{"id":"882976c8-09e2-4417-8b3e-393f504edcb0","arxiv_id":"2611.17193","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":1,"one_line_summary":"A resummation of the partition function into a polymer model with weights decaying exponentially in boundary length yields convergent cluster expansions for many-body tilts with radius decay on Z^2.","lead":"This paper develops a new cluster expansion method for statistical mechanics models on Z^2 that are perturbed by many-body interactions decaying exponentially in the radius of the interaction ball, rather than in its area. The result allows rigorous analysis of systems where standard cluster expansions fail, including high- and low-temperature Ising models with long-range many-body tilts.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek/deepseek-v4-flash","headline":"No significant objection identified","rationale":"The reader's weakest assumption (simple connectivity of Λ) is correctly identified and is indeed load-bearing. The paper acknowledges this and uses it throughout. My review confirms that the proof is rigorous and the central claim is well-supported. No additional concern outweighs this. The paper is a significant contribution and the verdict ACCEPT with HIGH confidence is appropriate.","tokens_in":42933,"tokens_out":358,"duration_ms":45773,"concrete_test":"Verify that the induction hypothesis in Step 3 applies to Int(γ) for each maximal polymer (γ,η) by checking that the conditioned measure µ_Int(γ)^η is indeed a generic nearest-neighbor Gibbs measure on a finite simply connected set, as per Eq. (2.1). This is already argued via Lemma 2.1; a formal check would confirm that the potentials V are unchanged and that the boundary condition η on γ provides the necessary boundary terms.","verdict_should_be":"UNCHANGED","load_bearing_attack":"After careful review, the central argument appears sound. The induction step relies on the domain Markov property to ensure that the conditioned measure on Int(γ) is a Gibbs measure of the same form, which is justified by Lemma 2.1. The geometry lemmas (2.4, 2.6) are correctly proven and used. The bijection in Step 2 is valid. The weight bound is correctly derived. The applications follow from the main theorem and standard cluster expansion techniques. The paper acknowledges its limitations (restriction to Z^2, suboptimal constants, non-analyticity). I find no logical gap or unstated assumption that would threaten the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek/deepseek-v4-flash","summary":"This paper studies the partition function of a nearest-neighbor Gibbs measure on a finite simply connected Λ ⊂ Z^2, tilted by many-body interactions g_{x,r} supported on ℓ^∞-balls of radius r, with ‖g_{x,r}‖_∞ ≤ exp(-C_*(r+1)): exponential decay in the radius rather than the area of the ball. Theorem 1 resums Z^g_Λ into a polymer model whose polymer supports are outer boundaries of connected unions of balls (boundary b∂^v K in Eq. 2.3), decorated by spins on them, with a symmetric compatibility relation and weights bounded by exp(-C_*|γ|/10) for C_* ≥ 35. The proof is organized in four steps: expansion of the tilt into compatible patterns (Step 1); peeling of outermost boundaries through a bijection onto compatible antichains, giving Eq. (2.14) (Step 2); an induction on |V| using the domain Markov property, applied to the simply connected interiors Int(γ) (Step 3); and the weight bound via a boundary-ring counting argument (Step 4), yielding C_0 = 35. Corollary 2 rewrites log Z^g_Λ as an expectation of exp(Σ_A Ψ_A) over connected sets A with exponentially decaying local terms. Applications include convergent cluster expansions and exponential decay of correlations for product/high-temperature measures (Prop. 3.1), low-temperature Ising measures (Prop. 3.4), existence of infinite-volume limits, and an explicit example in which the classical connected-unions-of-balls expansion diverges in finite volume (Prop. 3.5).","tokens_in":38318,"tokens_out":19350,"duration_ms":106387,"significance":"The main theorem is a genuine methodological contribution: it provides a resummation scheme that tames many-body tilts with radius (rather than area) decay, using only the domain Markov property of the base measure and no stability or correlation assumptions on it, and it extends the authors' earlier dimers result [17, Prop. 8.11] to arbitrary (± signed) interactions. The proof is essentially self-contained: Lemma 2.1 and Claim 3.6 are proved in the appendix, Lemma 2.5 is quoted from a published source with citation, and the geometry lemmas (2.4, 2.6) are proved in detail, with the figures demonstrating why each feature of the boundary definition (2.3) is necessary. The constant C_0 = 35 is explicit and honestly described as non-optimized; the restriction to simply connected Λ ⊂ Z^2 is stated in the hypotheses and is genuinely needed by the induction (Fact 2.2 and Lemma 2.4(iv)), so it is a scope condition rather than a hidden assumption. Proposition 3.5 gives a concrete, falsifiable example where the classical expansion diverges in finite volume, which strengthens the motivation for the new machinery.","major_comments":[],"minor_comments":[{"comment":"The phrase 'as well as to allow nonnegative weights w(·)' is misleading: since f(p) = exp(g_{x,r}) − 1 may be negative when g_{x,r} < 0, the constrained partition function E_c in (2.13) and hence the weights w in (2.16) are in general signed. What the construction actually requires is that E_i(γ,η) = Z^{η,g}_{Int(γ)} > 0 (Eq. 2.15) so that the ratio in (2.16) is well defined; I recommend rewording this sentence.","section":"Theorem 1 statement (before Eq. 1.8)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"Note that δ = C_*/20 − 2 is negative for every C_* < 40, so the exponential bound in (1.10) is non-trivial only in the δ > 0 regime; at the minimal admissible constant C_0 = 35 the statement reduces to the identity Z^g_Λ = E[exp ΣΨ_A] with the trivial bound. The proof handles this correctly via the δ ≤ 0 branch, but a remark making the threshold C_* > 40 explicit would help readers.","section":"Corollary 2 (§2.4)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The manuscript would benefit from a copyediting pass: it contains several typos and ligature artifacts, e.g., 'inhomogeneuous' (p. 1), 'speciffication' (Section 3.4 discussion), and 'the factorkthere' in the proof of Corollary 2.","section":"Throughout","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The count in §2.1 states that the boundary ring of a ball of radius r ≥ 1 has exactly 8r sites, while Step 4 uses the looser bound 'at most 8(r+1) elements (with room)'. Since Eq. (2.20) is the key step in the weight bound, it would help to explicitly connect the two counts and note that the looser bound covers r = 0.","section":"§2.1 vs. Step 4","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The captions argue convincingly that each feature of the boundary definition (2.3) is needed, but the figure files were not available in the reviewed text; please ensure they render legibly, in particular the distinction between the blue/green (and purple/orange) sites in the three panels.","section":"Figures 1–2","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"accept"},"author_rebuttal":null,"desk_editor":null,"pith_extraction":{"msc":["82B20","82B28","60K35"],"pacs":["05.50.+q","05.70.Fh","75.10.Hk"],"model":"deepseek/deepseek-v4-flash","headline":"Radius decay and boundary polymers: cluster expansion for long-range many-body tilts","keywords":["cluster expansion","polymer model","many-body interactions","radius decay","Ising model","boundary polymers","domain Markov property","Z^2 lattice"],"falsifier":"Construct a nearest-neighbor Gibbs measure on a finite simply connected Λ ⊂ $Z^{2}$ and a tilt g_{x,r} satisfying |g_{x,r}| ≤ exp(-35(r+1)) such that the partition function Z^g_Λ does not admit a representation of the form (1.8) with weights bounded by exp(-3.5|γ|). If the induction argument fails for some such measure, the theorem would be false.","tokens_in":39599,"feed_emoji":"🧊","tokens_out":1467,"duration_ms":16225,"temperature":0.7,"pith_summary":"The paper establishes a new cluster expansion for nearest-neighbor Gibbs measures on $Z^{2}$ tilted by many-body interactions that decay exponentially in the radius of the interaction ball, rather than in its volume. Standard cluster expansions fail in this regime because a ball of radius r carries exp(c $r^{2}$) configurations, making weights too large. The authors show that the partition function can be resummed into a polymer model whose polymers are outer boundaries of connected unions of balls, with weights decaying exponentially in the boundary length. This allows convergent cluster expansions for tilted Ising models at high and low temperatures, existence of infinite-volume limits, and exponential decay of correlations. The argument uses only the domain Markov property of the base measure, so it applies to any nearest-neighbor Gibbs measure.","feed_headline":"Boundary polymers tame many-body tilts with radius decay","feed_subtitle":"New cluster expansion for 2D lattice models works when interactions decay with distance, not area","key_machinery":"The central object is the outer boundary b∂vK of a finite connected set K ⊂ $Z^{2}$, defined as the set of sites in K that are ∗-adjacent to the infinite ∗-connected component of $Z^{2}$\\K. This boundary is always connected, its ∗-hull is simply connected, and two such boundaries are either disjoint, nested, or at distance at least 2. These geometric properties allow a recursive expansion: the partition function is first expanded into patterns (balls with spins), then grouped by their outermost boundaries, and each interior partition function is expressed via the same polymer representation by induction on the domain size. The weight of a polymer is the ratio of a constrained partition function (patterns whose outermost boundary is γ) to the unconstrained interior partition function, and this ratio is bounded using the exponential decay of the tilt and the fact that |γ| ≤ 8 Σ (r+1) over balls covering γ.","core_discovery":"For any nearest-neighbor Gibbs measure on a finite simply connected subset of $Z^{2}$, and any family of interaction functions g_{x,r} with |g_{x,r}| ≤ exp(-C_* (r+1)) for C_* ≥ 35, the partition function Z^g_Λ can be expressed as a sum over polymers (γ,η) where γ is the outer boundary of a connected union of balls and η is the spin configuration on γ. The polymers have weights bounded by exp(-C_*|γ|/10) and the sum runs over compatible families. This resummation turns a many-body problem with radius decay into a polymer model with area decay, enabling convergent cluster expansions.","pith_inferences":["The technique suggests a general method for handling many-body interactions that decay with the diameter of the interaction region rather than its volume: replace the region by its outer boundary, which has much smaller cardinality. This could extend to other two-dimensional lattice models with similar geometry.","The failure of analyticity in a global prefactor (Remark 3.8) indicates that the cluster expansion here is not merely a reparameterization of a standard one; it captures a genuinely different regime where the free energy has Griffiths-type singularities despite the existence of a convergent expansion for each fixed real coupling.","The constant C_0 = 35 is not optimized; the method likely works for much smaller constants, possibly down to the threshold where the naive expansion fails. This could be tested by refining the counting arguments.","The restriction to simply connected domains is load-bearing because the induction step requires that the interior of each boundary be simply connected. Extending to domains with holes would require a different recursion or a more general notion of boundary."],"forward_implications":["For product base measures (e.g., high-temperature Ising), the polymer representation gives an absolutely convergent cluster expansion for log Z^g_Λ, uniform in the domain.","For the low-temperature Ising model with all-plus boundary conditions and spin-flip symmetric tilt, the expansion yields a convergent polymer representation in terms of Peierls contours, with exponential decay of correlations.","The expansion implies exponential decay of correlations and existence of infinite-volume limits for tilted Ising models at both high and low temperatures, with explicit bounds.","A local perturbation of the tilt (e.g., freezing spins on a finite set) creates an analytic perturbation of the free energy, allowing recovery of local statistics via differentiation.","The expansion works for any nearest-neighbor Gibbs measure, not just product measures, because it only uses the domain Markov property."],"supporting_citations":[{"why":"Provides the abstract polymer model and the Kotecký–Preiss convergence criterion used to check the expansion.","marker":"[16]"},{"why":"Supplies the standard framework for cluster expansions, including the Ursell function and the convergence condition.","marker":"[7]"},{"why":"The authors' earlier work on Grassmann variables where a version of this resummation argument first appeared.","marker":"[17]"},{"why":"Zahradník's version of Pirogov–Sinai theory, which uses polymer boundaries with weights given by ratios of interior partition functions, a key inspiration.","marker":"[25]"},{"why":"Hammond's lemma on boundary connectedness, used to prove that the outer boundary of a connected set is connected.","marker":"[13]"}],"fun_headline_variants":["Boundary polymers crack many-body tilt problem","Cluster expansion tames long-range interactions with radius decay","Ball boundaries resum tough 2D lattice partition functions","Radius decay turns many-body tilts into polymer model","Outer boundaries shrink interactions: new cluster expansion works"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The domain Λ must be simply connected (no holes) because the induction step requires that the interior of each boundary polymer be simply connected, and the recursion would break if a hole appeared.","fun_headline_variants_meta":{"raw":{"variants":["Boundary polymers crack many-body tilt problem","Cluster expansion tames long-range interactions with radius decay","Ball boundaries resum tough 2D lattice partition functions","Radius decay turns many-body tilts into polymer model","Outer boundaries shrink interactions: new cluster expansion works"]},"model":"deepseek/deepseek-v4-flash","effort":"low","cost_usd":8.4e-05,"raw_usage":{"cost":8.43752e-05,"is_byok":false,"cost_details":{"upstream_inference_cost":8.43752e-05,"upstream_inference_prompt_cost":7.10892e-05,"upstream_inference_completions_cost":1.3286e-05},"total_tokens":1059,"prompt_tokens":986,"completion_tokens":73,"prompt_tokens_details":{"audio_tokens":0,"video_tokens":0,"cached_tokens":256,"cache_write_tokens":0},"completion_tokens_details":{"audio_tokens":0,"image_tokens":0,"reasoning_tokens":0}},"tokens_in":986,"tokens_out":73,"cost_source":"openrouter_usage","duration_ms":5236,"image_pages":[],"temperature":1.0,"billed_attempts":1,"reasoning_tokens":0,"transport_provider":"openrouter","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-11-13T09:21:11.014938+00:00","model_set":{"reader":"deepseek/deepseek-v4-flash"},"pipeline_version":"pith-pipeline@v1.4.1-daily-deepseek-or-medium","input_kind":null,"input_chars":null,"input_total_chars":null,"passes_missing":null,"published_standing":null,"published_confidence":null,"objection_audit":null,"run_record":null,"basis":{"kind":"not_recorded","source":null,"chars_given":null,"chars_total":null},"receipt":null,"falsifier":"Construct a nearest-neighbor Gibbs measure on a finite simply connected Λ ⊂ $Z^{2}$ and a tilt g_{x,r} satisfying |g_{x,r}| ≤ exp(-35(r+1)) such that the partition function Z^g_Λ does not admit a representation of the form (1.8) with weights bounded by exp(-3.5|γ|). If the induction argument fails for some such measure, the theorem would be false.","supporting_citations":[{"cited_title":"Friedli and Y","cited_arxiv_id":null,"evidence_quote":"Supplies the standard framework for cluster expansions, including the Ursell function and the convergence condition."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the abstract polymer model and the Kotecký–Preiss convergence criterion used to check the expansion."},{"cited_title":"Laslier and E","cited_arxiv_id":null,"evidence_quote":"The authors' earlier work on Grassmann variables where a version of this resummation argument first appeared."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Zahradník's version of Pirogov–Sinai theory, which uses polymer boundaries with weights given by ratios of interior partition functions, a key inspiration."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Hammond's lemma on boundary connectedness, used to prove that the outer boundary of a connected set is connected."}],"lane_accuracy":{"status":"not yet measured","lane":{"model":"deepseek/deepseek-v4-flash","input_kind":null},"url":"https://pith.science/accuracy"},"review_version":1}