{"id":"c6664dde-412d-45c7-b9bc-2c31303a3463","arxiv_id":"2607.10725","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"After rescaling, the centered Airy-line-ensemble height function converges in joint moments of compactly supported continuous linear statistics to a z-pullback of the Dirichlet GFF on the lower half-plane.","lead":"The paper proves that the rescaled, centered height function of the Airy line ensemble converges in joint moments to an explicit pullback of the Gaussian free field. This gives a field-level description of edge fluctuations in the KPZ class, matching the GFF picture already known for bulk height functions in random tilings.","discovery_kind":"extension","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified","rationale":"The paper supplies a complete, self-contained analytic proof of a natural field-level CLT for the Airy line ensemble. Every step—determinantal moment formula (Lemma 5.3), contour deformation to steep-descent paths (Section 5.2), tail estimates (Propositions 3.1–3.4), and identification with the GFF pullback (Section 6.2)—is written with explicit constants and uniform control. The only place where the argument could in principle fail is the discard of the regions x≥-T^{-ε}, but the moment bounds used there are classical and correctly applied. Consequently the reader’s ACCEPT verdict with high confidence stands; no adjustment is warranted.","tokens_in":23776,"tokens_out":450,"duration_ms":5691,"concrete_test":"Recompute the leading constant in the variance asymptotic (3.2) from Soshnikov’s original paper and verify that it remains 11/(12π^{2}); if the constant were larger by a factor that grows with T, the log-T moment bound of Proposition 3.1 would fail and the reduction step (6.3) would no longer vanish.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The reader's weakest assumption (Propositions 3.1–3.4) is correctly identified as the step that discards the near-zero and upper-tail regions in the decomposition (6.1)–(6.2). Those bounds rest on Soshnikov’s cumulant estimates (3.2)–(3.3) for the lower tail and on the classical Airy-function decay (3.10) for the upper tail; both are standard and are applied with explicit constants. The subsequent steepest-descent analysis (Sections 4–5) is fully explicit, the contour deformations are justified by residue calculus with a uniform separation η>T^{-δ}/8, and the resulting Gaussian moments match the Green function of the diffeomorphism z(t,x)=t+i√(-x). No hidden assumption or gap appears that would invalidate the joint-moment convergence of Theorem 1.1.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper proves that the centered and suitably rescaled height function of the Airy line ensemble converges, in the sense of joint moments of pairings against compactly supported continuous test functions, to an explicit pullback of the Gaussian free field with Dirichlet boundary conditions on the lower half-plane. The height function is defined by counting the number of Airy lines above a level x at time t. After the parabolic scaling (t,x)mapsto(T^{1/2}t,Tx), the centered field H_T is shown to satisfy that the random vectors (sqrt(pi)langle H_{T_n},phi_1rangle,...,sqrt(pi)langle H_{T_n},phi_mrangle) converge in all joint moments to a centered Gaussian vector whose covariance is the double integral of the Green function of the diffeomorphism z(t,x)=t+i sqrt(-x). The argument proceeds from the determinantal structure of the extended Airy kernel, moment bounds that discard the near-zero and upper-tail regions, and a fully explicit steepest-descent analysis of the phase function S on carefully chosen contours.","tokens_in":23951,"tokens_out":867,"duration_ms":9038,"significance":"The result supplies a field-level extension of Soshnikov’s CLT for the Airy point process and identifies the macroscopic fluctuation field of the Airy line ensemble with a concrete GFF pullback. Because the Airy line ensemble is the universal edge scaling limit for a wide class of KPZ models (Wigner matrices, lozenge tilings, non-intersecting paths, last-passage percolation), the theorem furnishes a canonical description of edge fluctuations that complements the bulk GFF results already known for random tilings. The proof is self-contained and analytic: it starts from the classical extended Airy kernel, uses only standard external inputs (Soshnikov’s cumulants and Airy-function asymptotics), and produces the Green-function covariance by direct contour analysis with no free parameters. The careful treatment of measurability and integrability of the continuous height pairings is a technical contribution that makes the continuous setting rigorous.","major_comments":[],"minor_comments":[{"comment":"In Definition 4.1 and the accompanying Figure 1 the contour gamma is described carefully, but a short sentence clarifying the orientation of the vertical segments when eta > sqrt(-x) (where those segments degenerate to points) would remove a possible ambiguity for the reader.","section":null},{"comment":"The constant C_1(m,A,epsilon) in Proposition 3.1 is asserted to exist for all T>=e; a parenthetical remark that the bound for odd m follows from Cauchy–Schwarz applied to the even case would make the argument fully self-contained without forcing the reader to reconstruct it.","section":null},{"comment":"In the display after (6.5) the factor 1/(2 pi^2) appears for the two-point integral I, while the final covariance (1.4) carries a factor -1/(2 pi). The conversion via Wick’s formula is correct, but an intermediate sentence equating the two normalizations would help the reader track the constants.","section":null},{"comment":"Lemma 2.9 establishes that the pairings are a.s. finite and have finite moments; a brief forward reference to this lemma in the statement of Theorem 1.1 (already present as Remark 1.2) could be strengthened by citing the precise moment bound used later in (6.8)–(6.9).","section":null},{"comment":"Typographical consistency: the manuscript alternates between “Airy line ensemble” and “Airy Line Ensemble” in section headings; a uniform choice would improve polish.","section":null}],"recommendation":"accept","confidential_remarks":"The manuscript is a clean, technically solid contribution that fits well in a probability journal of this calibre. The analytic work is thorough and the result is of genuine interest to the KPZ and random-matrix communities. I see no reason to delay acceptance."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"This paper does exactly what the title says: it upgrades Soshnikov’s one-dimensional CLT for the Airy point process to a full space-time field limit for the Airy line ensemble height function, and identifies the limit as the pullback of the Dirichlet GFF under the explicit map z(t,x)=t+i√(-x). That is new. Discrete models (Kenyon, Petrov, Borodin–Ferrari) already had bulk GFF fluctuations, but the continuous Airy ensemble itself only had the fixed-time counting statistics. Theorem 1.1 fills the gap with joint-moment convergence of the rescaled centered height against compactly supported continuous test functions.\n\nThe argument is standard and carefully executed. They start from the extended Airy kernel, write the joint moments via the usual determinantal expansion (Lemma 5.3), discard the near-zero and upper-tail regions with moment bounds that rest on Soshnikov’s cumulants plus classical Airy decay (Props 3.1–3.4), then run a fully explicit steepest-descent analysis on the phase function S (Sections 4–5). Contour deformations are justified by residue calculus with uniform separation, the estimates are written out with constants, and the resulting pairings match the Green function of the diffeomorphism. No free parameters, no circular normalizations. The technical overhead of working with a continuous height function (measurability, integrability of the pairings) is handled cleanly in Lemma 2.9.\n\nThe softest step is the reduction that throws away everything above x=-T^{-ε}. If those moments grew faster than any power of log T the argument would fail, but the bounds used are classical and applied with explicit constants; the stress-test note is right that nothing hidden appears. Convergence is only in joint moments, not in a stronger topology, which is the usual price for this style of analysis and is stated honestly.\n\nThis is for people who work on KPZ edge statistics, random matrices, or height-function fluctuations. It is self-contained enough that a serious referee can check every estimate. I would send it to peer review without hesitation and would cite the covariance formula when I need the Airy-to-GFF link.","headline":"Clean field-level CLT for the Airy height function; the first space-time GFF limit for the continuous edge object itself.","tokens_in":24584,"tokens_out":551,"would_cite":true,"duration_ms":6915,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60G15","60G55","60B20","82B31"],"pacs":[],"model":"grok-4.5","headline":"After scaling, the Airy line ensemble's height function converges in joint moments to an explicit pullback of the Gaussian free field.","keywords":["Airy line ensemble","Gaussian free field","height function","extended Airy kernel","determinantal point process","steepest descent","KPZ universality","joint moments"],"falsifier":"Compute the second moment of the linear statistic against a fixed positive continuous test function supported deep in the lower half-plane and check whether it converges to the explicit Green-function integral predicted by formula (1.4); any systematic discrepancy for large scaling parameter would falsify the claimed covariance.","tokens_in":24666,"feed_emoji":"∿️","tokens_out":1094,"duration_ms":10761,"temperature":0.7,"pith_summary":"The Airy line ensemble is the universal edge scaling limit for many models in the KPZ class, including random matrices, lozenge tilings, and last-passage percolation. This paper studies the global space-time fluctuations of its height function, which simply counts how many lines sit above a given level at a given time. After a natural parabolic scaling that zooms out, the centered height function is shown to converge, in the sense of joint moments against continuous compactly supported test functions, to a Gaussian free field pulled back by the map that sends the lower half-plane to the upper half-plane via square-root coordinates. The result upgrades Soshnikov's classical central-limit theorem for single-time counting statistics to a full two-dimensional random field, and it matches the bulk GFF fluctuations already known for the discrete models whose edges produce the Airy ensemble. The argument relies on the ensemble's determinantal structure, moment bounds that discard the upper and near-zero tails, and a steepest-descent analysis of the extended Airy kernel that extracts the Green function of the limiting field.","feed_headline":"Airy height fluctuations converge to a GFF pullback","feed_subtitle":"Scaled joint moments of the Airy line ensemble match the Green function of square-root coordinates","key_machinery":"The extended Airy kernel written in the form (2.4) whose phase function S admits explicit critical points and steep-descent contours; after discarding tails by moment bounds, the joint moments of H_T reduce to contour integrals that converge, by steepest descent, to the Green-function pairings of the z-pullback of the GFF.","core_discovery":"Theorem 1.1 asserts that the rescaled and centered height function H_T of the Airy line ensemble satisfies: for every m and every continuous compactly supported test functions φ_1,…,φ_m, the random vector of linear statistics (√π ⟨H_{T_n},φ_1⟩,…,√π ⟨H_{T_n},φ_m⟩) converges in joint moments, as T_n\to∞, to a centered Gaussian vector whose covariance is exactly the double integral over the lower half-plane of the Green function of the diffeomorphism z(t,x)=t+i√(-x).","pith_inferences":["The same steepest-descent contours and phase-function analysis should apply, with only notational changes, to the Airy_2 and Pearcey kernels that appear at soft edges of other matrix ensembles.","Because the limit is identified only through joint moments, tightness in a suitable Sobolev space of distributions would upgrade the result to convergence in law of the random fields themselves.","The appearance of the identical GFF pullback both in the bulk of discrete tilings and at the continuous Airy edge suggests a single global Gaussian structure that interpolates between bulk and edge."],"forward_implications":["Macroscopic height fluctuations of the Airy line ensemble are completely described by a single explicit Gaussian free field.","Soshnikov's one-dimensional CLT for Airy counting statistics is recovered as the equal-time special case of the field limit.","Any discrete model whose edge converges to the Airy line ensemble is expected to have the same GFF pullback as its macroscopic height fluctuation field.","The limiting covariance is given by an elementary closed-form integral involving only square roots and logarithms, making quantitative predictions immediate."],"fun_headline_variants":["Airy line heights converge to GFF pullback","Rescaled Airy height function yields GFF pullback","Airy ensemble heights match GFF via joint moments","Centered Airy fluctuations converge to free-field pullback","Airy height stats converge to square-root GFF Green"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"The moment bounds that force every contribution from heights near zero or above zero to vanish as the scaling parameter tends to infinity; if those tails were heavier, the reduction to the deep lower half-plane where the steepest-descent analysis works would fail.","fun_headline_variants_meta":{"raw":{"variants":["Airy line heights converge to GFF pullback","Rescaled Airy height function yields GFF pullback","Airy ensemble heights match GFF via joint moments","Centered Airy fluctuations converge to free-field pullback","Airy height stats converge to square-root GFF Green"]},"model":"grok-4.5","effort":"low","cost_usd":0.00703,"raw_usage":{"total_tokens":1675,"prompt_tokens":659,"num_sources_used":0,"completion_tokens":63,"cost_in_usd_ticks":70300000,"prompt_tokens_details":{"text_tokens":659,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":953,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":659,"tokens_out":63,"duration_ms":11672,"temperature":1.0,"reasoning_tokens":953,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-14T09:42:07.320047+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Compute the second moment of the linear statistic against a fixed positive continuous test function supported deep in the lower half-plane and check whether it converges to the explicit Green-function integral predicted by formula (1.4); any systematic discrepancy for large scaling parameter would falsify the claimed covariance.","supporting_citations":[],"review_version":1}