Pith. sign in

REVIEW 2 major objections 5 minor 33 references

The self-correcting memory of the 3D Wegner gauge theory remains exponentially long-lived under every sufficiently small local perturbation, proved by a new gauge-averaging method.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 09:50 UTC pith:BHOZNKMH

load-bearing objection Gauge averaging is a genuine new method and the stability theorem is a major step for self-correcting memories; the only real issue is a load-bearing but standard bound left unproved. the 2 major comments →

arxiv 2607.20605 v1 pith:BHOZNKMH submitted 2026-07-22 quant-ph cond-mat.stat-mechcond-mat.str-elhep-phmath-phmath.MP

Perturbatively Stable Self-Correcting Classical Memory from Gauge Averaging

Ryan Thorngren This is my paper
classification quant-ph cond-mat.stat-mechcond-mat.str-elhep-phmath-phmath.MP
keywords self-correcting memorygauge averagingWegner gauge theoryperturbative stabilitycluster expansioncontour countingtopological order1-form symmetry
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 self-correcting classical memory of the three-dimensional Wegner gauge theory is stable against every sufficiently small local perturbation of the Hamiltonian. The new 'gauge averaging' method replaces a symmetry-breaking perturbation by an exactly symmetric effective interaction: the Gibbs state is averaged over the model's local vertex symmetry, and a cluster expansion shows the result is exponentially local whenever the perturbation is weak relative to temperature. The proof then runs a contour-counting bottleneck estimate on the gauge-averaged measure, which transfers directly to the original perturbed measure because the topological sectors used as memory states are symmetric. If correct, this establishes a genuine self-correcting memory phase whose stability grows with temperature up to a critical value, and it places the predicted phase boundary within about ten percent of numerical estimates.

Core claim

The paper's central claim, Theorem 1, states that for all 0<T<1/log 5 and η>log 6+1 there is a constant C_stab(T,η)>0 such that every perturbation V with β|V|_η < C_stab leaves the Gibbs state of the 3D Wegner gauge theory with exponentially long-lived self-correcting memory. The proof works by gauge averaging: averaging the perturbed Gibbs state over the local vertex symmetry yields an exactly symmetric, exponentially local effective interaction V_eff (via a convergent cluster expansion), so the contour-counting bottleneck argument for the unperturbed model goes through. Because the memory sectors are gauge-symmetric, the exponential mixing-time bound transfers to the original perturbed mea

What carries the argument

The central object is 'gauge averaging', the operation ρ_G = E_g (g ρ g^{-1}) that averages the Gibbs state over the finite local symmetry group generated by the vertex star operators A_v = ∏_{e: v∈∂e} X_e. For classical commuting Hamiltonians this produces an effective symmetric interaction via e^{-βV_eff} = E_g(e^{-βV}). The argument is carried by three tools: a cluster expansion (with a standard convergence criterion) that proves V_eff is exponentially local and small when β|V|_η is below a threshold; the bottleneck lemma, which lower-bounds the mixing time of any local Markov chain by the ratio μ(A)/μ(∂A) for a symmetric 'bottle' A; and a contour-counting estimate that controls the cost

Load-bearing premise

The proof of the effective local interaction V_eff relies on an unproved combinatorial counting bound from the literature: the number of connected sets of l local generators containing a fixed generator grows no faster than (eΔ)^l; if this bound fails, the cluster expansion may not converge and the whole argument collapses.

What would settle it

Enumerate connected sets of l vertices in the infinite 3D cubic lattice that contain a fixed vertex, for l up to about 30, and check whether the count exceeds (6e)^l ≈ (16.31)^l; a single violation would invalidate the convergence proof of the gauge-averaged cluster expansion. Alternatively, simulate a standard single-spin-flip Markov chain on an L×L×L Wegner model with a small random perturbation satisfying β|V|_η below the claimed threshold and check whether the mixing time grows exponentially in L; observing polynomial growth would refute Theorem 1.

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

If this is right

  • The 3D Wegner gauge theory is a genuine self-correcting memory phase: below the proven temperature threshold, every sufficiently weak local perturbation leaves an exponentially long (in system size) memory lifetime.
  • The proven stability wedge grows with temperature, so the memory is fluctuation-stabilized rather than merely low-temperature-stable; robustness increases up to T_c = 1/log 5.
  • Gauge averaging provides a rigorous mechanism by which an explicitly broken extensive local symmetry is exactly restored in the effective Hamiltonian, giving a precise sense of emergent 1-form symmetry.
  • For a magnetic-field perturbation, the theorem predicts the phase boundary near the zero-temperature line with a slope bound within ~10% of the numerically measured value, a quantitative check of the method.
  • The proof applies to any local Markov dynamics with a finite range that satisfies detailed balance, not just single-site updates.

Where Pith is reading between the lines

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

  • This suggests a general criterion for fluctuation-stabilized memory: any classical Hamiltonian with an extensive local symmetry and a symmetric low-temperature ordered phase may remain self-correcting under weak symmetry-breaking perturbations, with a stability wedge set by the local symmetry's degree. Testing this on simpler models would separate the role of topological order from that of the sym
  • Because the convergence proof relies on a cited counting bound for connected sets of generators, one could numerically verify this bound on the 3D cubic lattice; if the true growth is faster, the cluster expansion might still converge for specific perturbations, suggesting the proven stability region is not optimal.
  • The same gauge-averaging trick may be useful for quantum systems at finite temperature: averaging over a local symmetry group in a quantum Gibbs state could similarly produce a symmetric effective Hamiltonian, potentially extending self-correcting memory arguments to certain quantum codes.
  • The quantitative comparison near T=h=0 hints that the true critical slope could be closer to 4.5 than the proven bound 4.9; testing whether gauge averaging becomes tight at higher order in the cluster expansion could sharpen the bound.

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 / 5 minor

Summary. The paper develops a method called “gauge averaging” to prove that the self-correcting classical memory of the 3D Wegner Z2 gauge theory is stable to arbitrary sufficiently small local perturbations. The method replaces a symmetry-breaking perturbation V by a formally gauge-averaged effective interaction V_eff, and gives a cluster-expansion theorem (Theorem 3) establishing that V_eff is exponentially localized and gauge invariant provided β|V|η is below a computable threshold. The paper then proves a Peierls/bottleneck theorem (Theorem 4) on the gauge-averaged measure, showing that for 0<T<1/log 5 and η>log 6+1, any finite-range detailed-balance Markov chain has exponentially long mixing time in system size. A special case, uniform magnetic field perturbation, is analyzed separately with an explicit formal cluster expansion and a quantitative comparison to the numerical critical slope.

Significance. If the theorems are correct, this is a significant result: it upgrades the 3D Wegner gauge theory from a model with self-correcting memory at a parameter point to a model with a self-correcting memory phase, stable to all sufficiently weak local perturbations. The gauge-averaging method is novel and potentially transferable to other extensively symmetric classical systems. The paper contains detailed, explicit derivations of the Kotecký–Preiss convergence criterion and the Peierls bottleneck bound, with no fitted parameters; the comparison with the numerical phase boundary in the field-perturbation case is a genuine falsifiable check. However, as detailed below, the central cluster-expansion convergence relies on a combinatorial bound that is stated without proof and with a questionable citation. This is a load-bearing gap, not merely a presentation issue.

major comments (2)
  1. [Appendix A, Eq. (59)] The convergence of the cluster expansion, and hence the existence of an exponentially local, symmetric V_eff, depends on the bound A(l) ≤ (eΔ)^l for the number of connected sets J of local generators of size l containing a fixed generator. This is stated without proof and attributed to “Lemma 9 of [26]”. Reference [26] is about dense random subgraphs of the hypercube, whose degree grows with the dimension/volume; it is not evident that the cited lemma applies to the bounded-degree symmetry graph used here. The bound is standard and likely true, but it is load-bearing: Eq. (58)–(63) and the KP criterion (43) all fail if it does not hold. A direct proof, or a correct citation for the bounded-degree graph case, must be supplied.
  2. [Theorem 4, Eqs. (95)–(103)] The geometric argument bounding δ_Γ V_eff is sketched and needs more precision. In particular, Eq. (99) assumes that any gauge-invariant support X contributing to a difference across a loop Γ must contain a contractible cycle linking Γ, and then bounds the contribution by paths of length at most (3/2)r from Γ. The primal/dual linking conventions, the choice of the bounding cube, and the path-length inequality need to be stated explicitly, especially on the 3-torus and for multi-component Γ. This bound underlies the central tension-renormalization inequality |δ_Γ V_eff| ≤ C_perim(m)|V_eff|_m|Γ| + R_m, and therefore controls the Peierls factor q in Eq. (104).
minor comments (5)
  1. [Special case: Magnetic Field Perturbation] The quantitative claims τ_c ≥ 1/5 and S_c ≤ 1/arctanh(1/5) ≈ 4.9 are advertised as a comparison with numerics, but the derivation is only sketched: it relies on electric–magnetic duality and a statement that the generated Wilson loops in V_eff “remain confined” without proof. If these claims are to remain in the paper, they should be labeled as heuristic or provided with a proof.
  2. [Appendix A, Eq. (59)] The citation “see Lemma 9 of [26]” seems mismatched with the bibliographic entry, which concerns random subgraphs of the hypercube rather than general bounded-degree graphs. The authors should reconcile this in revision, either by replacing the citation or by proving the bound directly.
  3. [Notation] The notation l(X), l(γ), l(γ) and “l∗” is overloaded; in particular l(γ) is used both for the size of the hull and later as a loop length. A table of notation or a change of symbols would improve readability.
  4. [Theorem 2 / Theorem 3] The statement of Theorem 2 requires b < η − m − log Δ − 1, which is vacuous for η ≤ m + log Δ + 1. This is not an error, but it should be stated that the useful regime is η > m + log Δ + 1, as is later used in Theorem 4.
  5. [Figure 1] The shaded “dragon” region is informal; no boundary is derived. A caption clarifying that this region is only where the method ceases to apply, not necessarily where the memory breaks down, would avoid overclaiming.

Circularity Check

0 steps flagged

No significant circularity: the stability claim is derived via a convergent cluster expansion and a Peierls bound; no step reduces to its inputs, and the numerical comparison is external. The only flagged dependency—the unproved combinatorial bound A(l)≤(eΔ)^l at Eq. (59)—is a correctness risk, not circularity.

full rationale

Walking the derivation chain: the unperturbed Peierls argument (Eq. (12)) is self-contained, using a path-counting bound and 5e^{-β}<1; the bottle/bottleneck sets are constructed geometrically in Appendix B and the bottleneck lemma is proved via conductance (Lemma 1). The gauge-averaging theorem defines V_eff from V through Eqs. (19)/(28) and proves locality via the Kotecký–Preiss criterion; the constants C_gauge are fixed by explicit inequalities (58)–(65), with no parameter fitted to the memory-stability conclusion. In Theorem 4, V_eff enters only as a renormalized loop tension, and q<1 follows from the chosen inequalities b < min(η−m−log6−1, (β−log5)/C_perim(m)) (Eq. (108))—again a derivation, not an assumption of the conclusion. The comparison with [9] (τ_c≈0.22 vs. τ_c≥1/5) is ex post and does not set any constant. The genuinely load-bearing external input is the combinatorial bound A(l)≤(eΔ)^l at Eq. (59), cited to [26, Lemma 9]; it is not stated or proved in the paper and its applicability to the Wegner generator graph is not demonstrated, so it is a legitimate rigor/correctness concern, but it is not a circular reduction and does not involve self-citation. Score 0.

Axiom & Free-Parameter Ledger

0 free parameters · 5 axioms · 0 invented entities

No fitted constants or new physical entities. The proof parameters η,m,b and constants C_gauge,C_stab are part of the theorem's hypotheses and norms, not empirical fits. The only 'entity' introduced is the derived effective interaction V_eff, which is a mathematical construct with no new degrees of freedom.

axioms (5)
  • standard math Kotecký–Preiss cluster expansion criterion and Fernández–Procacci bounds guarantee absolute convergence and locality of the polymer cluster expansion.
    Used in Theorem 3 to bound K_{b,m} and |V_eff|_m; assumptions are standard and cited to [19,25].
  • standard math Number of connected sets of size l in a bounded-degree graph containing a fixed vertex obeys A(l) ≤ (eΔ)^l (cited to [26, Lemma 9]).
    Load-bearing for convergence; cited but not stated in Appendix A near Eq. (59).
  • standard math Bottleneck lemma for reversible Markov chains: if all exits from A pass through ∂A, then t_mix ≥ (1/4) μ(A)/μ(∂A).
    Used to convert bottleneck-ratio bounds into mixing time; cited to [27, Thm 7.4].
  • standard math Peierls loop counting: number of long syndrome loops of length l is bounded by 5^l on the cubic lattice.
    Used in the unperturbed and perturbed Peierls estimates, Eq. (12).
  • domain assumption Electric-magnetic duality identifies the gauge-averaged magnetic-field expansion with the unperturbed Wegner polymer partition function.
    Used only in the special-case estimate τ_c ≥ 1/5; standard in lattice gauge theory but not proven in the paper.

pith-pipeline@v1.3.0-alltime-deepseek · 15574 in / 22492 out tokens · 184286 ms · 2026-08-01T09:50:31.978036+00:00 · methodology

0 comments
read the original abstract

We show that the self-correcting memory in 3d Wegner gauge theory is stable to arbitrary small enough perturbations of the Hamiltonian. Our proof relies on a new method we dub ``gauge averaging'', which gives conditions under which explicitly broken gauge symmetries are effectively restored by fluctuations. These conditions show that the self-correcting memory phase is fluctuation stabilized, with its robustness to perturbations increasing with increasing temperature, up to some $T_c > 0$.

Figures

Figures reproduced from arXiv: 2607.20605 by Ryan Thorngren.

Figure 1
Figure 1. Figure 1: FIG. 1. Schematic phase diagram of stable memory as a func [PITH_FULL_IMAGE:figures/full_fig_p001_1.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

33 extracted references · 3 canonical work pages

  1. [1]

    Full details will be given in the appendix as Theorem

  2. [2]

    Landon-Cardinal and D

    O. Landon-Cardinal and D. Poulin, Local topological or- der inhibits thermal stability in 2d, Physical review let- ters110, 090502 (2013)

    2013

  3. [3]

    B. J. Brown, D. Loss, J. K. Pachos, C. N. Self, and J. R. Wootton, Quantum memories at finite temperature, Re- views of Modern Physics88, 045005 (2016)

    2016

  4. [4]

    bottleneck

    Here we will sketch the general idea. To prove there is an exponential memory lifetime, fol- lowing [8] we use the notion of a “bottleneck”. This may be defined for a general distributionµon a configuration space Ω and applies to any suitable class of local Markov dynamics satisfying detailed balance with respect toµ. We sayµhas a bottleneck if there is a...

  5. [5]

    Bravyi and B

    S. Bravyi and B. Terhal, A no-go theorem for a two- dimensional self-correcting quantum memory based on stabilizer codes, New Journal of Physics11, 043029 (2009)

    2009

  6. [6]

    F. J. Wegner, Duality in generalized ising models and phase transitions without local order parameters, Journal of Mathematical Physics12, 2259 (1971)

    1971

  7. [7]

    L. E. Thomas, Bound on the mass gap for finite volume stochastic ising models at low temperature, Communica- tions in Mathematical Physics126, 1 (1989)

    1989

  8. [8]

    F. Martinelli, Lectures on glauber dynamics for discrete spin models, inLectures on probability theory and statis- tics: Ecole d’et´ e de probailit´ es de saint-flour xxvii-1997 (Springer, 2004) pp. 93–191

    1997

  9. [9]

    A. M. Somoza, P. Serna, and A. Nahum, Self-dual criti- cality in three-dimensional z 2 gauge theory with matter, Physical Review X11, 041008 (2021)

    2021

  10. [11]

    Stahl, B

    C. Stahl, B. Placke, V. Khemani, and Y. Li, Slow mixing and emergent one-form symmetries in three-dimensional z2 gauge theory, arXiv preprint arXiv:2601.06010 (2026)

    arXiv 2026

  11. [12]

    Roberts, B

    S. Roberts, B. Yoshida, A. Kubica, and S. D. Bartlett, Symmetry-protected topological order at nonzero temperature, Physical Review A96, 10.1103/phys- reva.96.022306 (2017)

    doi:10.1103/phys- 2017

  12. [13]

    J. B. Kogut, An introduction to lattice gauge theory and spin systems, Reviews of Modern Physics51, 659 (1979)

    1979

  13. [14]

    M. B. Hastings and T. Koma, Spectral gap and expo- nential decay of correlations, Communications in Math- ematical Physics265, 781–804 (2006)

    2006

  14. [15]

    Castelnovo and C

    C. Castelnovo and C. Chamon, Topological order in a three-dimensional toric code at finite temperature, Phys- ical Review B78, 10.1103/physrevb.78.155120 (2008)

    work page doi:10.1103/physrevb.78.155120 2008

  15. [16]

    Fradkin and S

    E. Fradkin and S. H. Shenker, Phase diagrams of lattice gauge theories with higgs fields, Physical Review D19, 3682 (1979)

    1979

  16. [17]

    surface tension

    (assuming low enough temperature 5e −β <1) µ(∂A[C])≤µ(A[C])L 3 L−1X l=⌈L/4⌉ 5le−βl ≤µ(A[C])L 3 (5e−β)L/4 1−5e −β (12) 4 Since there are 8 possible [C] onT 3, and eachA[C] is dis- joint, at least one of them hasµ(A[C])<1/2. Therefore we can apply the bottleneck lemma, and we find a mixing time which diverges exponentially as (eβ/5)L/4/L3. Note that this gi...

  17. [18]

    Tupitsyn, A

    I. Tupitsyn, A. Kitaev, N. Prokof’Ev, and P. Stamp, Topological multicritical point in the phase diagram of the toric code model and three-dimensional lattice gauge higgs model, Physical Review B—Condensed Matter and Materials Physics82, 085114 (2010)

    2010

  18. [19]

    Serna, A

    P. Serna, A. M. Somoza, and A. Nahum, Worldsheet patching, 1-form symmetries, and ”landau-star” phase transitions (2024), arXiv:2403.04025 [cond-mat.str-el]

    Pith/arXiv arXiv 2024

  19. [20]

    Peierls, On ising’s model of ferromagnetism, inMath- ematical proceedings of the cambridge philosophical soci- ety, Vol

    R. Peierls, On ising’s model of ferromagnetism, inMath- ematical proceedings of the cambridge philosophical soci- ety, Vol. 32 (Cambridge University Press, 1936) pp. 477– 481

    1936

  20. [21]

    Agrawal, L

    R. Agrawal, L. F. Cugliandolo, L. Faoro, L. B. Ioffe, and M. Picco, Geometric phase transition of the three- dimensional z2 lattice gauge model, Physical Review Let- ters135, 10.1103/33q3-g68k (2025)

    work page doi:10.1103/33q3-g68k 2025

  21. [22]

    Fern´ andez and A

    R. Fern´ andez and A. Procacci, Cluster expansion for abstract polymer models. new bounds from an old ap- proach, Communications in Mathematical Physics274, 123 (2007)

    2007

  22. [23]

    Lake and S

    E. Lake and S. Ro, Squeezing codes: robust fluctuation- stabilized memories (2026), arXiv:2509.20730 [cond- mat.stat-mech]

    Pith/arXiv arXiv 2026

  23. [24]

    Elitzur, Impossibility of spontaneously breaking local symmetries, Phys

    S. Elitzur, Impossibility of spontaneously breaking local symmetries, Phys. Rev. D12, 3978 (1975)

    1975

  24. [25]

    Gaiotto, A

    D. Gaiotto, A. Kapustin, N. Seiberg, and B. Willett, Generalized global symmetries, Journal of High Energy Physics2015, 10.1007/jhep02(2015)172 (2015)

    doi:10.1007/jhep02(2015)172 2015

  25. [26]

    S. D. Pace and X.-G. Wen, Exact emergent higher-form symmetries in bosonic lattice models, Physical Review B 108, 10.1103/physrevb.108.195147 (2023)

    work page doi:10.1103/physrevb.108.195147 2023

  26. [27]

    Liu, W.-T

    Y.-J. Liu, W.-T. Xu, F. Pollmann, and M. Knap, Information-theoretic principle of emergent 1-form sym- metries (2026), arXiv:2502.17572 [quant-ph]

    Pith/arXiv arXiv 2026

  27. [28]

    Koteck` y and D

    R. Koteck` y and D. Preiss, Cluster expansion for ab- stract polymer models, Communications in Mathemat- ical Physics103, 491 (1986)

    1986

  28. [29]

    McDiarmid, A

    C. McDiarmid, A. Scott, and P. Withers, The component structure of dense random subgraphs of the hypercube 6 (2021), arXiv:1806.06433 [math.CO]. [27] D. A. Levin and Y. Peres,Markov chains and mixing times(American Mathematical Society, 2026). Appendix A: Classical Gauge Averaging and Cluster Expansions

    Pith/arXiv arXiv 2021

  29. [30]

    dependent

    Definitions Definition 1.A local symmetry acting on a set Λ of degrees of freedom is a pair (G, G 0) of finite abelian groupG and a generating setG 0 •For eachx∈Λ there is some local symmetryg∈Gwhose support overlapsx. •For each operatorV X which is diagonal in the classical basis and supported inXand eachg∈G,gV X g−1 is also diagonal in the classical bas...

  30. [31]

    KP activity

    Proof of the Gauge Averaging Theorem Theorem 3.Let(G, G 0)be a local symmetry satisfying Definition 1. Suppose it has the property that each local generatorg 0 ∈G 0 has support overlapping the supports of at most∆other local generators. Then for all0< b < η−m−log ∆−1, there is a constantC gauge(η, m, b)>0such that if β|V| η < Cgauge(η, m, b), (35) then th...

  31. [32]

    If each component of∂Shas length< εL, then all but at mostMcomponents of∂S ′ have length< εL

  32. [33]

    long loops

    The combined length of these “long loops” is less thanL/2

  33. [34]

    small” setsXwithl(X)< Land “large

    The decoded homology classes are equal:[C(S)] = [C(S ′)]. Proof.There is a maximum numbera 0R3 of disjoint dual lattice loops (which occupy plaquettes) which neighbor an edge inB. We can takeM=a 0R3, then 1 will automatically be satisfied, because only these loops may differ between ∂Sand∂S ′. Furthermore, if all loops in∂Shave length< εL, the total lengt...