{"id":"87ec03a1-a48c-44ca-935e-4428e4a19007","arxiv_id":"2507.11445","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A generalized Ding-Zhuang argument proves persistence of long-range order under weak disorder for any d≥3 lattice system satisfying a Peierls condition and a local symmetry condition.","lead":"This paper builds a general mathematical framework, combining the Ding-Zhuang argument with Pirogov-Sinai theory, to prove that long-range order survives weak quenched disorder in dimensions three and higher for a wide class of lattice models. It offers physicists and probabilists a reusable pair of conditions, a Peierls condition and a local symmetry condition, that guarantee stable ordered phases and multiple coexisting Gibbs states.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The local-symmetry axiom (Definition 4) as stated does not imply the cross-region compatibility that Lemma 8 and Proposition 14 require, so the chaining estimate behind Theorem 15 is not yet a consequence of the paper's stated assumptions.","rationale":"Read in good faith, the paper is a serious generalization of Ding-Zhuang: the contour and polymer machinery is coherent, the concentration estimates are plausible, and the range of applications is substantial. The most vulnerable point is the chaining step for F^int. The reader's weakest assumption (Definition 4 is strong) is related but not identical: the issue is not merely that the axiom is hard to verify, but that as written it is insufficiently constraining for the proof of Lemma 8. This is load-bearing because Proposition 14 provides the probability estimate for F^int, one of the three events needed to get the uniform contour bound and hence the N_g Gibbs measures in Theorem 15. The gap is repairable: all models in Section 6 use global symmetries (spin flip, cyclic permutation, lattice translation), so adding a compatibility axiom would not invalidate the applications. The continuous-spin application in §6.5 contains a separate apparent sign/choice error around the constant in Eq. (105), but that concerns an example, not the central theorem, and is of the same 'repairable typo' class the reader flagged. The ergodicity/covariant-selection step at the end of Theorem 15 is also terse: if uncountable choices of limit points are made independently for each ω, the required η-covariance is not automatic; a measurable equivariant selection is needed. This is another reason the proof needs a small strengthening, but it is standard and not a fatal objection. Overall the conditional verdict stands; I would keep the reader's verdict unchanged, with the specific request to add a compatibility condition to Definition 4 and to re-verify Lemma 8.","tokens_in":32980,"tokens_out":28893,"duration_ms":373856,"concrete_test":"Re-derive Lemma 8 using only Definition 4. If the derivation is impossible, build two overlapping regions Λ1,Λ2 and two local symmetries satisfying Definition 4 (e.g. spin-flip on Λ1 and a translation-induced symmetry on Λ2 in the antiferromagnetic checkerboard model of §6.2) that disagree on Λ1∩Λ2; on a small finite torus compute Δ_{Λ1}F−Δ_{Λ2}F conditioned on the variables in Λ' of (38). If the conditional mean is nonzero, Lemma 8 fails as stated. Then add to Definition 4 the axiom that all τ_Λ are restrictions of one global local symmetry and verify that Lemma 8 and Proposition 14 go through unchanged.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 15 depends on Proposition 14, whose proof uses Lemma 8 to show that the free-energy differences Δ_{Λ1}F−Δ_{Λ2}F are subgaussian. Lemma 8 conditions on Λ' defined in Eq. (38) and asserts '(τ_{Λ1}η)_s=(τ_{Λ2}η)_s for all s∈Λ′', and then uses measure quasi-invariance to kill the conditional mean. But Definition 4 supplies, separately for each finite region Λ, a local transformation τ_Λ between H^{k1}_{η,Λ} and H^{k2}_{η,Λ}; it contains no axiom saying that τ_{Λ1} and τ_{Λ2} agree on Λ1∩Λ2 (or on the set Λ'). The paper even emphasizes in Step 3 of §4.1 that τ_{intγ} is a composition of transformations over the components of intγ, so for two different contours the associated transformations genuinely can differ on an overlap. The equality used in Lemma 8 therefore does not follow from the stated axioms. Without it, the chaining increments are not controlled and Proposition 14 is not established for the full generality of Theorem 15. A strengthening of Definition 4 to require one global local symmetry whose restrictions are τ_Λ, or an explicit proof of compatibility, would close the gap; the Section 6 applications all use global spin-flip or translation symmetries and would be unaffected.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a general framework for proving persistence of long-range order in disordered lattice systems, axiomatizing the Ding-Zhuang method through a Peierls condition and a local-symmetry condition. The main result, Theorem 15, asserts that in dimensions d≥3, for sufficiently low temperature and weak disorder, a system with N_g ground states admits at least N_g distinct η-covariant Gibbs measures, each with the corresponding ground-state probability bounded below by 1/2. The proof combines Pirogov-Sinai contour/polymer representations with concentration inequalities and a multiscale chaining argument. The framework is then applied to the random-field Ising and Potts models, an extended Edwards-Anderson model, a quenched Fredrickson-Andersen model, hard-core models on several lattices, and continuous-spin versions of the Ising and anisotropic Heisenberg models.","tokens_in":33174,"tokens_out":11256,"duration_ms":131811,"significance":"If the main theorem is correct, the paper delivers a useful unifying scheme: it recovers the Ding-Zhuang results for RFIM/RFPM and extends the method, in a single axiomatic setup, to bond disorder, hard-core models, other lattice structures, and continuous spin spaces. The explicit construction of local symmetries in Section 6 is a strength, as is the careful use of concentration and chaining estimates. The price of the generality is that the local-symmetry axiom (Definition 4) is strong and model-specific; the paper does not derive it from weaker structural conditions, and the theorem is vacuous when no such transformation exists. The central proof, however, currently contains load-bearing gaps that must be repaired before the theorem can be regarded as established.","major_comments":[{"comment":"The identity (τ_{Λ1}η)_s=(τ_{Λ2}η)_s for all s∈Λ′ is not a consequence of Definition 4. Definition 4 supplies, separately for each finite region Λ, a local transformation τ_Λ, and Section 4.1 explicitly constructs τ_intγ as a composition over the components of intγ; two different contours can therefore induce different transformations on a common overlap. Since Lemma 8 is the subgaussian input for the chaining argument in Proposition 14, the proof of Theorem 15 currently rests on an unstated cross-region compatibility property. Adding a compatibility axiom—for example, requiring that all τ_Λ are restrictions of one global local symmetry—would close the gap, and the applications in Section 6, which use global spin flips or translations, would satisfy such an axiom.","section":"Section 4.2, Lemma 8 and Eq. (38)"},{"comment":"The displayed bound P(...) ≤ exp(−ρ²/(128Cν n^{2−d/(d−1)})) does not decay in n for d≥3; as n→∞ the right-hand side tends to 1, so the subsequent summation over n cannot be made arbitrarily small. The exponent appears to have been inverted: combining the preceding line n²/diam(C0(n))² with the isoperimetric inequality would yield a bound of the form exp(−c n^{d/(d−1)}) or at least exp(−cn). Please correct the displayed inequality and verify the constants in the final summation, since this is the step that makes Proposition 14 valid.","section":"Section 4.4, Proposition 14, Eq. (65)"},{"comment":"The line 'Since Ξ^{k0}_{τintγ0 η,Λ}/Ξ^{k0}_{η,Λ}=Z^{k0}_{τintγ0 η,Λ}/Z^{k0}_{η,Λ}' is inconsistent with the definition of Ξ in Proposition 1, which contains the η-dependent prefactor exp[(e_g|Λ|+S^k_Λ(η))/T]. The quotient of the Ξ's differs from the quotient of the Z's by the factor exp[(S^k_Λ(τintγη)−S^k_Λ(η))/T], and this additional factor is not controlled by the event F_int, which only bounds the Z-ratio. The derivation of the contour-occurrence estimate exp(−ρ|γ̄|/4T) therefore needs to be reworked or the definitions adjusted.","section":"Section 5.1, proof of Theorem 15, around Eq. (71)"}],"minor_comments":[{"comment":"The compatibility condition repeats 'γ≥γ′' twice; it should read 'γ≥γ′ or γ′≥γ'.","section":"Section 3.2, Eq. (19)"},{"comment":"The probability tail bounds in (34) and the surrounding text omit the factor 2 present in Theorems 2 and 3, and the symbol ν is used for both the bounded and Gaussian cases; the notation should be aligned.","section":"Section 4.2, Proposition 6 and Corollary 7"},{"comment":"The line 'P(η∈ Fc ∩ Ic ∩ Fint) ≥ 1−3ρ > 3/4' overloads ρ, which is already the Peierls constant; a separate tolerance parameter should be used.","section":"Section 5.1, proof of Theorem 15"},{"comment":"The random-field notation is inconsistent: the text uses η^{Γ,b}_s, η^b,Γ_s, and η^c_s for what appears to be the same quantity; define one symbol and use it throughout.","section":"Section 6.3, Eqs. (95)–(96)"},{"comment":"There are typos such as 'we can also generated the four sublattice', and the description of the sublattices for the hexagonal close-packed lattice would benefit from a clearer definition.","section":"Section 6.4"}],"recommendation":"major_revision","confidential_remarks":"The three major comments are all load-bearing for Theorem 15, but they appear repairable within the manuscript's scope: a compatibility axiom for local symmetries, a corrected exponent in Proposition 14, and a rewritten comparison in the proof of Theorem 15. I would ask the authors to address these before acceptance, and to double-check the applications in Section 6 against the strengthened axiom."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper is worth taking seriously. It turns the Ding-Zhuang argument into an abstract framework — Peierls condition plus local symmetry — and runs it through Pirogov-Sinai, with a genuinely broad set of applications: RFIM, RFPM, Edwards-Anderson with random fields, hard-core/FA-1B, and continuous-spin models. The axiomatization is useful, and the translation-based symmetries for chessboard ground states are a real extension beyond spin-flip.\n\nThe main theorem as stated is not quite proven, and the stress-test note is right. Definition 4 asserts, for each finite region Λ, a local transformation τ_Λ. It does not say that τ_{Λ1} and τ_{Λ2} agree on Λ1∩Λ2. Lemma 8 needs exactly that agreement, and without it the subgaussian bound for Δ_{Λ1}F−Δ_{Λ2}F does not follow from the stated axioms. This is a genuine gap in the proof of Proposition 14, and hence in Theorem 15. It is repairable: require one global local symmetry of the disorder measure whose restriction to Λ is τ_Λ, or add a compatibility axiom. The applications all use global spin-flip or translation symmetries, so nothing in Section 6 is threatened.\n\nOther issues are minor. The compatibility condition in (19) has a typo (\"γ≥γ\" repeated), the \"1−3ρ\" probability in the proof of Theorem 15 should presumably be something like 1−3δ, and the chaining exponent in (65) deserves a cleaner derivation. The ergodicity step at the end of Theorem 15 is terse but standard; it needs a sentence or two of detail.\n\nWho is it for: rigorous statistical mechanics, especially disordered lattice systems and low-temperature phase diagrams. With the compatibility gap closed, this becomes a very useful reference. As is, it should go to a serious referee, not be desk-rejected. The referee should ask for a strengthened axiom, a revised Lemma 8/Proposition 14, and the typo fixes. I would bring it to a reading group: the framework is interesting and the gap is instructive.","headline":"A serious and readable generalization of the Ding-Zhuang argument, but the local-symmetry axiom as stated is too weak to support the chaining estimate; the main theorem needs a compatibility condition or a strengthened axiom.","tokens_in":33807,"tokens_out":3821,"would_cite":true,"duration_ms":46880,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82B20","82B26","82B44"],"pacs":[],"model":"deepseek-v4-flash","headline":"In three or more dimensions, weak disorder fails to destroy long-range order whenever the clean system has a Peierls condition and a local symmetry.","keywords":["disordered systems","long-range order","Ding-Zhuang argument","Pirogov-Sinai theory","Peierls condition","random-field Ising model","quenched disorder","Gibbs states"],"falsifier":"A concrete falsifier would be a Hamiltonian on $\\mathbb{Z}^3$ that satisfies the Peierls condition and the local-symmetry axioms of Definition 4 but whose low-temperature Gibbs measures show no magnetization exceeding $1/2$ at arbitrarily small disorder variance—e.g., a numerical study of the quenched Fredrickson-Andersen 1-blocked model, whose hard-core limit the paper claims is ordered; observing no ordered phase there would contradict the theorem's prediction.","tokens_in":32700,"feed_emoji":"🧲","tokens_out":10953,"duration_ms":119845,"temperature":0.7,"pith_summary":"This paper asks when long-range order in a lattice spin system survives the addition of frozen-in randomness. It claims that in dimensions $d \\ge 3$, order persists at low temperature and weak disorder provided two conditions hold: the disorder-free model has a Peierls condition (an excitation of size $n$ costs at least $\\rho n$ energy) and the disorder is almost invariant under local maps that swap ground states. The proof turns the Ding-Zhuang multiscale argument into a general theorem, Theorem 15, integrated with the Pirogov-Sinai contour representation. The payoff is a single checkable criterion that reproduces and unifies known results for random-field, random-bond, hard-core, and continuous-spin disordered models.","feed_headline":"Weak disorder can't destroy order in 3D","feed_subtitle":"One symmetry condition plus a Peierls bound guarantees multiple ordered phases under weak randomness.","key_machinery":"The load-bearing object is the local symmetry operation (Definition 4): for any two ground states $b_{k_1}, b_{k_2}$ and any finite region $\\Lambda$, there is a pair of transformations $(\\bar\\tau_\\Lambda, \\tau_\\Lambda)$ acting on spins inside $\\Lambda$ and on quenched parameters in a slightly enlarged region. The transformation must be local, injective, Lipschitz, and quasi-invariant in two senses: it changes the Hamiltonian only along the internal boundary of $\\Lambda$ (energy quasi-invariance), and it pushes forward the i.i.d. disorder measure to itself up to boundary effects (measure quasi-invariance). The Ding-Zhuang argument—a multiscale Peierls-type proof for disordered systems—uses these symmetries to compare partition functions of contour interiors with ground-state references, bounding three random events—$F^c$, $I^c$, and $F^{\\mathrm{int}}$—by subgaussian concentration and a coarse-grained chaining estimate. The output is an exponential decay bound on the probability of contours, which beats the entropy of contour counting and produces the long-range order of Theorem 15.","core_discovery":"Theorem 15 states that for a statistical mechanical system on $\\mathbb{Z}^d$ with $d \\ge 3$ whose disorder-free Hamiltonian satisfies the Peierls condition and admits a local symmetric operation, there are constants $T_0>0$ and $\\varepsilon_0>0$ such that for $T<T_0$ and $\\varepsilon<\\varepsilon_0$ the system has at least $N_g$ distinct $\\eta$-covariant Gibbs measures $\\{\\mu^k_\\eta\\}_{k=1}^{N_g}$. For almost every disorder realization each such measure is ordered: the limiting density of sites where the spin equals the ground-state value $b_k$ exceeds $1/2$. Conceptually the paper establishes a stability criterion: a Peierls energy condition plus an approximate local symmetry of the disorder distribution guarantees persistence of long-range order and phase coexistence. The criterion is then checked on a diverse set of models, including random-field Ising and Potts models, the Edwards-Anderson model, a quenched Fredrickson-Andersen 1-blocked model, hard-core models on several three-dimensional lattices, and continuous-spin variants such as the anisotropic Heisenberg model.","pith_inferences":["The axioms might be checkable for models the paper does not treat, such as random-field clock or $O(n)$ models with discrete spin partitions, provided a suitable map $\\bar\\tau_\\Lambda$ can be built; this is an extrapolation of the stated theorem, not a claim of the paper.","The measure quasi-invariance condition is essentially an exact local symmetry of the disorder law; physical disorder distributions that are only approximately symmetric on larger scales would need a quantitative stability analysis, which the paper leaves implicit.","Because the proof uses subgaussian concentration of the disorder, it likely extends to weakly dependent disorder with finite-range correlations by modifying the chaining step; again, the authors do not make this claim.","The theorem can be read as a design principle: a material with a clean model satisfying the Peierls condition and with disorder generated by a locally symmetric random process should keep its ordered phase, which could inform experiments on disordered magnets."],"forward_implications":["For any model in the class, the number of coexisting low-temperature Gibbs states is at least the number of periodic ground states, so the theorem turns ground-state counting into a phase-coexistence statement.","In dimensions $d\\ge 3$, order and phase coexistence are stable under small disorder for all models satisfying the two axioms, giving a uniform reason for phenomena previously proved model by model.","The local symmetry need not be a global symmetry of the clean model; translations, reflections, and cyclic permutations of internal states all qualify, as the applications to the antiferromagnetic and hard-core models show.","When the random field that couples to the order parameter is absent and only bond disorder is present, the dimension restriction relaxes to $d\\ge 2$ (Remark 2), so the mechanism distinguishes random fields from random bonds.","For continuous-spin models, the same conclusion holds once the state space is partitioned into ground-state and metastable regions and the extended Peierls condition (Definition 9) and local symmetry (Definition 10) are satisfied."],"supporting_citations":[{"why":"Supplies the original Ding-Zhuang argument for the random-field Ising and Potts models, which this paper axiomatizes and generalizes.","marker":"[32]"},{"why":"Provides the Pirogov-Sinai contour representation and the ground-state framework that the paper uses to rewrite partition functions as polymer models.","marker":"[6]"},{"why":"Introduces the multiscale and coarse-graining analysis that underlies the chaining estimate controlling fluctuation-stabilized internal regions.","marker":"[33]"},{"why":"Contains the McDiarmid and Gaussian concentration inequalities and the chaining tail inequality used in Section 4 to bound contour probabilities.","marker":"[40]"},{"why":"Gives the refined coarse-graining lemmas (Lemma 10 and Proposition 11) used to bound covering numbers of contour regions.","marker":"[42]"},{"why":"Supplies the general framework of $\\eta$-covariant Gibbs measures and the no-go constraints that motivate the dimension restriction $d\\ge 3$.","marker":"[29]"},{"why":"Provides the earlier rigorous proof of long-range order in the 3D random-field Ising model, the benchmark the simplified method extends and unifies.","marker":"[30]"},{"why":"Supplies the isoperimetric inequality applied in Proposition 14 to control the diameter of contour regions in the chaining argument.","marker":"[44]"}],"fun_headline_variants":["3D order persists under weak disorder: new proof","Generalized Ding-Zhuang: order survives weak disorder","Weak disorder can't break 3D long-range order","Peierls condition plus symmetry: order in 3D"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole theorem rests on the existence, for every finite region, of a local transformation of spins and disorder that almost preserves both the Hamiltonian and the disorder distribution while swapping the ground states; if the random-field distribution or the couplings lack such a symmetry, or boundary errors accumulate, the theorem gives nothing.","fun_headline_variants_meta":{"raw":{"variants":["3D order persists under weak disorder: new proof","Generalized Ding-Zhuang: order survives weak disorder","Weak disorder can't break 3D long-range order","Peierls condition plus symmetry: order in 3D"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00062,"raw_usage":{"total_tokens":2852,"prompt_tokens":898,"completion_tokens":1954,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":514,"completion_tokens_details":{"reasoning_tokens":1887}},"tokens_in":514,"tokens_out":1954,"duration_ms":17761,"temperature":1.0,"reasoning_tokens":1887,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:09:46.012356+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete falsifier would be a Hamiltonian on $\\mathbb{Z}^3$ that satisfies the Peierls condition and the local-symmetry axioms of Definition 4 but whose low-temperature Gibbs measures show no magnetization exceeding $1/2$ at arbitrarily small disorder variance—e.g., a numerical study of the quenched Fredrickson-Andersen 1-blocked model, whose hard-core limit the paper claims is ordered; observing no ordered phase there would contradict the theorem's prediction.","supporting_citations":[{"cited_title":"Theoreti- cal and Mathematical Physics25(3), 1185–1192 (1975) https://doi.org/10.1007/ BF01040127","cited_arxiv_id":null,"evidence_quote":"Provides the Pirogov-Sinai contour representation and the ground-state framework that the paper uses to rewrite partition functions as polymer models."},{"cited_title":"Journal of Statistical Physics34(5-6), 863–870 (1984) https://doi.org/10","cited_arxiv_id":null,"evidence_quote":"Introduces the multiscale and coarse-graining analysis that underlies the chaining estimate controlling fluctuation-stabilized internal regions."},{"cited_title":"Lecture notes in progress, available at https://web.math.princeton.edu/ rvan/APC550.pdf (2016)","cited_arxiv_id":null,"evidence_quote":"Contains the McDiarmid and Gaussian concentration inequalities and the chaining tail inequality used in Section 4 to bound contour probabilities."},{"cited_title":"Phase Transitions in Multidimensional Long-Range Random Field Ising Models","cited_arxiv_id":"2307.14150","evidence_quote":"Gives the refined coarse-graining lemmas (Lemma 10 and Proposition 11) used to bound covering numbers of contour regions."},{"cited_title":"Communications in Mathematical Physics98(2), 145–176 (1985) https://doi.org/ 10.1007/BF01220505","cited_arxiv_id":null,"evidence_quote":"Provides the earlier rigorous proof of long-range order in the 3D random-field Ising model, the benchmark the simplified method extends and unifies."}],"review_version":1}