraic framework that contains previously known AOD-compatible Kasai codes, supplies analytic bounds, and yields concrete rate-1/2 codes with d>w that are short enough for near-term atom-array hardware. The closed-form bounds in Section S3 and the explicit move schedules in Section S5 are genuine strengths, as is the use of an end-to-end case study for [[132,30,12]]. However, the central distance claims rest on an exhaustive enumeration whose code and algorithmic details are not provided, and the 'all smaller than previously known' abstract claim needs qualification. The result is significant if reproducible, but the certification step is currently a black box.

major comments (1)
  1. [S6.1, Table S3] The exact-distance certifications for [[1752,880,14]] and [[2232,1120,16]] rest entirely on an exhaustive enumeration described only as 'our own implementation' in Section S6.1. No code, pseudocode, pruning strategy, complexity analysis, or completeness proof is given. For n=2232 and d=16, a naive search over all supports of weight at most 15 is astronomically large (C(2232,15) is on the order of 10^35), so the feasibility and correctness of the claimed certification depend on undocumented structure or pruning. The manuscript itself states that a 2160-qubit instance has only its X-sector enumeration complete, which underscores that this is a substantial custom computation. Because the 'barrier-breaking' status of the two headline codes is exactly the claim that d>w=12, the central result cannot be verified from the paper alone. I recommend that the authors release the enumerator or provide a detailed algorithm with complexity analysis and validation tests, together with certification logs for each reported instance.
minor comments (5)
  1. [Abstract and main text] The abstract says the two barrier-breaking codes are 'all smaller than previously known hardware compatible rate-1/2 codes,' but the main text acknowledges that the pair-partition code [[1524,766,14]] of Ref. [10] is shorter at d=14. Please clarify the exact comparison class (e.g., AOD-compatible codes only) and align the abstract with that qualification.
  2. [Main text, Table S5] The abstract describes [[132,30,12]] as 'almost girth-6,' while the main text and Table S5 report girth 4 (t4=660). Please use consistent terminology and explicitly state that the code is girth 4 despite being called almost girth-6.
  3. [S3.1, Proposition 7] The proof says 'All rows sum up to 1^{1×n},' but the sum of the active rows is the all-ones vector only when the column weight J is odd; for even J (e.g., J=2) the sum is zero. Please state the parity condition and verify the claimed J−1 rank-deficiency count for even J, or restrict the proposition to the odd case.
  4. [S6.2, Figure S3] The extrapolated logical error rates at p=10^-3 depend on two fit forms with parameters c0,c1,c2 and d_eff. The text already calls this an extrapolation, but the figure caption should note explicitly that the two fit forms can disagree and that the quoted 10^-10 LER for [[672,336,12]] is from the chosen form.
  5. [Table S3] Several certified frontier codes listed in Table S3, such as [[1764,886,14]] and [[2328,1168,16]], are not discussed in the main text. Please add a sentence explaining their status relative to the headline claims and why they are not included in the abstract.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: all bounds and certified distances follow from the defined group-product lift and independent exhaustive enumeration.

full rationale

The derivation chain is self-contained rather than circular. The GALA construction is defined in Definitions 15 and 16 as a lift over H_k × C_m or H_k ⋉ C_m^k, and the rate bound (Eq. S35), distance caps (Prop. 8, Cor. 11), girth inheritance (Lemma 12), and quotient distance bounds (Cor. 10) are proved from the group data in Section S3, not fitted to the reported codes. The code-search section is transparently two-stage: QDistRnd yields only an upper bound on distance, and every reported instance is then certified exactly by exhaustive exclusion of all logical operators of weight below d together with an explicit weight-d witness (Section S6.1). That certification does not assume the claimed distance, so the headline d=12, 14, 16 values are not definitionally encoded in the search inputs. The self-citation to Ref. [2] in Proposition 5 (using Lemma 1 of [2] to identify AOD-compatible Kasai codes as GALA instances) is contextual: it supports the 'contains previously discovered Kasai codes' claim and baseline comparisons, but it is not needed to establish the new bounds or the new certified instances, and the cited result is a mathematical statement from prior work rather than a fitted parameter renamed as a prediction. The paper also flags its own limitation that the 2160-qubit instance is not quoted because its Z-sector enumeration is incomplete, which is inconsistent with any attempt to force the claimed distances. The main caveat is reproducibility: the exact-distance enumerator is not shipped and its pruning is not described in detail, so the d=14 and d=16 certifications cannot be independently checked from the paper alone. That is a verification gap, not circularity. No step in the paper's argument reduces to its own inputs by construction.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The construction's algebraic framework is self-contained given the group choices; the main external inputs are the search-selected group elements and the physical/hardware assumptions.

free parameters (3)
  • Bottom group exponents f_i, g_i for reported code instances = Listed in Tables S3-S5 (e.g., for [[132,30,12]]: F = {x^2, x^4, x^3, x^6, x^3, x^9}, G = {x^9, x^2, x^8, x^5, x^8, x^7})
    Chosen by random/greedy search over C_m to maximize girth and distance; these choices are not derived from first principles but are inputs to the construction.
  • Fit parameters c0, c1, c2 for LER polynomial-exponent form = Not explicitly quoted (fit curves shown in Fig. S3)
    Fitted to Stim simulation data to extrapolate logical error rates at p=1e-3.
  • Effective distance d_eff in threshold-theorem fit = Reported as 9.5 for [[132,30,12]] and 13.5 for [[672,336,12]]
    Fitted parameter in C(p/p_th)^d_eff used for LER extrapolation; it is not the true code distance.
assumptions (5)
  • standard math The CSS orthogonality condition H_X H_Z^T = 0 is sufficient for a valid CSS code.
    Standard quantum error correction background (Section S1.1).
  • domain assumption The two-block circulant lift with active/latent rows defines a CSS code when active rows commute; latent rows are not stabilizers.
    This is the core construction of Kasai codes, assumed and proved via commutator sums.
  • domain assumption Physical error model with all gates and measurements at error rate p, no idle errors, and the greedy coloration syndrome-extraction circuit faithfully represents hardware behavior.
    Used in LER simulations; standard in prior work [2].
  • domain assumption AOD movement model (12 um spacing, 5500 m/s^2 acceleration) used for cycle times.
    Assumed from [22] for time estimates.
  • standard math QDistRnd provides an upper bound on distance; exact certification is done by custom exhaustive search.
    The paper relies on this to claim exact distances.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Designer Codes from GALA: Compact, Self-Dual, and Rate-1/2 QEC on Reconfigurable Atom Arrays." pith.science (2026). https://pith.science/paper/4ENEJTDZ

@misc{pith2026260807431,
  author       = {Pith},
  title        = {Pith review of: Designer Codes from GALA: Compact, Self-Dual, and Rate-1/2 QEC on Reconfigurable Atom Arrays},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4ENEJTDZ}},
  note         = {Machine review of arXiv:2608.07431}
}
abstract

High rate quantum low-density parity-check codes on reconfigurable neutral-atom arrays can reduce the overhead of quantum error correction, but near-term devices support only hundreds of qubits with limited reconfigurability from a few crossed acousto-optic deflectors (AOD). A practical code must be compact in addition to low-overhead, with checks and logical gates mapping onto hardware-compatible physical instructions. We introduce the GALA codes, or Group-Action Lifts with Active orthogonality, that lifts over a product group $G = H_k \times C_m$ (or $H_k \ltimes C_m^k$). The small non-abelian factor $H_k$ supplies active orthogonality, reaching $1/2$ rate with above-weight distance, while the large abelian factor $C_m$ supplies symmetries that give code automorphisms and explicit, simple AOD move schedules. Hardware compatibility and logical capability thereby become customizable inputs to a code search rather than properties verified post-hoc, making GALA designer codes by construction. The GALA family contains several previously discovered rate-1/2 Kasai codes of Ref. [arXiv:2601.08824, arXiv:2604.16209] while exposing simpler parameter bounds, logical operations, and ZX-dual variants with AOD-compatible fold-transversal Clifford gates. Our search yields a compact self-dual $[[132, 30, 12]]$ with a small number of $4$-cycles (almost girth-6) and below $10^{-8}$ logical error rate (LER) for memory at $10^{-3}$ physical error rate, 3.1ms syndrome-extraction cycle and transversal Clifford gates; a girth-6, rate-$1/2$ $[[672, 336, 12]]$ with a 6.76ms cycle with below $10^{-10}$ LER (extrapolated) and rate-$1/2$ barrier-breaking $[[1752, 880, 14]]$ and $[[2232, 1120, 16]]$ with exactly certified distances greater than check weights and all smaller than previously known hardware compatible rate-1/2 codes.

Figures

Figures reproduced from arXiv: 2608.07431 by the authors.

Figure 1
Figure 1. FIG. 1. GALA codes simultaneously achieve low qubit over [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. GALA: a family of two-block codes lifted over a product group [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 4
Figure 4. FIG. 4. GALA codes (lower-right is better). [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 19 canonical work pages

  1. [1]

    2026 , eprint=

    Towards Ultra-High-Rate Quantum Error Correction with Reconfigurable Atom Arrays , author=. 2026 , eprint=

    work page 2026

  2. [2]

    2026 , eprint=

    Breaking the Orthogonality Barrier in Quantum LDPC Codes , author=. 2026 , eprint=

    work page 2026

  3. [3]

    2025 , eprint=

    Automorphism gadgets in homological product codes , author=. 2025 , eprint=

    work page 2025

  4. [4]

    2026 , eprint=

    Spacetime-Efficient and Hardware-Compatible Complex Quantum Logic Units in qLDPC Codes , author=. 2026 , eprint=

    work page 2026

  5. [5]

    and Burton, Simon , year=

    Breuckmann, Nikolas P. and Burton, Simon , year=. Fold-Transversal Clifford Gates for Quantum Codes , volume=. doi:10.22331/q-2024-06-13-1372 , journal=

    doi:10.22331/q-2024-06-13-1372 2024

  6. [6]

    Quantum Quasi-Cyclic LDPC Codes , url=

    Hagiwara, Manabu and Imai, Hideki , year=. Quantum Quasi-Cyclic LDPC Codes , url=. doi:10.1109/isit.2007.4557323 , booktitle=

    work page arXiv 2007

  7. [7]

    and Shabashov, Vadim A

    Pryadko, Leonid P. and Shabashov, Vadim A. and Kozin, Valerii K. , year=. QDistRnd: A GAP package for computing the distance of quantum error-correcting codes , volume=. Journal of Open Source Software , publisher=. doi:10.21105/joss.04120 , number=

    doi:10.21105/joss.04120

  8. [8]

    2026 , eprint=

    Pair-Partition Constructions for CPM-Based Quantum LDPC Codes , author=. 2026 , eprint=

    work page 2026

Show all 27 references
  1. [9]

    Nature , volume=

    Logical quantum processor based on reconfigurable atom arrays , author=. Nature , volume=. 2024 , publisher=

    2024

  2. [10]

    Nature Physics , volume=

    Constant-overhead fault-tolerant quantum computation with reconfigurable atom arrays , author=. Nature Physics , volume=. 2024 , publisher=

    2024

  3. [11]

    and Gambetta, Jay M

    Bravyi, Sergey and Cross, Andrew W. and Gambetta, Jay M. and Maslov, Dmitri and Rall, Patrick and Yoder, Theodore J. , year=. High-threshold and low-overhead fault-tolerant quantum memory , volume=. Nature , publisher=. doi:10.1038/s41586-024-07107-7 , number=

    doi

  4. [12]

    2026 , eprint=

    Shor's algorithm is possible with as few as 10,000 reconfigurable atomic qubits , author=. 2026 , eprint=

    2026

  5. [13]

    2025 , eprint=

    Tour de gross: A modular quantum computer based on bivariate bicycle codes , author=. 2025 , eprint=

    2025

  6. [14]

    2026 , eprint=

    Fast and Parallel High-Rate STAR Architecture for Megaquop Quantum Simulation , author=. 2026 , eprint=

    2026

  7. [15]

    Matching Generalized-Bicycle Codes to Neutral Atoms for Low-Overhead Fault-Tolerance , url=

    Viszlai, Joshua and Yang, Willers and Lin, Sophia Fuhui and Liu, Junyu and Nottingham, Natalia and Baker, Jonathan M and Chong, Frederic T , year=. Matching Generalized-Bicycle Codes to Neutral Atoms for Low-Overhead Fault-Tolerance , url=. doi:10.1109/qce65121.2025.00080 , booktitle=

  8. [16]

    Tillich, Jean-Pierre and Z. Quantum. IEEE Transactions on Information Theory , volume=. 2014 , doi=

    2014

  9. [17]

    Degenerate quantum

    Panteleev, Pavel and Kalachev, Gleb , journal=. Degenerate quantum. 2021 , doi=

    2021

  10. [18]

    IEEE Transactions on Information Theory , volume=

    Balanced product quantum codes , author=. IEEE Transactions on Information Theory , volume=. 2021 , doi=

    2021

  11. [19]

    Nature , volume=

    High-fidelity parallel entangling gates on a neutral-atom quantum computer , author=. Nature , volume=. 2023 , doi=

    2023

  12. [20]

    and Webster, Paul and Vasmer, Michael , year=

    Quintavalle, Armanda O. and Webster, Paul and Vasmer, Michael , year=. Partitioning qubits in hypergraph product codes to implement logical gates , volume=. doi:10.22331/q-2023-10-24-1153 , journal=

    2023 doi

  13. [21]

    High-rate

    Aditya Bhardwaj and Muzhou Ma and Nadine Meister and Robbie King and Dolev Bluvstein and John Preskill and Madelyn Cain and Qian Xu and Hsin-Yuan Huang , year=. High-rate. 2607.28795 , archivePrefix=

    arXiv

  14. [22]

    2026 , eprint=

    Quantum error correction at ultra-low overhead , author=. 2026 , eprint=

    2026

  15. [23]

    2022 , eprint=

    Homomorphic Logical Measurements , author=. 2022 , eprint=

    2022

  16. [24]

    2024 , eprint=

    Fast and Parallelizable Logical Computation with Homological Product Codes , author=. 2024 , eprint=

    2024

  17. [25]

    ,booktitle =

    Zhou, Hengyun and Duckering, Casey and Zhao, Chen and Bluvstein, Dolev and Cain, Madelyn and Kubica, Aleksander and Wang, Sheng-Tao and Lukin, Mikhail D. ,booktitle =. doi:10.1145/3695053.3731039 ,isbn =

  18. [26]

    Stim: a fast stabilizer circuit simulator , volume=

    Gidney, Craig , year=. Stim: a fast stabilizer circuit simulator , volume=. doi:10.22331/q-2021-07-06-497 , journal=

    2021 doi

  19. [27]

    and Mariantoni, Matteo and Martinis, John M

    Fowler, Austin G. and Mariantoni, Matteo and Martinis, John M. and Cleland, Andrew N. , year=. Surface codes: Towards practical large-scale quantum computation , volume=. Physical Review A , publisher=. doi:10.1103/physreva.86.032324 , number=

    doi

Pith tools

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