Pith. sign in

REVIEW 3 major objections 6 minor 19 references

Improved Methods for Determining Quantum Error Correcting Code Performance and Fault Tolerance

T0 review · 3 major / 6 minor · reviewed 2026-07-30 · grok-4.5

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

Pith's one-line read Typical quantum error-correction failures hide a small malignant core under correctable fluff; pruning and subregion MCMC exploit that structure to certify fault tolerance and speed low-error simulations.

desk verdict Practical QEC methods paper: pruning finds low-weight cores and subregion MCMC is a real 2–10× speedup over BV when the core heuristic fits, with one incomplete loop on the unknown-w_min use case. read the letter →

arxiv 2607.27153 v1 pith:ZOF366N2 submitted 2026-07-29 quant-ph

Michael Mullan , Matthew Weippert , Winton Brown This is my paper · ORCID
classification quant-ph
keywords quantumerrorcorrectionMarkovchainMonteCarlofaulttolerancesurfacecodelogicalrateminimum-weightfailuresMetropolis-Hastingscircuit-levelnoise
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

Utility-scale quantum algorithms need logical error rates so low that direct Monte Carlo cannot reach them, so engineers extrapolate from higher rates using the minimum weight of uncorrectable errors. That weight is often unknown once the decoder and syndrome circuit are fixed, and ordinary Monte Carlo almost never sees the rare low-weight failures. The paper shows that, in the error-rate window near threshold, a typical failure is a small malignant core surrounded by many easily correctable “fluff” errors. A simple pruning procedure strips the fluff so the core can be inspected and the true minimum weight recovered; a new family of Metropolis–Hastings moves, subregion MCMC, then resamples a controlled fraction of the lattice at each step and converges to the failure distribution far faster than single-location Metropolis. Together the two tools let implementers both debug subtle non-fault-tolerance and obtain reliable logical-error curves at the rates needed for large algorithms.

What carries the argument

Subregion MCMC: a Metropolis–Hastings proposal that selects each circuit location independently with probability p_r and resamples the chosen subregion from the noise model at flip rate p_f. With the core-resampling choice (p_r, p_f) ≈ (1/w_min, p_j) the acceptance probability is unity for uncorrectable patterns and the chain explores distinct logical cores efficiently; the method continuously interpolates between ordinary Monte Carlo (p_r = 1) and single-location Metropolis (p_r ∼ 1/N).

What would settle it

On a code family whose failures are known not to be localized cores (or at physical rates where the expected number of errors falls below t), measure the number of circuit simulations needed for Gelman–Rubin R̂ ≤ 1.05; if subregion MCMC with p_r = 1/w_min no longer requires fewer simulations than single-location Metropolis, or if pruning never recovers weight ≤ t when non-fault-tolerant hooks are deliberately inserted, the central claim fails.

Watch

Extended reading notes

Core claim

At physical error rates not far below threshold, uncorrectable error patterns consist of a low-weight malignant core coexisting with a large number of easily correctable fluff errors. Removing random small subsets of errors repeatedly (pruning) isolates that core and thereby reveals whether an implementation achieves the expected minimum failure weight. The same structure motivates subregion MCMC: by resampling a random fraction of locations (ideally about one core error per step) one obtains a Metropolis–Hastings chain that mixes between distinct logical failures orders of magnitude faster than single-site flips while still sampling the correct conditional distribution of failures.

Load-bearing premise

The claim that failures are spatially localized cores plus fluff, and that resampling roughly one core location per step remains optimal, must hold for the codes, decoders, and sparse-error regimes of interest.

Editorial extensions

If this is right

  • A few hundred pruned Monte Carlo failures suffice to certify that an implementation meets its design distance or to expose the exact low-weight bug.
  • Logical-error curves at 10^{-10} and below become obtainable on a desktop for surface and concatenated Bacon–Shor codes under full circuit noise.
  • The continuum of proposal distributions between Monte Carlo and single-site MCMC can be tuned by two scalar parameters without redesigning the sampler.
  • Any code whose failure geometry is core-plus-fluff inherits the same 2–10× reduction in simulation count relative to Bravyi–Vargo MCMC.

Reading between the lines

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

  • Adaptive schedules that raise p_r when the expected error count drops below w_min could close the sparse-error regime the authors defer.
  • The same core-extraction idea may let decoders themselves reject fluff on the fly, turning a diagnostic into an online decoder improvement.
  • If the mixing-time advantage scales with distance as suggested by the surface-code data, subregion MCMC becomes the practical default for any large-distance threshold study.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper addresses estimation of quantum error-correcting code performance at low physical error rates, where direct Monte Carlo is infeasible and asymptotic scaling P(p)=α p^{w_min} depends on a minimum uncorrectable weight that is often implementation-dependent and unknown. Motivated by the observation that typical failures near threshold consist of a low-weight malignant core plus many easily correctable “fluff” errors, the authors introduce (i) a pruning algorithm (Alg. 1) that iteratively removes subsets of errors from high-p Monte Carlo failures to recover candidate cores and thereby diagnose w_min or decoder/circuit bugs, and (ii) subregion MCMC, a Metropolis–Hastings family that resamples a random fraction p_r of circuit locations at a flip rate p_f, interpolating between single-site Bravyi–Vargo (BV) MCMC and full Monte Carlo. With the heuristic core-resampling choice (p_r,p_f)=(1/w_min,p_j), they report logical rates matching Monte Carlo when Gelman–Rubin R̂≤1.05 (Fig. 5) and roughly 2–10× fewer circuit simulations to convergence than BV across surface-code distances and concatenated Bacon–Shor codes (Fig. 7), under full circuit-level noise in QVM.

Significance. If the empirical speedups and the pruning diagnostic hold more broadly, the work supplies practical engineering tools for verifying fault tolerance and for pushing logical-error estimates into the 10^{-10} regime without heroic Monte Carlo budgets. The fluff/core decomposition is a useful organizing idea; the parameterized proposal family is a clean, reusable contribution within the Metropolis–Hastings framework; and the methods are validated against independent Monte Carlo baselines and a planted non-fault-tolerant hook bug (Figs. 1–3). These are concrete, reproducible advances over BV MCMC for circuit-level QEC simulation.

major comments (3)
  1. [§3.2, Fig. 7] §3.2 “Core Resampling” and Fig. 7: the reported 2–10× decode reduction versus BV is measured with (p_r,p_f)=(1/w_min,p_j), which presupposes knowledge of w_min (or the design distance). The introduction and §2 motivate the work precisely by the fact that w_min is often unknown a priori and can be <t+1 for non-fault-tolerant implementations (Fig. 1a). The manuscript never demonstrates the combined pipeline—high-p Monte Carlo → Alg. 1 prune → feed discovered |E| into p_r → subregion MCMC at lower p—on an instance whose true w_min is initially unknown or strictly less than t+1. Without that, or without a clear sensitivity study for misspecified p_r, the load-bearing speed claim is only partially supported for the use-case advertised in §§1–2.
  2. [§3.3] §3.3 explicitly defers the sparse-error regime (expected physical errors ≲ t), which is exactly where MCMC is most needed relative to Monte Carlo and where the fluff/core heuristic and the p_r≈1/w_min mixing argument are least justified. The conjecture that larger subregions cut decorrelation time is stated without quantitative mixing-time or effective-sample-size comparisons beyond R̂ as a proxy. A minimal addition—either data at lower p or an explicit scoping statement that the speedup is demonstrated only when the expected error count still exceeds t—would keep the central claim from overreaching.
  3. [§3.1] §3.1: statistical error on the splitting ratios is deferred; R̂ is used both as convergence diagnostic and as a proxy for uncertainty, with a citation to the R̂–ESS relation. Because temporal correlations and possible false convergence (multiple disconnected components, incomplete burn-in; cf. Fig. 6) can bias the Bennett-style ratios, the logical-error points in Fig. 5 lack rigorous error bars. At minimum the text should state what can and cannot be claimed about uncertainty from R̂ alone, or report a simple ESS/MCSE estimate on the chains already generated.
minor comments (6)
  1. [§3.3, Fig. 7] Fig. 7 switches to a bit-flip error model “since BV MCMC was originally described in terms of bit or phase flips,” while the rest of the paper (and Fig. 5) uses depolarizing noise. A sentence clarifying that the relative speedup is expected to carry over, or a single depolarizing comparison point, would remove ambiguity.
  2. [§2, Algorithm 1] Alg. 1 leaves the removal strategy (subset size distribution, K, M) as free parameters with only empirical guidance (“<5”, “1000–10000 rounds”, “a few hundred patterns”). A short sensitivity paragraph or default settings used for Figs. 2–3 would aid reproducibility.
  3. [§3.2] Eq. (12)–(14): the cancellation that makes A independent of p_r is neat but compressed; a one-line remark that locations with e_ℓ=e′_ℓ contribute factors of 1 would help readers verify the algebra.
  4. [Table 1] Table 1 caption and body: “Prob.<w_min” at p=10^{-3} is useful; stating the Poisson or binomial model used for the tail probabilities would make the table self-contained.
  5. Minor typography: “simualtedrounds”, “thathave”, “inthe”, missing spaces after periods in several places (e.g., near Eq. (1) and the start of §3), and inconsistent italicization of p_r, p_f, w_min.
  6. [§1, References] References [4] and [16] are concurrent arXiv preprints on related rare-event QEC simulation; a sentence distinguishing subregion MCMC from those proposal adaptations would help place the contribution.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: methods paper with independent Monte Carlo baselines and an explicit heuristic parameter, not a self-forced prediction.

full rationale

The paper’s load-bearing claims are (i) a pruning procedure that empirically reduces high-weight Monte Carlo failures to low-weight cores and (ii) a Metropolis–Hastings family (subregion MCMC) whose mixing is measured by the number of circuit simulations needed to reach Gelman–Rubin R̂ ≤ 1.05, then validated by matching independent Monte Carlo logical error rates via the standard splitting/Bennett ratio estimator. Neither claim is obtained by defining an output in terms of itself, nor by fitting a constant to data and relabeling the fit a prediction. The free parameter choice (p_r, p_f) = (1/w_min, p_j) is stated as a heuristic motivated by the fluff-vs-core observation; the reported 2×–10× speedup is an empirical runtime comparison against Bravyi–Vargo single-location Metropolis under that choice, not a quantity forced by construction. Citations (Bravyi–Vargo, Bennett, Gelman–Rubin, Blossom V) are external and not load-bearing uniqueness results from the same authors. Scope limitations (need for known w_min; sparse-error regime deferred) affect correctness/generality, not circularity of the derivation. Steps list is therefore empty.

Assumptions & free parameters 4 free parameters · 5 assumptions · 2 invented entities

The work rests on standard MCMC/Metropolis–Hastings theory, the circuit-level independent depolarizing (or bit-flip) noise model, and the empirical fluff/core structure of failures. Free parameters are the MCMC hyperparameters (region rate, flip rate, chain count, R̂ threshold) and pruning knobs; they are chosen heuristically rather than fitted to produce the logical-rate curves. No new physical entities are postulated.

free parameters (4)
  • region rate p_r (default 1/w_min) = 1/w_min
    Controls fraction of circuit locations resampled each MCMC step; core-resampling choice is heuristic, not derived from a mixing-time optimum.
  • flip rate p_f (default p_j) = p_j
    Base error rate used when resampling the subregion; set equal to the target physical rate so acceptance of uncorrectable proposals is 1.
  • Gelman–Rubin threshold R̂ and chain count M = M=50, R̂≤1.05
    Convergence stopping rule (M=50, R̂≤1.05 typical); used as proxy for statistical error without a full ESS-based error bar.
  • pruning removal size and round count K = <5 errors/round; K~10^3–10^4
    Heuristic knobs for Algorithm 1; paper reports removal size <5 and K in 1000–10000 as working in practice.
assumptions (5)
  • standard math Metropolis–Hastings with detailed balance yields the unique stationary distribution π_j(E|F) when the chain on uncorrectable patterns is irreducible/ergodic.
    Invoked throughout §3.1; standard MCMC theory.
  • domain assumption Circuit noise factors into independent per-location errors (Eqs. 10–11); gate error rates are monotonic in the base rate p_j for splitting.
    Standard circuit-level depolarizing/bit-flip model used for all simulations and acceptance ratios.
  • ad hoc to paper Typical uncorrectable patterns near and below threshold consist of a low-weight malignant core plus many easily correctable fluff errors.
    Central motivating observation (§2, Fig. 2); empirically illustrated for surface-code MWPM and used to set p_r=1/w_min, but not proved for general codes/decoders.
  • domain assumption Gelman–Rubin R̂ near 1 and discarding the first half of each chain suffice as practical convergence and error proxies for logical-rate ratios.
    §3.1 cites Gelman/Shirley and Vats–Knudson; paper explicitly defers rigorous MCMC standard-error analysis.
  • domain assumption Logical failure ratio P(p_j)/P(p_{j-1}) is accurately estimated from samples of the two nearby failure distributions via Bennett-style free-energy estimation.
    Splitting method as in Bravyi–Vargo; accuracy requires sufficient distribution overlap (heuristic from Ref. [5]).
invented entities (2)
  • subregion MCMC proposal family independent evidence
    purpose: New Metropolis–Hastings proposal that resamples a random circuit subregion at rate p_r with flip rate p_f, interpolating Monte Carlo and single-site MCMC.
    Defined in §3.2; not a physical entity but a new algorithmic object. Independent evidence is the empirical speedup and MC agreement in Figs. 5–7.
  • fluff vs malignant core decomposition of failure patterns
    purpose: Conceptual split used to motivate both pruning and the core-resampling parameter choice.
    Descriptive structure observed in simulations (§2); not a new physical object. Falsifiable in the weak sense that other codes could lack localized cores.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Improved Methods for Determining Quantum Error Correcting Code Performance and Fault Tolerance." pith.science (2026). https://pith.science/paper/ZOF366N2

@misc{pith2026260727153,
  author       = {Pith},
  title        = {Pith review of: Improved Methods for Determining Quantum Error Correcting Code Performance and Fault Tolerance},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZOF366N2}},
  note         = {Machine review of arXiv:2607.27153}
}
read the original abstract

One of the central challenges in quantum error correction is determining the performance of a code in the low-error regimes needed to implement utility-scale computations. While performance at these error rates is not amenable to direct Monte Carlo simulation, it can be extrapolated from simulations at higher logical error rates, assuming the logical error rate scales predictably with increasing distance or decreasing physical error rate. However, the expected scaling depends sensitively on the minimum weight of uncorrectable error patterns. In many cases, the minimum weight is unknown since it depends not only on the theoretical code distance, but also on details of the implementation. Markov chain Monte Carlo (MCMC) methods, as adapted to quantum error correction by Bravyi and Vargo, provide a way to estimate logical failure rates in these low-error regimes via simulation. While offering significant gains over Monte Carlo, the described Metropolis algorithm makes small changes to the current logical failure patterns which results in slow convergence. In this paper, we argue that typical failure patterns include a large number of easily correctable errors that coexist alongside a malignant core. This observation motivates two new approaches to better evaluate code performance. First, we describe a pruning algorithm designed to obviate these correctable errors and focus on the problematic low-weight core. Second, we develop a novel family of Metropolis-Hastings algorithms, referred to as subregion MCMC. This technique is parameterized by the fraction of the error pattern that is resampled at each step, effectively interpolating between Monte Carlo and single step MCMC. We show that a judicious choice of this parameter results in far faster convergence than prior work.

Figures

Figures reproduced from arXiv: 2607.27153 by the authors.

Figure 1
Figure 1. A d = 11 surface code decoded using MWPM. In (a) the decoder fails to take into account one of the hooks. The first five highest p points are fit to Eq. (1) with wmin = 6, which predicts a logical error rate of P(p) = 9.47×10−9 at p = 3×10−4 ; however the Monte Carlo simulation predicts P(p) = 1.54×10−8 . Notice that, until points six and seven are included, this fit looks perfectly fault tolerant, and it is not obv… view at source ↗
Figure 2
Figure 2. A randomly chosen failure pattern on a d = 7 surface code. Red circles indicate where an error was first generated; these may subsequently propagate via gates or the decoder. Note that the ratio of fluff errors to malignant core errors is worse at higher distances and physical error rates. 5 [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. The same error pattern shown in Fig [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: The first eight logical failures explored by subregion MCMC, along with the number of times each [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 5
Figure 5. Figure 5: Encoded error rates for the d = 11 surface code and the d = 5 on d = 5 concatenated Bacon-Shor code generated using subregion MCMC and standard Monte Carlo for comparison. Here we seed M = 50 chains and run until Rˆ ≤ 1.05. Comparing the Markov graph for the BV MCMC, G…
Figure 6
Figure 6. Figure 6: Encoded error rates for the d = 11 surface code, generated using subregion MCMC with Rˆ = 1.2. The blue points are generated in the usual way, via downward splitting, while the green points are generated via upward splitting. Notice that since the chains were not given…
Figure 7
Figure 7. Figure 7: (a) A comparison of the runtime of subregion and BV MCMC for surface codes of distances [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

19 extracted references · 1 linked inside Pith

  1. [1]

    Quantum accuracy threshold for concatenated distance-3 codes.Quantum Information & Computation, 6(2):97–165, 2006

    Panos Aliferis, Daniel Gottesman, and John Preskill. Quantum accuracy threshold for concatenated distance-3 codes.Quantum Information & Computation, 6(2):97–165, 2006

    2006

  2. [2]

    Operator quantum error-correcting subsystems for self-correcting quantum memories.Physical Review A—Atomic, Molecular, and Optical Physics, 73(1):012340, 2006

    Dave Bacon. Operator quantum error-correcting subsystems for self-correcting quantum memories.Physical Review A—Atomic, Molecular, and Optical Physics, 73(1):012340, 2006

    2006

  3. [3]

    Efficient estimation of free energy differences from Monte Carlo data

    Charles H Bennett. Efficient estimation of free energy differences from Monte Carlo data. Journal of Computational Physics, 22(2):245–268, 1976

    1976

  4. [4]

    Fail fast: techniques to probe rare events in quantum error correction.arXiv preprint arXiv:2511.15177, 2025

    Michael E Beverland, Malcolm Carroll, Andrew W Cross, and Theodore J Yoder. Fail fast: techniques to probe rare events in quantum error correction.arXiv preprint arXiv:2511.15177, 2025

    arXiv 2025

  5. [5]

    Simulation of rare events in quantum error correc- tion.Physical Review A—Atomic, Molecular, and Optical Physics, 88(6):062308, 2013

    Sergey Bravyi and Alexander Vargo. Simulation of rare events in quantum error correc- tion.Physical Review A—Atomic, Molecular, and Optical Physics, 88(6):062308, 2013

    2013

  6. [6]

    Topological quantum memory.Journal of Mathematical Physics, 43(9):4452–4505, 2002

    Eric Dennis, Alexei Kitaev, Andrew Landahl, and John Preskill. Topological quantum memory.Journal of Mathematical Physics, 43(9):4452–4505, 2002. 17

    2002

  7. [7]

    Hybrid Monte Carlo.Physics letters B, 195(2):216–222, 1987

    Simon Duane, Anthony D Kennedy, Brian J Pendleton, and Duncan Roweth. Hybrid Monte Carlo.Physics letters B, 195(2):216–222, 1987

    1987

  8. [8]

    Inference from simulations and monitoring conver- gence

    Andrew Gelman and Kenneth Shirley. Inference from simulations and monitoring conver- gence. In Steve Brooks, Andrew Gelman, Galin Jones, and Xiao-Li Meng, editors,Hand- book of Markov Chain Monte Carlo, chapter 6, pages 163–174. Chapman and Hall/CRC, 2011

    2011

Show all 19 references
  1. [9]

    Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images.IEEE Transactions on pattern analysis and machine intelligence, (6):721–741, 1984

    Stuard Geman and Donald Geman. Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images.IEEE Transactions on pattern analysis and machine intelligence, (6):721–741, 1984

    1984

  2. [10]

    Introduction to Markov chain Monte Carlo

    Charles Geyer. Introduction to Markov chain Monte Carlo. In Steve Brooks, Andrew Gelman, Galin Jones, and Xiao-Li Meng, editors,Handbook of Markov Chain Monte Carlo, chapter 1, pages 3–48. Chapman and Hall/CRC, 2011

    2011

  3. [11]

    How to factor 2048 bit rsa integers with less than a million noisy qubits

    Craig Gidney. How to factor 2048 bit rsa integers with less than a million noisy qubits. arXiv preprint arXiv:2505.15917, 2025

    arXiv

  4. [12]

    Monte Carlo sampling methods using markov chains and their appli- cations

    W Keith Hastings. Monte Carlo sampling methods using markov chains and their appli- cations. 1970

    1970

  5. [13]

    Dynamical subset sampling of quantum error-correcting protocols.Physical Review Research, 6(1):013177, 2024

    Sascha Heußen, Don Winter, Manuel Rispler, and Markus Müller. Dynamical subset sampling of quantum error-correcting protocols.Physical Review Research, 6(1):013177, 2024

    2024

  6. [14]

    Blossom v: a new implementation of a minimum cost perfect matching algorithm.Mathematical Programming Computation, 1(1):43–67, 2009

    Vladimir Kolmogorov. Blossom v: a new implementation of a minimum cost perfect matching algorithm.Mathematical Programming Computation, 1(1):43–67, 2009

    2009

  7. [15]

    Even more efficient quantum computations of chem- istry through tensor hypercontraction.PRX quantum, 2(3):030305, 2021

    Joonho Lee, Dominic W Berry, Craig Gidney, William J Huggins, Jarrod R McClean, Nathan Wiebe, and Ryan Babbush. Even more efficient quantum computations of chem- istry through tensor hypercontraction.PRX quantum, 2(3):030305, 2021

    2021

  8. [16]

    Rare event simulation of quantum error-correcting circuits.arXiv preprint arXiv:2509.13678, 2025

    Carolyn Mayer, Anand Ganti, Uzoma Onunkwo, Tzvetan Metodi, Benjamin Anker, and Jacek Skryzalin. Rare event simulation of quantum error-correcting circuits.arXiv preprint arXiv:2509.13678, 2025

    2025

  9. [17]

    Equation of state calculations by fast computing machines.The journal of chemical physics, 21(6):1087–1092, 1953

    Nicholas Metropolis, Arianna W Rosenbluth, Marshall N Rosenbluth, Augusta H Teller, and Edward Teller. Equation of state calculations by fast computing machines.The journal of chemical physics, 21(6):1087–1092, 1953

    1953

  10. [18]

    Slice sampling.The Annals of Statistics, 31(3):705–767, 2003

    Radford M Neal. Slice sampling.The Annals of Statistics, 31(3):705–767, 2003

    2003

  11. [19]

    Revisiting the Gelman-Rubin diagnostic.Statistical Science, 36(4):518 – 529, 2021

    Dootika Vats and Christina Knudson. Revisiting the Gelman-Rubin diagnostic.Statistical Science, 36(4):518 – 529, 2021. 18

    2021

Pith tools

Reviewed July 30, 2026 · model on record in the stance chip above.