Fluctuations in random field Ising models

Nabarun Deb; Seunghyun Lee; Sumit Mukherjee

arxiv: 2503.21152 · v3 · submitted 2025-03-27 · 🧮 math.PR · math-ph· math.MP

Fluctuations in random field Ising models

Seunghyun Lee , Nabarun Deb , Sumit Mukherjee This is my paper

Pith reviewed 2026-05-22 23:31 UTC · model grok-4.3

classification 🧮 math.PR math-phmath.MP

keywords central limit theoremBerry-Esseen boundsrandom field Ising modelStein's methodexponential familyhigh-temperature regimespin glass

0 comments

The pith

A general CLT with explicit Berry-Esseen bounds holds for linear statistics of spins from quadratic exponential families in the high-temperature regime.

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

The paper proves a central limit theorem for linear statistics of the form inner product of a vector q with an observation sigma drawn from an exponential family whose sufficient statistic is quadratic. The result supplies quantitative bounds on the distance to normality and is established in the high-temperature regime. The same machinery is specialized to random-field Ising models on several graph families and to the Hopfield model, producing both quenched and annealed statements. A reader cares because the bounds give concrete control on fluctuations without model-specific heavy lifting.

Core claim

For observations from an exponential family whose sufficient statistic is a quadratic form, the linear statistic inner product q dot sigma obeys a central limit theorem with explicit Berry-Esseen error bounds whenever the parameters lie in the high-temperature regime. The proof combines Stein's method via exchangeable pairs with Chevet-type concentration inequalities. The statement applies directly to random-field Ising models with discrete or continuous spins and yields both quenched and annealed CLTs on Erdős-Rényi, regular, dense bipartite graphs and on the Hopfield spin glass.

What carries the argument

Stein's method of exchangeable pairs combined with Chevet-type concentration inequalities applied to the quadratic sufficient statistic in the high-temperature regime.

If this is right

Quenched and annealed CLTs hold for linear statistics on Erdős-Rényi, regular, and dense bipartite graphs.
The same bounds apply to both discrete and continuous spin versions of the random-field Ising model.
Annealed and quenched central limit theorems are obtained for the Hopfield spin glass model.
The method produces explicit rates rather than mere existence of a limiting normal distribution.

Where Pith is reading between the lines

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

The same exchangeable-pair argument may extend to other quadratic-interaction models once a high-temperature condition is verified.
The explicit error bounds could be fed into downstream statistical procedures that require uniform control on fluctuations across many linear statistics.
Because the proof separates the high-temperature assumption from the graph structure, it suggests a route to CLTs on graphs whose maximum degree grows slowly with system size.

Load-bearing premise

The interaction parameters must lie inside the high-temperature regime so that the quadratic sufficient statistic satisfies the moment and contraction conditions needed for the exchangeable-pair construction to produce the stated error bound.

What would settle it

A Monte Carlo experiment on a small regular graph that computes the Kolmogorov distance between the normalized linear statistic and the standard normal and finds it larger than the paper's explicit upper bound, for parameters inside the claimed high-temperature region, would falsify the quantitative claim.

read the original abstract

This paper establishes a CLT for linear statistics of the form $\langle \mathbf{q},\boldsymbol{\sigma} \rangle$ with quantitative Berry-Esseen bounds, where $\boldsymbol{\sigma}$ is an observation from an exponential family with a quadratic form as its sufficient statistic, in the \enquote{high-temperature} regime. We apply our general result to random field Ising models with both discrete and continuous spins. To demonstrate the generality of our techniques, we apply our results to derive both quenched and annealed CLTs in various examples, which include Ising models on some graph ensembles of common interest (Erd\H{o}s-R\'{e}nyi, regular, dense bipartite), and the Hopfield spin glass model. Our proofs rely on a combination of Stein's method of exchangeable pairs and Chevet type concentration inequalities.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper gives quantitative Berry-Esseen bounds for linear statistics in quadratic exponential families via Stein's method, applied to Ising models on several graphs and the Hopfield model, but the high-temperature mapping needs explicit checks.

read the letter

This paper gives a general central limit theorem with Berry-Esseen bounds for linear statistics of the form where sigma comes from a quadratic exponential family, restricted to the high-temperature regime. They then apply it to random field Ising models on Erdős-Rényi, regular, and dense bipartite graphs, plus the Hopfield model, getting both quenched and annealed versions. The new part is the general result for quadratic families and the explicit bounds coming from Stein exchangeable pairs plus Chevet inequalities. That combination is not entirely routine, and it lets them handle several graph ensembles in one framework. The abstract frames it as more than just applying known tools to new examples. The potential issue is whether the high-temperature condition is verified quantitatively for each application. The bounds rely on the quadratic sufficient statistic satisfying moment and contraction properties that hold only in that regime. If the paper just states the assumption without mapping it to concrete ranges for inverse temperature, field strength, or edge probability, then the claims for the specific models rest on unverified steps. The stress-test note flags this, and it seems like the least secure part. On the other hand, if they do provide those ranges, the result is solid. Overall this is a technical contribution aimed at researchers in probability on graphs and spin glass models. Someone looking for explicit rates in CLTs for these systems would find it useful. The methods are standard enough that the paper should be checked carefully in review, but it looks like it deserves referee time because the general statement and the range of examples are new.

Referee Report

2 major / 2 minor

Summary. The paper establishes a CLT with quantitative Berry-Esseen bounds for linear statistics ⟨q, σ⟩ where σ is drawn from an exponential family whose sufficient statistic is a quadratic form, under a high-temperature regime. The general result is proved via Stein's method of exchangeable pairs combined with Chevet-type inequalities and is applied to random-field Ising models (discrete and continuous spins) on Erdős-Rényi, regular, and dense bipartite graphs as well as the Hopfield model, yielding both quenched and annealed statements.

Significance. If the high-temperature conditions are made fully explicit and verified for the listed graph ensembles, the work supplies a reusable quantitative fluctuation theorem for a broad class of quadratic exponential families. The combination of Stein exchangeable-pair contraction with Chevet bounds is a standard but cleanly executed approach; the breadth of examples (quenched/annealed, discrete/continuous, several graph models) adds value for applications in statistical mechanics on random structures.

major comments (2)

[§2–3 (general theorem) and §4 (applications)] The high-temperature regime is defined via a contraction condition on the exchangeable-pair covariance (presumably in the general theorem of §2 or §3). For the Erdős-Rényi, regular-graph, and Hopfield applications, the manuscript must supply explicit, uniform bounds on the inverse temperature β, external field, and edge probability (or connectivity parameter) that guarantee the required contraction constant is strictly less than 1 and that the third-moment term controlled by Chevet remains O(1/√n). If these mappings remain schematic rather than quantitative, the Berry-Esseen claim does not automatically transfer to the concrete models.
[Theorem 1.1 / main CLT statement] The statement of the Berry-Esseen bound (likely Theorem 1.1 or the main result in §3) is stated under the abstract high-temperature assumption; the paper must verify that the quadratic sufficient statistic satisfies the moment and Lipschitz conditions uniformly across the listed ensembles, or else the quantitative rate cannot be claimed for all examples simultaneously.

minor comments (2)

[Introduction and §4] Notation for the quadratic form and the linear statistic ⟨q, σ⟩ should be introduced once with a single consistent symbol set rather than re-defined in each application subsection.
[Abstract and §1] The abstract and introduction should clarify whether the high-temperature condition is uniform in the graph size n or only holds for sufficiently small β depending on n.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and the constructive major comments. The points raised concern the explicitness of the high-temperature conditions and the uniform verification of auxiliary assumptions across examples. We address each below and will incorporate the requested quantitative details in a revised version.

read point-by-point responses

Referee: [§2–3 (general theorem) and §4 (applications)] The high-temperature regime is defined via a contraction condition on the exchangeable-pair covariance (presumably in the general theorem of §2 or §3). For the Erdős-Rényi, regular-graph, and Hopfield applications, the manuscript must supply explicit, uniform bounds on the inverse temperature β, external field, and edge probability (or connectivity parameter) that guarantee the required contraction constant is strictly less than 1 and that the third-moment term controlled by Chevet remains O(1/√n). If these mappings remain schematic rather than quantitative, the Berry-Esseen claim does not automatically transfer to the concrete models.

Authors: We agree that the applications in §4 currently verify the high-temperature contraction only at the level of asymptotic regimes rather than with fully explicit numerical thresholds. In the revision we will add, for each graph ensemble, explicit upper bounds on β (and on the external-field strength where relevant) that guarantee the exchangeable-pair contraction constant is bounded by a fixed number strictly less than 1, uniformly in the graph size. We will likewise supply the corresponding explicit control on the Chevet third-moment term, confirming it is O(1/√n) under those same parameter restrictions. These bounds will be stated in a new subsection of §4 and cross-referenced from the statement of the general theorem. revision: yes
Referee: [Theorem 1.1 / main CLT statement] The statement of the Berry-Esseen bound (likely Theorem 1.1 or the main result in §3) is stated under the abstract high-temperature assumption; the paper must verify that the quadratic sufficient statistic satisfies the moment and Lipschitz conditions uniformly across the listed ensembles, or else the quantitative rate cannot be claimed for all examples simultaneously.

Authors: The general theorem in §3 assumes that the quadratic sufficient statistic satisfies a uniform third-moment bound and a Lipschitz condition with respect to the underlying measure. In the current draft these are checked only qualitatively for each model. We will revise §4 to include, for every ensemble (Erdős-Rényi, regular, dense bipartite, and Hopfield), a uniform verification that the required moment and Lipschitz constants are bounded independently of n under the same explicit high-temperature restrictions already mentioned. This will be presented in a table or short lemma that applies simultaneously to the quenched and annealed statements. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation rests on standard Stein exchangeable-pair method and Chevet inequalities applied under an externally verifiable high-temperature assumption.

full rationale

The paper's central result is a quantitative CLT with Berry-Esseen bounds obtained by combining Stein's method of exchangeable pairs with Chevet-type concentration inequalities. These are established external tools whose hypotheses (moment bounds and contraction of the exchangeable-pair covariance) are stated as assumptions on the quadratic sufficient statistic in the high-temperature regime. The applications to specific graph ensembles require only that the model parameters be chosen so those hypotheses hold; the paper does not define the target bound in terms of a fitted parameter, rename a known result, or rest the uniqueness of its approach on a self-citation chain. No step reduces the claimed bound to its own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The result rests on the standard properties of Stein's method and Chevet inequalities together with the domain assumption that the model is in the high-temperature regime; no free parameters or new entities are introduced in the abstract.

axioms (1)

domain assumption The quadratic sufficient statistic satisfies the conditions required for the exchangeable-pair construction in the high-temperature regime
Invoked to guarantee the Berry-Esseen bound holds for the linear statistics.

pith-pipeline@v0.9.0 · 5670 in / 1296 out tokens · 32108 ms · 2026-05-22T23:31:32.293397+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Assumption 2.1 (high-temperature). … ∥A_n∥ ≤ ρ … ∥A_n∥_4 ≤ ρ … ∥A_n∥_∞ ≤ ρ … SHT … commonly referred to as the high-temperature Dobrushin uniqueness condition
IndisputableMonolith/Foundation/AlexanderDuality.lean alexander_duality_circle_linking unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Theorem 2.4 … d_KS(T*_n(σ), W_n) ≲ √R1n + √(α_n R2n) + … Stein’s method of exchangeable pairs and Chevet-type concentration inequalities

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.