Pith. sign in

REVIEW 2 major objections 6 minor 69 references

Reviewed by Pith at T0; open to challenge.

T0 means a machine referee read the full paper against a public rubric. The mark states how deep the mechanical check went, never who wrote it. the ladder, T0–T4 →

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

T0 review · glm-5.2

One SVD of a small matrix predicts qLDPC routing depth

2026-07-08 14:12 UTC pith:TTUBJYFX

load-bearing objection Spectral results are clean and new; the O(log N) setup claim is unproven and the paper admits it the 2 major comments →

arxiv 2607.06177 v1 pith:TTUBJYFX submitted 2026-07-07 quant-ph cs.DS

Using Tanner Spectral Reduction to Improve Multi-Layer Optical Lattice Routing for Hypergraph-Product and Bivariate Bicycle qLDPC Codes

Joshua M. Courtney This is my paper
classification quant-ph cs.DS
keywords textbasetannertimesgraphroutingmulti-layerspectral
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper proves that the routing-relevant spectrum of hypergraph-product (HGP) quantum error-correcting codes is fully determined by a single singular value decomposition of the small base parity-check matrix. Specifically, the spectral ratio governing routing depth satisfies a closed-form identity: beta_HGP equals (1 + beta_base)/2, where beta_base is the ratio of the two largest singular values of the base matrix. The paper also proves an exact diameter identity (the full Tanner graph diameter is twice the base graph diameter) and, for bivariate-bicycle codes, reduces the full spectral analysis from a costly matrix diagonalization to a set of independent 2x2 decompositions via Fourier methods. These spectral results then feed into a routing protocol for multi-layer optical-lattice hardware, where stacked acousto-optic planes execute syndrome-extraction edges in parallel with no atom motion, collapsing the per-cycle routing depth from 8 sequential steps to 1.

Core claim

The central mathematical discovery is that the full Tanner graph spectrum of an HGP code built from a base matrix H decomposes into 'product modes' formed from pairs of singular values of H and 'boundary modes' from individual singular values, with the overall spectral ratio satisfying the exact identity beta_HGP = (1 + beta_base)/2. This means the routing depth of the large quantum code is computable from one SVD of the much smaller classical base matrix, without ever diagonalizing the full Tanner graph. For bivariate-bicycle codes, an analogous reduction uses the abelian torus symmetry to decompose the spectrum into lm independent 2x2 singular value problems, each arising from evaluating a

What carries the argument

The spectral ratio identity beta_HGP = (1 + beta_base)/2, the diameter identity D_T = 2*D_base, and the Fourier block-diagonalization of the bivariate-bicycle symbol matrix into 2x2 blocks.

Load-bearing premise

The 50-300x wall-clock advantage depends on hardware that does not yet exist: it requires 8 or more independently addressable stacked optical layers with 5-30 microsecond reprogramming time and no atom motion during the syndrome cycle, while current demonstrations achieve at most 2 layers. Separately, the O(log N) setup cost rests on an unproven conjecture whose key hypothesis is explicitly shown to fail for the canonical routing scheme.

What would settle it

Find an HGP code where the top singular value of the base matrix is not simple (i.e., sigma_1 = sigma_2), which would break the closed-form spectral ratio identity. Or, demonstrate that the empirical routing constant k_emp grows with N beyond 100,000, contradicting the O(log N) scaling claim.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Code designers can screen candidate qLDPC code families for favorable routing properties using only a small base-matrix SVD, avoiding expensive diagonalization of Tanner graphs with thousands of nodes.
  • The multi-layer AOL routing protocol provides a concrete architectural target for neutral-atom hardware groups: building 6-8 independently addressable optical layers would collapse syndrome-extraction routing to a single pattern activation per cycle.
  • The Fourier spectral reduction for bivariate-bicycle codes makes spectral analysis tractable at scales where full diagonalization is infeasible, enabling systematic search over polynomial families for good spectral gaps.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the chromatic index chi' of the Tanner graph is the binding constraint on per-cycle depth, then any qLDPC code family with lower maximum degree would saturate the multi-layer speedup with fewer optical layers, suggesting that code design and hardware layer count should be co-optimized.
  • The failure of the (1+beta_base)/2 formula to generalize to bivariate-bicycle codes suggests that the product-mode structure is specific to the hypergraph-product construction, and that other code families with different algebraic symmetries may require their own bespoke spectral reductions.
  • If hypothesis (H2) of Conjecture 4.2 cannot be satisfied by any 2D routing scheme, then the O(log N) setup cost may require genuinely higher-dimensional routing paths, which would further motivate the 3D hardware architecture beyond just the per-cycle advantage.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 6 minor

Summary. This manuscript characterizes the Tanner graph spectra of hypergraph-product (HGP) and bivariate-bicycle (BB) qLDPC codes and uses these results to construct a multi-layer acousto-optic lattice (AOL) routing protocol for neutral-atom quantum architectures. The spectral results include a closed-form spectral ratio β_HGP = (1+β_base)/2 (Theorem 3.2), a diameter identity D_T = 2D_base (Theorem 3.5), and a Fourier reduction of the BB Tanner spectrum to lm independent 2×2 SVDs (Theorem 3.6). These compose into a routing protocol with per-cycle depth ⌈χ'/L_layers⌉ and a one-time setup cost T_Valiant = O(log N) (Theorem 4.1). The spectral results are clean, proven from first principles, and numerically validated to machine precision. The routing protocol's per-cycle bound is a straightforward application of König's theorem. The O(log N) setup cost, however, rests on Conjecture 4.2 / Theorem 4.3, which is explicitly conditional on hypotheses (H1)–(H3), of which (H2) is shown to fail for the canonical routing scheme.

Significance. The spectral results (Proposition 3.1, Theorems 3.2, 3.5, 3.6) are the strongest contribution: they are parameter-free derivations from the HGP stabilizer block structure and Fourier diagonalization of circulants, with proofs in the main text and Appendix A. Numerical validation is thorough—26 HGP instances with eigenvalue matching to double precision and 52 BB codes with maximum discrepancy 3.1×10⁻¹⁴. The BB spectral reduction (Theorem 3.6) is a practically useful tool, reducing spectral analysis from O((lm)³) to O(lm). The per-cycle routing bound ⌈χ'/L_layers⌉ is sound, following from König's theorem and multi-layer parallelism. The wall-clock advantage estimates (50–300×) are transparently conditioned on hardware assumptions (R2-A, R2-B) that are not yet demonstrated, and the paper is commendably explicit about this gap. The O(log N) setup cost is the weakest link: it is supported only by empirical extrapolation (k_emp ≈ 0.55 over N ≤ 100,000) after the paper itself proves that hypothesis (H2) fails for the canonical routing scheme.

major comments (2)
  1. Theorem 4.1 (§4.2) states T_Valiant = O(log N) as part of the cost decomposition, but this is unproven. The paper's own analysis in §4.3 shows that Theorem 4.3 (which would give the O(log N) bound) is conditional on hypotheses (H1)–(H3), and that (H2) fails for the canonical 2D routing scheme (Lemma 4.5 gives μ₀ = O(√N), not O(1)). The abstract, Table 1, and Theorem 4.1 all present T_Valiant = O(log N) as established, while the body text (§4.3, §6.6) acknowledges the gap. The headline claim should be consistently qualified: Theorem 4.1 should state that the O(log N) setup cost is conjectural (supported by empirical k_emp ≈ 0.55 over N ∈ [225, 100,000]) rather than presenting it as a theorem. As written, Theorem 4.1 overclaims.
  2. Table 1 and the abstract present the multi-layer 3D AOL scheme with T_Valiant = O(log N) setup cost alongside demonstrated hardware schemes (Xu et al., Bravyi et al.) without clearly flagging that the O(log N) is unproven and the hardware (L_layers ≥ 8) is not demonstrated. The 'Hardware demonstrated' column for this work reads 'partial (L_layers ≲ 2 today),' which is appropriate, but the setup cost column should note its conditional status. Consider adding a footnote or parenthetical in both Table 1 and the abstract to distinguish proven results from conjectural/empirical ones.
minor comments (6)
  1. §2.3: The hardware model uses L for both the grid dimension (N = L²) and L_layers for the number of AOL planes. Using L for the grid dimension is standard but the proximity to L_layers creates occasional ambiguity. Consider renaming the grid dimension to L_grid or similar.
  2. Table 3 caption: 'The constant tends to decrease with L_layers' — the word 'constant' is ambiguous here since k_emp is itself the constant being discussed. Consider rephrasing to 'k_emp tends to decrease with L_layers.'
  3. §4.3, Conjecture 4.2: The conjecture states c ≈ 1, but Theorem 4.3's union bound requires c > 2(ln 2)/3 ≈ 0.46. The relationship between the conjectured c ≈ 1 and the empirical k_emp ≈ 0.55 should be made explicit—does k_emp ≈ 0.55 imply c ≈ 0.55, which is above the 0.46 threshold? Clarifying this would strengthen the empirical-to-theoretical bridge.
  4. §5.4, Table 5: Rows for N ≥ 10⁶ are labeled as 'asymptotic extrapolations' in the caption, but the table format makes them visually indistinguishable from measured rows. Consider separating extrapolated rows with a horizontal rule or placing them in a separate sub-table.
  5. §6.2: The wall-clock comparison uses τ_AOL ∈ [5, 30] µs and τ_move ≈ 200 µs. The 50–300× range is derived from 1.6ms / [5–30]µs, but the lower bound (50×) corresponds to τ_AOL = 30µs while the upper bound (300×) corresponds to τ_AOL ≈ 5µs. The paper should clarify that the 50–300× range reflects uncertainty in τ_AOL, not in the chromatic collapse factor.
  6. Reference [12] (Courtney 2026, arXiv:2605.02498) is cited as the source of the analytical bound that is 'empirically loose by ~25×.' Since this is the author's own prior work, a brief note on what is new in this manuscript versus [12] would help readers assess the contribution. The shifted-Cayley overlay analysis in §6.5 appears to be carried over from [12].

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for a careful and accurate reading. The referee correctly identifies that the spectral results (Proposition 3.1, Theorems 3.2, 3.5, 3.6) are the strongest contribution and are rigorously proven, while the O(log N) setup-cost claim in Theorem 4.1 is only conditionally supported. We agree with both major comments: Theorem 4.1 overclaims by presenting T_Valiant = O(log N) as established when the manuscript's own analysis shows hypothesis (H2) fails for the canonical routing scheme, and the abstract/Table 1 should clearly flag the conjectural status of this bound. We will revise accordingly.

read point-by-point responses

  1. Referee: Theorem 4.1 states T_Valiant = O(log N) as established, but the paper's own analysis shows Theorem 4.3 is conditional on (H1)-(H3) and (H2) fails for the canonical 2D routing scheme (Lemma 4.5 gives mu_0 = O(sqrt(N))). The headline claim should be consistently qualified as conjectural rather than presented as a theorem.

    Authors: The referee is correct. Theorem 4.1 as stated overclaims. The per-cycle cost bound ceil(chi'/L_layers) is rigorously proven via Konig's theorem and multi-layer parallelism, and the cost decomposition T_total = T_Valiant + R * ceil(chi'/L_layers) is correct as an algebraic identity. However, the specific scaling T_Valiant = O(log N) is not proven: Theorem 4.3, which would yield this bound, is explicitly conditional on hypotheses (H1)-(H3), and the manuscript's own Lemma 4.5 shows that (H2) fails for the canonical 2D routing scheme (mu_0 = O(sqrt(N)), not O(1)). The O(log N) scaling is supported only by empirical measurement (k_emp ~ 0.55 over N in [225, 100,000]) and by the conjecture that a check-node-bridge routing scheme satisfying (H2) exists, which we have not constructed. We will revise Theorem 4.1 to state the cost decomposition as a theorem while explicitly marking the T_Valiant = O(log N) scaling as conjectural, supported by empirical evidence, and conditional on the future construction of an (H2)-satisfying routing scheme. The body text in Sections 4.3 and 6.6 already acknowledges this gap; the revision ensures the theorem statement itself is consistent with that acknowledgment. revision: yes

  2. Referee: Table 1 and the abstract present T_Valiant = O(log N) alongside demonstrated hardware schemes without clearly flagging that the O(log N) is unproven and the hardware (L_layers >= 8) is not demonstrated. The setup cost column should note its conditional status.

    Authors: The referee is right that the abstract and Table 1 should distinguish proven results from conjectural/empirical ones. The per-cycle bound ceil(chi'/L_layers) is proven; the T_Valiant = O(log N) setup cost is not. We will add a parenthetical qualifier in both the abstract and Table 1 indicating that the O(log N) setup cost is conjectural (empirically supported with k_emp ~ 0.55 over N <= 100,000, conditional on Conjecture 4.2 / Theorem 4.3). The 'Hardware demonstrated' column already reads 'partial (L_layers <= 2 today),' which is appropriate; we will add a corresponding note to the setup-cost column. The 50-300x wall-clock estimates are already transparently conditioned on hardware assumptions (R2-A, R2-B) and will remain so. revision: yes

Circularity Check

0 steps flagged

No significant circularity: spectral results are first-principles derivations; one minor self-citation to [12] for the routing bound is not load-bearing for the central claims.

full rationale

The paper's core spectral results are derived from first principles. Proposition 3.1 follows from explicit SVD algebra on the HGP stabilizer block structure (Appendix A, Eq. 16). Theorem 3.2 (β_HGP = (1+β_base)/2) follows by identifying the Perron eigenvalue 2σ₁ and the second eigenvalue σ₁+σ₂ from the multiset in Eq. (3)—straightforward algebra, not a definition. Theorem 3.5 (D_T = 2D_base) is proved by independent upper and lower bound arguments using projection structure. Theorem 3.6 (BB spectral reduction) uses standard Fourier diagonalization of circulant matrices. None of these reduce to their inputs by construction. The routing protocol (Theorem 4.1) composes König's theorem (external, classical, properly cited [44–47]) with the spectral ratio for the per-cycle depth ⌈χ'/L_layers⌉, which is sound. The one self-citation is to Courtney [12] (arXiv:2605.02498, same author) for the analytical Valiant routing bound. This citation is not load-bearing for the central spectral or per-cycle results: the paper itself shows the [12] bound is loose by ~25×, proposes Conjecture 4.2 to improve it, proves Theorem 4.3 conditionally on (H1)–(H3), and then explicitly demonstrates that (H2) fails for the canonical 2D routing scheme (Lemma 4.5 gives μ₀ = O(√N), not O(1)). The O(log N) setup claim in Theorem 4.1 is not rigorously proven—it rests on empirical k_emp ≈ 0.55 measured up to N=100,000—but this is a proof gap / correctness risk, not circularity. The empirical constant k_emp = T_Valiant/log²N is a normalization of directly measured routing depths, not a fitted parameter renamed as a prediction. No result in the paper reduces to its inputs by construction.

Axiom & Free-Parameter Ledger

4 free parameters · 5 axioms · 2 invented entities

The spectral results rest on standard mathematical axioms (Valiant's bound, König's theorem) and one domain assumption (simple σ₁) that is argued generic. The hardware assumptions (R2-A, R2-B) and the unverified hypothesis (H2) are the load-bearing ad hoc elements. The k_emp constant is empirically measured, not fitted to a target. No invented entities are pulled from a hat without falsifiable handles—the protocol is testable if hardware is built.

free parameters (4)
  • k_emp = ≈0.55 (L_layers=8), ≈0.51 (L_layers=16)
    Empirically measured constant relating T_Valiant to log₂N. Not fitted to a target result but measured from simulation; the paper explicitly labels N≥10⁶ rows as conjectured extrapolation.
  • d₀ (overlay degree) = 8
    Degree of random overlay graphs used in Valiant routing union. Chosen by hand; affects β_union and hence T_Valiant.
  • τ_AOL (pattern reprogramming time) = 5–30 µs
    Hardware parameter from Bluvstein et al. Methods [11], not fitted by the authors but assumed for wall-clock estimates.
  • τ_move (per-cycle atom move time) = ≈200 µs
    Hardware parameter from demonstrated short-range moves [11, 39], used for wall-clock comparison baseline.
axioms (5)
  • standard math Valiant's two-phase routing bound (Eq. 1): r_t(G) ≤ 4(d'+6)/(d'·log₂(1/β))·log₂N + 19·log₂N
    Classical result from Valiant [18, 19] and Leighton-Maggs-Rao [41]. Invoked in §2.1 as the routing-depth bound that β_HGP feeds into.
  • standard math König's theorem: χ′(G) = Δ(G) for bipartite G
    Classical graph theory result [44]. Invoked in §4.2 to establish χ′(G_T) = 2δ_c for bipartite Tanner graphs.
  • domain assumption Simple top singular value hypothesis: σ₁(H) > σ₂(H)
    Required for Theorem 3.2. The paper argues (Remark 3.4) this is generic for random biregular bases and was satisfied in all tested instances.
  • ad hoc to paper Multi-layer AOL hardware with L_layers independent pattern channels and no atom motion in application phase (R2-A, R2-B)
    The 50–300× wall-clock advantage depends on this hardware capability. Current demonstrations are at L_layers≲2 (§6.4). Invoked in §6.2 and §6.4.
  • ad hoc to paper Hypothesis (H2): per-source bound sp ≤ μ₀ for absolute constant μ₀ = O(1)
    Required for Theorem 4.3 (the O(log N) setup bound). The paper explicitly states (§4.3) this is NOT satisfied by canonical 2D routing, where Lemma 4.5 gives μ₀ = O(√N).
invented entities (2)
  • Multi-layer 3D AOL routing protocol (Algorithm 1) independent evidence
    purpose: Parallelize syndrome-extraction chromatic colors across stacked AOL planes
    The protocol is validated by classical simulation (Tables 3–8) and the per-cycle step-count reduction is a direct consequence of König's theorem. However, the wall-clock advantage requires hardware not yet built. Falsifiable: if L_layers≥8 AOL hardware is built, the 8× step-count reduction is testable.
  • Check-node-bridge routing (proposed, §6.6) no independent evidence
    purpose: Routing scheme with O(log N) canonical paths to satisfy hypothesis (H2)
    Explicitly left to future work (§6.6). No construction is given; it is a plausible approach to making Theorem 4.3 unconditional.

pith-pipeline@v1.1.0-glm · 29804 in / 3867 out tokens · 750287 ms · 2026-07-08T14:12:30.383490+00:00 · methodology

0 comments
read the original abstract

We characterize the Tanner graph spectrum of hypergraph-product (HGP) / lifted-product (LP) codes and bivariate-bicycle (BB) codes, informing qubit routing for three-dimensional reconfigurable qubit architectures. Syndrome-extraction routing depth on HGP/LP Tanner graphs reduces to a single SVD on the base parity-check matrix, using a spectral ratio $\beta_\text{HGP} = (1 + \beta_\text{base})/2$ where $\beta_\text{base} = \sigma_2(H)/\sigma_1(H)$ for the base parity-check matrix, and a diameter identity $D_T = 2 D_\text{base}$ where $D_\text{base}$ is the base Tanner graph diameter. Fourier spectral reduction reveals that the BB Tanner graph spectrum equals the union, over the $l \times m$ grid of characters of $\mathbb{Z}_l \times \mathbb{Z}_m$, of the singular values of a single $2 \times 2$ symbol matrix built from the two defining polynomials. This reduces spectral analysis from an $O((lm)^3)$ diagonalization of the $4lm$-node Tanner graph to $lm$ independent $2 \times 2$ SVDs. These results compose into a multi-layer three-dimensional AOL routing protocol with one-time setup cost $T_\text{Valiant} = O(\log N)$ atom rearrangements amortizable over a memory experiment of $R$ rounds. For a Tanner graph chromatic index $\chi'$ and $L_\text{layers}$ stacked AOL planes, the per-syndrome-cycle depth is $\lceil \chi'/L_\text{layers} \rceil$ AOL pattern activations with no atom motion, an $8\times$ step-count reduction at $L_\text{layers} \geq \chi' = 8$. Contingent on multi-layer AOL hardware, this yields an estimated $\sim50-300\times$ per-cycle wall-clock advantage over a single-layer AOD baseline (degrading to $\sim5-100\times$ under AOD-crosstalk overhead), reducing to equality in the single-layer limit. This paper therefore presents a route toward practical routing improvement for future quantum hardware incorporating multi-layer reconfigurable qubit architectures.

Figures

Figures reproduced from arXiv: 2607.06177 by Joshua M. Courtney.

Figure 1
Figure 1. Figure 1: Spectrum decomposition of the HGP code [[225, 9, 4]] (seed 1, full-rank base H). Top: histogram of all 441 Tanner graph eigenvalues, stacked by Proposition 3.1 classification, as product modes ±(σi ±σj ) (blue, 333 eigenvalues, 76%) and boundary modes ±σk (orange, 108 eigenvalues, 24%). Zero unmatched eigenvalues. Bottom: the predicted multiset positions as tick marks, aligned with the top-panel histogram.… view at source ↗
Figure 2
Figure 2. Figure 2: Multi-layer 3D AOL routing protocol (Algorithm [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Per-cycle wall-clock vs. number of stacked AOL pattern channels [PITH_FULL_IMAGE:figures/full_fig_p018_3.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

69 extracted references · 69 canonical work pages · 6 internal anchors

  1. [1]

    Bravyi, A

    S. Bravyi, A. W. Cross, J. M. Gambetta, D. Maslov, P. Rall, and T. J. Yoder, Nature627, 778 (2024)

    work page 2024

  2. [2]

    Q. Xu, J. P. Bonilla Ataides, C. A. Pattison, N. Raveendran, D. Bluvstein, J. Wurtz, B. Vasić, M. D. Lukin, L. Jiang, and H. Zhou, Nature Physics20, 1084 (2024)

    work page 2024

  3. [3]

    N. P. Breuckmann and J. N. Eberhardt, PRX quantum2, 040101 (2021)

    work page 2021

  4. [4]

    Fault-Tolerant Quantum Computation with Constant Overhead

    D. Gottesman, arXiv preprint arXiv:1310.2984 (2013)

    work page internal anchor Pith review Pith/arXiv arXiv 2013

  5. [5]

    M. A. Tremblay, N. Delfosse, and M. E. Beverland, Physical Review Letters129, 050504 (2022)

    work page 2022

  6. [6]

    Bounds on stabilizer measurement circuits and obstructions to local implementations of quantum LDPC codes

    N. Delfosse, M. E. Beverland, and M. A. Tremblay, arXiv preprint arXiv:2109.14599 (2021)

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  7. [7]

    Pecorari, S

    L. Pecorari, S. Jandura, G. K. Brennen, and G. Pupillo, Nature Communications16, 1111 (2025)

    work page 2025

  8. [8]

    Poole, T

    C. Poole, T. Graham, M. Perlin, M. Otten, and M. Saffman, arXiv preprint arXiv:2404.18809 (2024)

    work page arXiv 2024

  9. [9]

    Optimal Routing Protocols for Reconfigurable Atom Arrays

    N. Constantinides, A. Fahimniya, D. Devulapalli, D. Bluvstein, M. J. Gullans, J. Porto, A. M. Childs, and A. V. Gorshkov, arXiv preprint arXiv:2411.05061 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  10. [10]

    H. Zhou, C. Duckering, C. Zhao, D. Bluvstein, M. Cain, A. Kubica, S.-T. Wang, and M. D. Lukin, inProceedings of the 52nd Annual International Symposium on Computer Architecture (2025) pp. 1432–1448

    work page 2025

  11. [11]

    Bluvstein, S

    D. Bluvstein, S. J. Evered, A. A. Geim, S. H. Li, H. Zhou, T. Manovitz, S. Ebadi, M. Cain, M. Kalinowski, D. Hangleiter,et al., Nature626, 58 (2024)

    work page 2024

  12. [12]

    J. M. Courtney, Permutation routing on ramanujan hypergraphs with applications to neutral atom quantum architectures (2026), arXiv:2605.02498 [quant-ph]

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  13. [13]

    Lubotzky, R

    A. Lubotzky, R. Phillips, and P. Sarnak, Combinatorica8, 261 (1988)

    work page 1988

  14. [14]

    A. W. Marcus, D. A. Spielman, and N. Srivastava, SIAM Journal on computing47, 2488 (2018)

    work page 2018

  15. [15]

    G. A. Margulis, Problemy peredachi informatsii24, 51 (1988)

    work page 1988

  16. [16]

    Friedman,A proof of Alon’s second eigenvalue conjecture and related problems(American Mathematical Soc., 2008)

    J. Friedman,A proof of Alon’s second eigenvalue conjecture and related problems(American Mathematical Soc., 2008)

    work page 2008

  17. [17]

    Hoory, N

    S. Hoory, N. Linial, and A. Wigderson, Bulletin of the American Mathematical Society43, 439 (2006)

    work page 2006

  18. [18]

    L. G. Valiant, SIAM journal on computing11, 350 (1982)

    work page 1982

  19. [19]

    L. G. Valiant and G. J. Brebner, inProceedings of the thirteenth annual ACM symposium on Theory of computing(1981) pp. 263–277

    work page 1981

  20. [20]

    Tillich and G

    J.-P. Tillich and G. Zémor, IEEE Transactions on Information Theory60, 1193 (2013)

    work page 2013

  21. [21]

    A. A. Kovalev and L. P. Pryadko, Physical Review A—Atomic, Molecular, and Optical Physics 88, 012311 (2013)

    work page 2013

  22. [22]

    Panteleev and G

    P. Panteleev and G. Kalachev, IEEE Transactions on Information Theory68, 213 (2021)

    work page 2021

  23. [23]

    N. P. Breuckmann and J. N. Eberhardt, IEEE Transactions on Information Theory67, 6653 (2021)

    work page 2021

  24. [24]

    Panteleev and G

    P. Panteleev and G. Kalachev, inProceedings of the 54th annual ACM SIGACT symposium on theory of computing(2022) pp. 375–388. 22

    work page 2022

  25. [25]

    Leverrier and G

    A. Leverrier and G. Zémor, in2022 IEEE 63rd Annual Symposium on Foundations of Com- puter Science (FOCS)(IEEE, 2022) pp. 872–883

    work page 2022

  26. [26]

    Leverrier, J.-P

    A. Leverrier, J.-P. Tillich, and G. Zémor, in2015 IEEE 56th Annual Symposium on Founda- tions of Computer Science(IEEE, 2015) pp. 810–824

    work page 2015

  27. [27]

    Dinur, M.-H

    I. Dinur, M.-H. Hsieh, T.-C. Lin, and T. Vidick, inProceedings of the 55th annual ACM symposium on theory of computing(2023) pp. 905–918

    work page 2023

  28. [28]

    Saffman, T

    M. Saffman, T. G. Walker, and K. Mølmer, Reviews of modern physics82, 2313 (2010)

    work page 2010

  29. [29]

    Browaeys and T

    A. Browaeys and T. Lahaye, Nature Physics16, 132 (2020)

    work page 2020

  30. [30]

    Bernien, S

    H. Bernien, S. Schwartz, A. Keesling, H. Levine, A. Omran, H. Pichler, S. Choi, A. S. Zibrov, M. Endres, M. Greiner,et al., Nature551, 579 (2017)

    work page 2017

  31. [31]

    Ebadi, T

    S. Ebadi, T. T. Wang, H. Levine, A. Keesling, G. Semeghini, A. Omran, D. Bluvstein, R. Samajdar, H. Pichler, W. W. Ho,et al., Nature595, 227 (2021)

    work page 2021

  32. [32]

    Scholl, M

    P. Scholl, M. Schuler, H. J. Williams, A. A. Eberharter, D. Barredo, K.-N. Schymik, V. Lien- hard, L.-P. Henry, T. C. Lang, T. Lahaye,et al., Nature595, 233 (2021)

    work page 2021

  33. [33]

    Barredo, S

    D. Barredo, S. De Léséleuc, V. Lienhard, T. Lahaye, and A. Browaeys, Science354, 1021 (2016)

    work page 2016

  34. [34]

    Barredo, V

    D. Barredo, V. Lienhard, S. De Leseleuc, T. Lahaye, and A. Browaeys, Nature561, 79 (2018)

    work page 2018

  35. [35]

    Y.-H. Lu, N. Song, T. Xiang, J. Ho, T.-C. Lee, Z. Yan, and D. M. Stamper-Kurn, Optica Quantum4, 241 (2026)

    work page 2026

  36. [36]

    Z. Guo, R. A. van Herk, E. J. Vredenbregt, and S. J. Kokkelmans, arXiv preprint arXiv:2510.09398 (2025)

    work page arXiv 2025

  37. [37]

    L. R. Picard and M. Endres, arXiv preprint arXiv:2510.07633 (2025)

    work page arXiv 2025

  38. [38]

    Endres, H

    M. Endres, H. Bernien, A. Keesling, H. Levine, E. R. Anschuetz, A. Krajenbrink, C. Senko, V. Vuletic, M. Greiner, and M. D. Lukin, Science354, 1024 (2016)

    work page 2016

  39. [39]

    Bluvstein, H

    D. Bluvstein, H. Levine, G. Semeghini, T. T. Wang, S. Ebadi, M. Kalinowski, A. Keesling, N. Maskara, H. Pichler, M. Greiner,et al., Nature604, 451 (2022)

    work page 2022

  40. [40]

    L. Voss, S. J. Xian, T. Haug, and K. Bharti, Physical Review A111, L060401 (2025)

    work page 2025

  41. [41]

    F. T. Leighton, B. M. Maggs, and S. B. Rao, Combinatorica14, 167 (1994)

    work page 1994

  42. [42]

    Leighton, B

    T. Leighton, B. Maggs, and A. W. Richa, Combinatorica19, 375 (1999)

    work page 1999

  43. [43]

    Upfal, Journal of the ACM (JACM)39, 55 (1992)

    E. Upfal, Journal of the ACM (JACM)39, 55 (1992)

    work page 1992

  44. [44]

    König, Mathematische Annalen77, 453 (1916)

    D. König, Mathematische Annalen77, 453 (1916)

    work page 1916

  45. [45]

    V. G. Vizing, Diskret analiz3, 25 (1964)

    work page 1964

  46. [46]

    Diestel, inGraph theory(Springer, 2024) pp

    R. Diestel, inGraph theory(Springer, 2024) pp. 179–226

    work page 2024

  47. [47]

    R. Cole, K. Ost, and S. Schirra, Combinatorica21, 5 (2001)

    work page 2001

  48. [48]

    Bernstein, Ann

    S. Bernstein, Ann. Sci. Inst. Sav. Ukraine, Sect. Math1, 38 (1924)

    work page 1924

  49. [49]

    S.Boucheron, G.Lugosi,andP.Massart,Concentration Inequalities: A Nonasymptotic Theory of Independence(Oxford University Press, 2013)

    work page 2013

  50. [50]

    McDiarmidet al., Surveys in combinatorics141, 148 (1989)

    C. McDiarmidet al., Surveys in combinatorics141, 148 (1989)

    work page 1989

  51. [51]

    Joag-Dev and F

    K. Joag-Dev and F. Proschan, The Annals of Statistics , 286 (1983)

    work page 1983

  52. [52]

    D. P. Dubhashi and D. Ranjan, BRICS Report Series3(1996)

    work page 1996

  53. [53]

    Bluvstein, A

    D. Bluvstein, A. A. Geim, S. H. Li, S. J. Evered, J. P. Bonilla Ataides, G. Baranes, A. Gu, T. Manovitz, M. Xu, M. Kalinowski,et al., Nature649, 39 (2026)

    work page 2026

  54. [54]

    J. N. Eberhardt and V. Steffan, IEEE Transactions on Information Theory71, 1140 (2024)

    work page 2024

  55. [55]

    M. H. Shaw and B. M. Terhal, Physical review letters134, 090602 (2025)

    work page 2025

  56. [56]

    A matching decoder for bivariate bicycle codes

    K. Sahay, D. J. Williamson, and B. J. Brown, arXiv preprint arXiv:2602.22770 (2026)

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  57. [57]

    T. J. Yoder, E. Schoute, P. Rall, E. Pritchett, J. M. Gambetta, A. W. Cross, M. Carroll, and M. E. Beverland, arXiv preprint arXiv:2506.03094 (2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  58. [58]

    Wang and F

    M. Wang and F. Mueller, Quantum10, 2009 (2026)

    work page 2009

  59. [59]

    Viszlai, W

    J. Viszlai, W. Yang, S. F. Lin, J. Liu, N. Nottingham, J. M. Baker, and F. T. Chong, in 2025 IEEE International Conference on Quantum Computing and Engineering (QCE), Vol. 1 (IEEE, 2025) pp. 688–699

    work page 2025

  60. [60]

    S. J. Evered, D. Bluvstein, M. Kalinowski, S. Ebadi, T. Manovitz, H. Zhou, S. H. Li, A. A. Geim, T. T. Wang, N. Maskara,et al., Nature622, 268 (2023)

    work page 2023

  61. [61]

    R. B.-S. Tsai, X. Sun, A. L. Shaw, R. Finkelstein, and M. Endres, PRX Quantum6, 010331 (2025). 23

    work page 2025

  62. [62]

    Peper, Y

    M. Peper, Y. Li, D. Y. Knapp, M. Bileska, S. Ma, G. Liu, P. Peng, B. Zhang, S. P. Horvath, A. P. Burgers,et al., Physical Review X15, 011009 (2025)

    work page 2025

  63. [63]

    Li and J

    Y. Li and J. D. Thompson, PRX Quantum5, 020363 (2024)

    work page 2024

  64. [64]

    Sinclair, J

    J. Sinclair, J. Ramette, B. Grinkemeyer, D. Bluvstein, M. D. Lukin, and V. Vuletić, Physical Review Research7, 013313 (2025)

    work page 2025

  65. [65]

    Grinkemeyer, E

    B. Grinkemeyer, E. Guardado-Sanchez, I. Dimitrova, D. Shchepanovich, G. E. Mandopoulou, J. Borregaard, V. Vuletić, and M. D. Lukin, Science387, 1301 (2025)

    work page 2025

  66. [66]

    Ramette, J

    J. Ramette, J. Sinclair, N. P. Breuckmann, and V. Vuletić, npj Quantum Information10, 58 (2024)

    work page 2024

  67. [67]

    L. Li, X. Hu, Z. Jia, W. Huie, W. K. C. Sun, Aakash, Y. Dong, N. Hiri-O-Tuppa, and J. P. Covey, Nature Physics21, 1826 (2025)

    work page 2025

  68. [68]

    H. J. Manetsch, G. Nomura, E. Bataille, X. Lv, K. H. Leung, and M. Endres, Nature647, 60 (2025)

    work page 2025

  69. [69]

    N.-C. Chiu, E. C. Trapp, J. Guo, M. H. Abobeih, L. M. Stewart, S. Hollerith, P. L. Stroganov, M. Kalinowski, A. A. Geim, S. J. Evered,et al., Nature646, 1075 (2025). 24

    work page 2025