{"id":"fe0f59fb-cf29-48ed-aa54-f08c3cbfaede","arxiv_id":"2505.11092","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Under diffusive scaling, the empirical density of the generalized KMP, discrete KMP, and harmonic models converges to the solution of the heat equation with a model-dependent diffusion coefficient.","lead":"This paper proves that three families of random spin models, including generalized KMP and harmonic models, converge to the heat equation at large scales. The result extends hydrodynamic limit theory to systems with unbounded occupation variables by exploiting a monotonicity property called attractiveness.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proof of absolute continuity uses wrong LLN mean: under ν_ρ the limit is 2sρ for gKMP and Harm, so (4.8) is false as written; repairable by replacing ρhat with (2s∨1)ρhat.","rationale":"The paper's central claim is a clean and well-scoped hydrodynamic limit for three attractive gradient spin models, and the entropy-method strategy is standard and largely correct. The reader's verdict of ACCEPT with high confidence is close to justified, and the weakest assumption identified (stochastic domination by an invariant measure) is indeed the main limitation and is honestly stated. However, a careful check of the absolute-continuity argument reveals a real, specific error: the law-of-large-numbers bound in Section 4.2.2 ignores the factor 2s in the mean of the invariant measures for gKMP and Harm. As written, this makes the asserted Q*-almost-sure bound |<π_t,G>| ≤ ρhat||G||_L1 false for s>1/2, even for the simple initial measure μ_N = ν_{ρhat}. This does not undermine the hydrodynamic limit itself, because the error is confined to an intermediate bound and can be fixed by replacing ρhat with the correct finite constant; the rest of the proof, including the characterization of limit points as weak solutions of the heat equation and the uniqueness argument, does not depend on the specific constant. Since the theorem is very likely true but the published proof contains an incorrect line in a key lemma, the appropriate verdict is CONDITIONAL acceptance pending the stated correction. No other load-bearing defects were found: the attractiveness proofs for dKMP and Harm match the cited criteria, the carré-du-champ estimates are valid, and the tightness and martingale arguments are sound apart from the constant issue.","tokens_in":20253,"tokens_out":23451,"duration_ms":228173,"concrete_test":"Independently compute E_{ν_ρ}[η_0] from (2.3) for gKMP and from (2.9) for Harm; for both, the m=1 moment is 2sρ. Then test the proof's claim by taking μ_N = ν_{ρhat} with s=1 and ρhat=1: the initial empirical measure converges to 2 du, so <π_t^N,1> converges in probability to 2, contradicting the asserted bound <π_t,1> ≤ 1 in (4.8). Finally, replace the constant ρhat in (4.8) by 2sρhat (or ρhat ∨ 2sρhat) and verify that the subsequent passage through Proposition B.1 and the change-of-variable estimate in Section 4.2.3 uses only finiteness of the constant, so the corrected proof goes through unchanged.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The proof of Theorem 2.8 contains a concrete parametrization error in Section 4.2.2. The authors claim in (4.8) that every limit point Q* satisfies |<π_t,G>| ≤ ρhat ||G||_L1 for all continuous G, and justify this by the law of large numbers under the dominating invariant measure ν_{ρhat}. But for the generalized KMP model, (2.3) with m=1 gives E_{ν_ρ}[η_0] = 2sρ, not ρ; for the Harmonic model, (2.9) with m=1 gives the same factor 2sρ. Consequently, under ν_{ρhat}, the empirical measure <π_t^N, |G|> converges in probability to 2s ρhat ||G||_L1. For s>1/2 — which is allowed for both gKMP and Harm — the event {<π_t^N,|G|> − ρhat||G||_L1 > ε} has probability tending to 1, not 0, so the asserted limsup bound in Section 4.2.2 is false. A direct example is μ_N = ν_{ρhat}: Assumption 2.6 holds trivially, yet the limiting measure is 2sρhat du, which violates (4.8) when 2s>1. The error is isolated to the absolute-continuity step: Proposition B.1 only needs some finite constant, so replacing ρhat by 2sρhat (or ρhat ∨ 2sρhat) repairs the argument. The diffusion coefficients, the martingale estimates, and the uniqueness argument are unaffected, so the main theorem is likely still correct, but the proof as written contains an incorrect step.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves hydrodynamic limits for three one-dimensional conservative spin models with unbounded occupation variables: the generalized KMP model, the discrete KMP model, and the harmonic models. Under diffusive scaling and for initial laws satisfying Assumption 2.6 (association to a profile plus stochastic domination by an invariant measure), Theorem 2.8 states that the empirical measure converges in probability to the unique weak solution of the heat equation, with diffusion coefficients D=1/2 for the two KMP-type models and D=1/(2s) for the harmonic models. The strategy combines the gradient identity (2.13) with a proof of attractiveness (Section 3) and the entropy method; tightness uses a common carré du champ bound (Lemma 4.1), and the limit points are characterized via Dynkin martingales. The spin restriction s≥1/2 for the harmonic models and the initial-measure restriction are stated explicitly by the authors.","tokens_in":20651,"tokens_out":20217,"duration_ms":183904,"significance":"The paper is a solid contribution to the hydrodynamic limit theory for unbounded spin systems. Its strengths are explicit: the diffusion coefficients are computed directly from the generators in (2.12), the gradient identities are given in detail, and the attractiveness of all three models is proved in full, including a novel verification for the harmonic family via the Gobron-Saada criterion. The common carré du champ estimate (4.6) reduces the tightness proof to a uniform second-moment bound and is transparently adapted to each model. If the identified gap in Section 4.2.2 is repaired, the main theorem provides a unified treatment of three models that previous methods handled only under stronger assumptions. The authors are also honest about the limitations imposed by Assumption 2.6 and by the harmonic spin range.","major_comments":[{"comment":"The inequality (4.8) is asserted with the wrong constant. For the generalized KMP model, (2.3) with m=1 gives E_{ν_ρ}[η_0]=2sρ, and for the harmonic model, (2.9) with m=1 gives the same value 2sρ; the discrete KMP model has mean ρ. Therefore, under the dominating measure ν_{ρhat}, the law of large numbers gives <π_t,|G|> → 2s ρhat ||G||_{L^1} for gKMP and Harm, not ρhat ||G||_{L^1}. For s>1/2, which is allowed for both gKMP and Harm, the event used in (4.8) has probability tending to 1 rather than 0; for example μ_N=ν_{ρhat} satisfies Assumption 2.6 and produces a limit violating (4.8). This is a genuine error in a load-bearing step, because (4.8) is the only input that yields absolute continuity of the limit measures. The repair is local: replace ρhat everywhere in (4.8) and in the subsequent supremum argument by C=(2s∨1)ρhat. Since Proposition B.1 only requires a finite constant, the rest of the characterization, the martingale estimates, and the uniqueness argument are unaffected.","section":"Section 4.2.2, Eq. (4.8)"}],"minor_comments":[{"comment":"The set {gKMP, dKPM, Harm} should read {gKMP, dKMP, Harm}; the same typo appears in the paragraph immediately before Theorem 2.8.","section":"Theorem 2.8"},{"comment":"The displayed equality for D_gKMP(η_x−η_{x+1})^2−L(η_xη_{x+1}) omits the cross term −2Ips η_x η_{x+1}; the subsequent inequality remains valid because this omitted term is negative, but the equality as written is false.","section":"Appendix A.1"},{"comment":"The notation ||ΔG||_∞ in Proposition 4.2 refers to the discrete Laplacian Δ_N G defined there, but the same symbol Δ is also used for the continuum Laplacian in Definition 2.7; a footnote or a repeated definition would remove the ambiguity.","section":"Section 4.1 and Proposition 4.2"}],"recommendation":"major_revision","confidential_remarks":"The error in Eq. (4.8) is real but localized; I would not treat it as evidence against the main theorem, and with the constant C=(2s∨1)ρhat the argument appears complete. The paper is within the scope of the journal and the contribution is substantial. I recommend returning it for a revision rather than rejecting."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Punchline: the paper does what it says — hydrodynamic limits for generalized KMP, discrete KMP, and harmonic models — and the main novelty, attractiveness for the harmonic family, is real. It deserves a serious referee, but the proof has a wrong constant in the absolute-continuity step that needs fixing.\n\nThe genuinely new part is the verification of attractiveness and gradient structure for these three unbounded spin systems. Theorem 3.5 (attractiveness for harmonic models with s≥1/2) is the main technical work; the monotonicity argument via digamma is plausible. The gradient identity (2.13) is clean, and the diffusion coefficients (2.12) are computed from the generators, not fitted. The paper is honest about the cost: Assumption 2.6 restricts initial measures to be stochastically dominated by an invariant measure, and the harmonic restriction s≥1/2 is stated as technical.\n\nThe soft spot is real but local. In Section 4.2.2, equation (4.8) claims every limit point satisfies |<π_t,G>| ≤ ρ_hat ||G||_L1. The proof uses the LLN under ν_{ρ_hat}, but for generalized KMP and harmonic models, E_{ν_ρ}[η_0] = 2sρ, not ρ (equations (2.3) and (2.9) with m=1). For s>1/2 — allowed for gKMP and needed for harmonic — <π^N,|G|> converges to 2sρ_hat ||G||, so the event in (4.8) has probability going to one for G with positive integral. The statement is false as written. The fix is immediate: replace ρ_hat by C=ρ_hat·(2s∨1) (or simply 2sρ_hat for s≥1/2, with ρ_hat for dKMP). Proposition B.1 only needs a finite constant, so absolute continuity of limit points and the identification with the heat equation are unaffected. This is a repairable error, not a fatal gap.\n\nCitation pattern looks fair; the relevant literature on zero-range, KMP, and symmetric inclusion is cited, and self-citation is not load-bearing. No invented parameters. For readers in mathematical statistical physics, this is a useful addition: it clears a class of models where standard techniques struggle. I'd send it to peer review; the only required revision is the constant in (4.8) and a comment on why the stochastic domination assumption forces the initial density bound.","headline":"A genuine advance on hydrodynamics for unbounded spin models, but the absolute-continuity proof has a wrong LLN constant that must be fixed before this is cited.","tokens_in":21128,"tokens_out":4697,"would_cite":true,"duration_ms":41646,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60K35","82C22"],"pacs":[],"model":"deepseek-v4-flash","headline":"Under diffusive scaling, three unbounded spin models converge to the heat equation, with model-dependent diffusion coefficients.","keywords":["hydrodynamic limit","KMP model","harmonic model","attractiveness","entropy method","gradient condition","unbounded occupation variables","heat equation"],"falsifier":"A direct numerical check would evolve each model on a large torus from an initial profile satisfying the domination assumption and compare $N^{-1}\\sum_x \\eta^\\sigma_x(N^2t)\\delta_{x/N}$ with the heat-kernel solution with diffusivity $D_\\sigma$; a mismatch that persists as $N$ grows would contradict Theorem 2.8. For the harmonic spin restriction, an algebraic search for ordered configurations with $s<1/2$ violating inequalities (3.1)-(3.2) would either confirm that the restriction is necessary or suggest it can be removed.","tokens_in":20074,"feed_emoji":"🎲","tokens_out":8570,"duration_ms":80754,"temperature":0.7,"pith_summary":"The paper proves that three stochastic lattice models with unbounded occupation variables—generalized Kipnis–Marchioro–Presutti energy model, its discrete particle counterpart, and the family of harmonic models—all have the same large-scale behavior: under diffusive time scaling, the empirical density converges to the solution of the plain heat equation, with diffusion coefficient $1/2$ for the two KMP-type models and $1/(2s)$ for the harmonic model with spin $s$. What makes the result worth attention is that previous hydrodynamic-limit technology did not readily apply, because the occupation variables are unbounded and the invariant measures have only exponential tails. The paper removes the obstacle by proving that all three models are attractive, a monotonicity property that lets the entropy method control unbounded configurations through comparison with an equilibrium measure. The price is an explicit restriction: initial states must be stochastically dominated by such an equilibrium measure, and the harmonic models require $s\\ge 1/2$. If the theorem stands, the microscopic details of these very different processes wash out in the continuum limit and leave pure linear diffusion.","feed_headline":"Three unbounded spin models all converge to heat diffusion","feed_subtitle":"Their empirical density obeys the heat equation, with diffusion constants 1/2 and 1/(2s).","key_machinery":"The machinery is the combination of the gradient identity and attractiveness inside the entropy method, a standard route that proves hydrodynamic limits by tightness plus characterization of limit points. The gradient identity $W^{\\sigma}_{x,x+1}=D_\\sigma(\\eta_{x+1}-\\eta_x)$ turns the generator's action on occupation variables into $N^2 L_N^\\sigma \\eta_x = D_\\sigma \\Delta_N \\eta_x$, which produces the diffusion term in the limiting equation. Attractiveness—preservation of the pointwise partial order between coupled configurations—is proved for generalized KMP by a basic Beta coupling and for discrete KMP and Harmonic models by checking the rate inequalities of Theorem 2.9 of [10]; it lets every expectation under the unknown initial measure be bounded by an expectation under the invariant measure $\\nu^{\\sigma}_{\\hat\\rho}$. A model-independent bound on the carré du champ, $D_\\sigma(\\eta_x-\\eta_{x+1})^2 - L^{\\sigma}_{x,x+1}(\\eta_x\\eta_{x+1}) \\le D_\\sigma(\\eta_x^2+\\eta_{x+1}^2)$, is the tightness input that controls the martingale oscillations uniformly in $N$.","core_discovery":"The central discovery is Theorem 2.8: for each of the three models, if the initial measures are associated to a bounded density profile $\\rho_0$ and are stochastically dominated by the invariant product measure $\\nu^{\\sigma}_{\\hat\\rho}$ for some $\\hat\\rho>0$ (and if the harmonic spin satisfies $s\\ge 1/2$), then the empirical measure $N^{-1}\\sum_x \\eta_x^\\sigma(N^2 t)\\delta_{x/N}$ converges in probability to the unique weak solution of $\\partial_t \\rho = D_\\sigma \\Delta \\rho$, with $D_{\\mathrm{gKMP}}=D_{\\mathrm{dKMP}}=1/2$ and $D_{\\mathrm{Harm}}=1/(2s)$. The proof's contribution is to verify the two structural properties that make the entropy method run: the gradient identity, which writes the instantaneous current as a discrete gradient with constant $D_\\sigma$, and attractiveness, which gives monotonicity under coupling. The harmonic models require $s\\ge 1/2$ because that is where the attractiveness proof works.","pith_inferences":["The restriction to dominated initial measures is probably not essential: approximating arbitrary bounded profiles by dominated ones and using the linearity of the heat equation should give the same limit; if true, the theorem would cover all physically natural initial states.","The three models share constant diffusivity and quadratic mobility, so the next-order fluctuations around the heat profile should be Gaussian with a covariance fixed by the mobility; this is a concrete prediction that could be checked by computing the associated non-equilibrium fluctuation field.","The attractiveness criterion used for the harmonic family suggests an easy numerical test for $s<1/2$: if ordered configurations can be found whose jump rates violate inequalities (3.1)-(3.2), that pinpoints why the spin restriction appears; if no violation exists, the restriction may be removable.","The continuous harmonic models mentioned in the paper have formally identical transport coefficients; once their well-posedness is settled, the same heat equation should emerge."],"forward_implications":["At macroscopic scale the evolution is deterministic diffusion: the random empirical measure converges in probability to the heat-kernel solution, so fluctuations vanish in the $N\\to\\infty$ limit.","The diffusion coefficient is density-independent, so the hydrodynamic equation is linear even though the microscopic jump rates are nonlinear functions of the occupation variables.","The result extends without conceptual change to any dimension and to the infinite lattice, where the same heat equation with the same $D_\\sigma$ is predicted.","Because the weak solution of the heat equation is unique, the whole sequence of empirical measures converges—subsequence extraction is only an intermediate step."],"supporting_citations":[{"why":"introduces the original KMP model and its discrete dual, which the paper generalizes to the gKMP and dKMP processes.","marker":"[13]"},{"why":"proposes the generalized KMP family and the spin-labelled discrete KMP models via duality, supplying the processes and their product invariant measures.","marker":"[3]"},{"why":"introduces the harmonic models as integrable stochastic particle processes, the third family whose hydrodynamic limit is proved.","marker":"[8]"},{"why":"provides the iff attractiveness criterion (Theorem 2.9) used to prove monotonicity for the discrete KMP and harmonic models from their jump rates.","marker":"[10]"},{"why":"supplies the entropy method that controls the empirical measures and characterizes their limit points.","marker":"[11]"},{"why":"gives the standard tightness criteria, the notion of association to a profile, and uniqueness of weak solutions of the heat equation used to close the argument.","marker":"[12]"},{"why":"supplies the Skorohod/uniform topology facts used to pass open-set estimates to limit points.","marker":"[1]"}],"fun_headline_variants":["Three unbounded spin models: heat equation limit proven","Attractiveness proves heat diffusion for gradient spins","Gradient spin models: universal heat limit under diffusive scaling","Unbounded spin models obey heat equation: new proof","Hydrodynamic limit for attractive gradient spin models"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the initial random configuration is statistically no larger, site by site, than an equilibrium configuration at some positive density; all moment estimates for the unbounded occupation variables use this comparison, so without it the proof has no control.","fun_headline_variants_meta":{"raw":{"variants":["Three unbounded spin models: heat equation limit proven","Attractiveness proves heat diffusion for gradient spins","Gradient spin models: universal heat limit under diffusive scaling","Unbounded spin models obey heat equation: new proof","Hydrodynamic limit for attractive gradient spin models"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000453,"raw_usage":{"total_tokens":2274,"prompt_tokens":937,"completion_tokens":1337,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":553,"completion_tokens_details":{"reasoning_tokens":1261}},"tokens_in":553,"tokens_out":1337,"duration_ms":10192,"temperature":1.0,"reasoning_tokens":1261,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:57:39.075762+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct numerical check would evolve each model on a large torus from an initial profile satisfying the domination assumption and compare $N^{-1}\\sum_x \\eta^\\sigma_x(N^2t)\\delta_{x/N}$ with the heat-kernel solution with diffusivity $D_\\sigma$; a mismatch that persists as $N$ grows would contradict Theorem 2.8. For the harmonic spin restriction, an algebraic search for ordered configurations with $s<1/2$ violating inequalities (3.1)-(3.2) would either confirm that the restriction is necessary or suggest it can be removed.","supporting_citations":[{"cited_title":"Kipnis, C","cited_arxiv_id":null,"evidence_quote":"introduces the original KMP model and its discrete dual, which the paper generalizes to the gKMP and dKMP processes."},{"cited_title":"Carinci, C","cited_arxiv_id":null,"evidence_quote":"proposes the generalized KMP family and the spin-labelled discrete KMP models via duality, supplying the processes and their product invariant measures."},{"cited_title":"Non-compactquantumspinchainsasintegrable stochastic particle processes.Journal of Statistical Physics, 180(1):135–171, 2020","cited_arxiv_id":null,"evidence_quote":"introduces the harmonic models as integrable stochastic particle processes, the third family whose hydrodynamic limit is proved."},{"cited_title":"Gobron and E","cited_arxiv_id":null,"evidence_quote":"provides the iff attractiveness criterion (Theorem 2.9) used to prove monotonicity for the discrete KMP and harmonic models from their jump rates."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the entropy method that controls the empirical measures and characterizes their limit points."},{"cited_title":"Kipnis and C","cited_arxiv_id":null,"evidence_quote":"gives the standard tightness criteria, the notion of association to a profile, and uniqueness of weak solutions of the heat equation used to close the argument."}],"review_version":1}