{"id":"e72ecdd7-d675-4c5c-878a-2b5bd71b0e05","arxiv_id":"2501.18566","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Large Boltzmann planar maps with a stable exponent alpha converge in the Gromov-Hausdorff-Prokhorov sense to an explicit random compact metric space S_alpha of Hausdorff dimension 2 alpha.","lead":"This paper proves that a family of random discrete surfaces called Boltzmann planar maps with large faces converge, after rescaling, to a newly constructed random continuous space S_alpha. It identifies a new universality class of random geometries beyond the well-known Brownian sphere, with topology ranging from a Sierpinski carpet in the dilute phase to a gasket with touching faces in the dense phase.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The full Theorem 1.1 relies on the Section 7.4.2 re-rooting reduction from a pointwise degree condition to the tail condition (1.2); that reduction is not visible in the reviewed excerpt and is the least secure link in the claim's full generality.","rationale":"The reader identified the non-genericity condition (1.2) as the weakest assumption, and I agree that it is the most delicate hypothesis in the statement. However, the theorem is explicitly conditional on (1.2), so the condition itself is not a flaw in the central claim. The load-bearing issue is whether the proof actually covers every q satisfying (1.2), given that the paper says the detailed proof is first done under a stronger pointwise condition and then extended by a re-rooting trick. Since Section 7.4.2 is not available in the reviewed excerpt, this reduction is an unverified link between the theorem's full statement and the detailed arguments. I therefore recommend conditional acceptance: the paper should be accepted if Section 7.4.2 indeed contains a correct reduction, and otherwise the theorem as stated would be overclaimed. This is a genuine missing-proof check rather than a manufactured objection, and it aligns with the reader's moderate confidence rather than exposing a known error.","tokens_in":70818,"tokens_out":16201,"duration_ms":170119,"concrete_test":"Read Section 7.4.2 and independently re-derive the re-rooting argument for a weight sequence satisfying (1.2) but not the pointwise condition w_q(deg(root face)=k) ∼ C k^{-α-1}. Verify that the construction yields the same space S_alpha with normalization (s_q n)^{-1/(2α)} for every such q; if the argument needs any condition beyond (1.2), or if s_q is not recovered exactly, then Theorem 1.1 is not proven for all non-generic q.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Theorem 1.1 asserts convergence for every admissible, critical, non-generic weight sequence q of exponent α. The introduction states that the proof is first carried out under the more stringent condition w_q(deg(root face)=k) ∼ C k^{-α-1}, and that Le Gall's re-rooting trick then yields the full statement from (1.2). This is a reduction from a pointwise asymptotic to a tail asymptotic, and it must preserve the universal limit S_alpha and the normalization (s_q n)^{-1/(2α)}. The excerpt does not contain the details of Section 7.4.2, so this step is an omitted proof in the material under review. The paper itself emphasizes, immediately after (1.5), that the non-genericity condition (1.2) is fine-tuned and depends on all values of q, and it explicitly leaves extensions to merely regularly varying tails as a conjecture. Thus the boundary of the proven universality class is exactly this Section 7.4.2 reduction: if the re-rooting argument requires any regularity beyond (1.2), or if it produces a q-dependent limit or a different scaling constant, then Theorem 1.1 as stated is not established for all non-generic q. This is not an accusation of error, but it is the place where the central claim's full generality is least secure.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs, for each alpha in (1,2), a random compact metric measure space (S_alpha, D*_alpha, Vol_alpha) from a normalized alpha-stable Levy excursion decorated by independent Brownian bridges, and claims that large critical non-generic Boltzmann bipartite planar maps with weight sequence q of exponent alpha converge in distribution to this space after scaling distances by (s_q n)^{-1/(2 alpha)}. The claimed limit is universal, depending on q only through alpha, and has Hausdorff dimension 2 alpha. The introduction describes a proof strategy: tightness from [97]; identification of the point equivalence via faces; uniqueness of geodesics via a two-source construction; surgery along geodesics; and a priori ball-volume estimates. The paper also proves that in the dilute phase alpha in [3/2,2) the limit is homeomorphic to the Sierpinski carpet, while in the dense phase alpha in (1,3/2) faces may touch.","tokens_in":71066,"tokens_out":10941,"duration_ms":115897,"significance":"If Theorem 1.1 holds, this is a major advance: it establishes the first universal scaling limits for non-generic Boltzmann maps with large faces, constructs the stable carpets/gaskets, and develops a rich continuum theory (Gaussian free field on looptrees, exact two-point function, geodesic classification). The paper is notable for the exact computation N(sup Z > 1) = alpha(alpha-1)/2 via Bessel and hypergeometric functions, for the spinal decomposition of the label process, and for the topology theorems, including the Sierpinski carpet statement in the dilute phase. The introduction is exemplary in outlining a long and intricate proof. Credible strengths include the detailed treatment of the coding process and the robust a priori estimates leading to a dimension bound. The main caveat is that the proof of full Theorem 1.1 relies on deferred re-rooting and surgery arguments that are not verified in the material under review.","major_comments":[{"comment":"Theorem 1.1 is stated for every admissible, critical, non-generic weight sequence q satisfying the tail condition (1.2). The text after (1.5) says that the proof is first carried out under the stronger pointwise condition w_q(deg(root face)=k) ~ C k^{-alpha-1}, and that the full statement follows from Le Gall's re-rooting trick, with details deferred to Section 7.4.2. Section 7.4.2 is not included in the material under review. This reduction is load-bearing: it must preserve the universal limiting space S_alpha and the normalization (s_q n)^{-1/(2 alpha)} for every q satisfying the fine-tuned condition (1.2). If the re-rooting argument requires additional regularity or yields a q-dependent limit or a different scaling constant, then Theorem 1.1 as stated is not established. Please provide the full details of Section 7.4.2, or restrict the statement of Theorem 1.1 to the pointwise condition and present the extension to the tail condition (1.2) as a conditional result.","section":"Section 1 (after (1.5)); Section 7.4.2"},{"comment":"The introduction states that the equality D = D* for every subsequential limit is proven via the identification of equivalence classes (Theorem 8.1), the two-point construction with delays (Sections 10-11), the uniqueness of the typical geodesic (Theorem 9.1), and the surgery argument with bad-point estimates (Proposition 9.3 and Section 12). These sections are only summarized heuristically in the introduction and are not part of the reviewed text. Since these are the core steps that go beyond [97] and establish the uniqueness of the subsequential limit, the central claim of Theorem 1.1 cannot be fully verified from the supplied material. This is a structural verification gap rather than an identified mathematical error, and the authors should ensure that the complete proof is available in the version under review.","section":"Proof of Theorem 1.1; Sections 8-12"}],"minor_comments":[{"comment":"The entry for PMroot contains duplicated 'of' and a typo: 'isometry classes of of (rooted) iweighted geodesic compact metric spaces' should read 'isometry classes of (rooted) weighted geodesic compact metric spaces'.","section":"Index of notation"},{"comment":"In several displays, the symbols 'eδ' and 'ed' are used where '~δ' and '~d' are intended (e.g., in the computation of the Brownian bridge covariance and in the final display of the proof). Please correct these typographical errors.","section":"Section 3.2, proof of Proposition 3.3"},{"comment":"The vanishing of the constants A' and B' in the hypergeometric expansion is asserted with 'a (tedious) computation shows that A'=B'=0', but the computation is not provided. Since the resulting value alpha(alpha-1)/2 determines the phase transition at alpha=3/2 and feeds into Proposition 5.6, please include the calculation in an appendix or give a precise reference.","section":"Section 5.2, proof of Proposition 5.4"},{"comment":"The term 'admissible' in 'admissible, critical and non-generic weight sequence' is not defined in the introduction; please give a definition or a reference at the point of first use.","section":"Theorem 1.1 statement"}],"recommendation":"major_revision","confidential_remarks":"The version of the manuscript provided to me was truncated after Section 6.3, so my major comments reflect the impossibility of verifying the proof of Theorem 1.1 from the supplied text, in particular the Section 7.4.2 reduction. The paper appears to be a very significant contribution if the deferred arguments are correct. I also note that the construction of S_alpha is intimately tied to the coding convergence result [97] by two of the authors; this reliance is not circular, but it means the paper does not independently prove tightness from first principles. The recommendation of major revision is driven by the need to see the deferred re-rooting and surgery arguments; if those are present in the full submission, the remaining issues are local and a minor revision would suffice."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"I read the arXiv version, focusing on the introduction, the construction of the label process Z, the continuum part (Sections 2–6), and the proof outline for the map side. This is the real thing: it constructs a family of explicit limiting spaces S_alpha from a stable Lévy excursion plus Brownian bridges, and proves that every non-generic critical Boltzmann map with exponent alpha converges to S_alpha. The Brownian sphere uniqueness program is transferred to the stable setting, and the topology dichotomy at alpha = 3/2 (Sierpinski carpet versus dense phase with touching faces) is a genuine new phenomenon. No circularity in the use of [97]: tightness and coding-process convergence are external benchmarks, and the paper's job is uniqueness of the metric limit plus the geometry of S_alpha.\n\nThe strongest part I could verify is the continuum analysis. The reformulation of Z as a Gaussian free field on the stable looptree is natural, the Hölder continuity bounds are clean, and the exact computation N(sup Z > 1) = alpha(alpha-1)/2 via Bessel and hypergeometric identities is careful. The use of Shepp's covering theorem to get the record structure on loops is clever and consistent. If the later chapters deliver what the outline promises, the proof of D = D* is as substantial as the Brownian sphere analogue.\n\nThe soft spot is exactly where you flagged: the reduction from the pointwise degree condition w_q(deg = k) ~ C k^{-alpha-1} to the tail condition (1.2) is relegated to Section 7.4.2, which is not in the material I saw. The theorem is stated for all non-generic q satisfying (1.2), so the re-rooting trick must preserve the limiting space and the normalization (s_q n)^{-1/(2 alpha)}. That is load-bearing. I cannot check it from what I read; the paper says it is 'minimal effort,' and Le Gall's earlier use of the same trick makes me suspect it works, but it remains unverified here. That is a reason for caution, not for rejection.\n\nI would send this to a serious referee. The result is a major advance in random planar maps, and the paper is written with care. The referee will need time on Section 7.4.2 and on the geodesic surgery sections, but that is what referees are for. I will cite this in my own work on scaling limits.","headline":"A very strong paper: explicit stable limits for non-generic Boltzmann maps, with the re-rooting reduction in Section 7.4.2 as the main unverified link.","tokens_in":71615,"tokens_out":2648,"would_cite":true,"duration_ms":30768,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60D05","60F17","60G52","05C80","60G15"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper constructs, for each α∈(1,2), a universal random compact metric space S_α — the α-stable carpet in the dilute phase and α-stable gasket in the dense phase — and proves that all non-generic critical Boltzmann planar maps of…","keywords":["Boltzmann planar maps","stable Lévy processes","looptrees","scaling limits","universality","Sierpinski carpet","Gaussian free field","Hausdorff dimension"],"falsifier":"Take two non-generic weight sequences with the same α and simulate large conditioned Boltzmann maps; compute the rescaled Gromov–Hausdorff–Prokhorov limits (for example via distance matrices between a growing cloud of sampled vertices) and the Hausdorff dimension of the limiting space. If the two limits differ or the dimension is not 2α, Theorem 1.1 is false. A smaller-scale check: evaluate N(sup Z>1) by Monte Carlo on a discretized stable looptree with independent Brownian bridges on each loop; the paper's two-point function identities imply the exact value α(α−1)/2, so any discrepancy would trace the failure to the continuum construction.","tokens_in":70591,"feed_emoji":"🗺️","tokens_out":8706,"duration_ms":91677,"temperature":0.7,"pith_summary":"This paper proves a universality theorem for random planar maps with very large faces. Fix α∈(1,2); any non-generic critical Boltzmann map whose root-face degree distribution has tail ~ c $k^{{-α}}$ converges, after graph distances are multiplied by (s_q n)^{-1/(2α)}, to a single random compact metric space S_α, the α-stable carpet for α≥3/2 and the α-stable gasket for α<3/2. The limit does not depend on the weight sequence q except through α, and has Hausdorff dimension 2α. This identifies a new one-parameter family of universal random geometries outside the Brownian sphere, and gives their topology: the Sierpinski carpet in the dilute phase, and a space whose faces may touch in the dense phase.","feed_headline":"Large random maps with big faces share one new fractal limit","feed_subtitle":"Every critical Boltzmann map of exponent α rescales to the same 2α-dimensional space, carpet-like or dense.","key_machinery":"The load-bearing objects are the α-stable looptree L coded by a spectrally positive α-stable Lévy excursion (loops glued along jumps), and the label process Z, defined as the Gaussian process on L with covariance given by the resistance metric — equivalently, Brownian motion indexed by the looptree. The proof works by (1) encoding discrete maps by labeled trees via a classical labeled-tree bijection, (2) passing to the coding-process limit (X,Z), (3) using fine properties of Z — absence of one-sided records on the skeleton, density of records on loops, and the exact two-point function — to prove that every subsequential limit identifies exactly the same points as the explicit pseudo-distance D*, and (4) a two-source labeled-tree construction to control geodesics between typical points and a surgery argument showing D and D* agree.","core_discovery":"The central discovery is that the scaling limit of non-generic critical Boltzmann maps with exponent α exists and is described explicitly by a stable Lévy excursion X decorated by a Gaussian label process Z, the Brownian motion indexed by the stable looptree. The limiting space S_α is [0,1]/∼_{D*} with D* built from Z; the argument shows every subsequential limit D equals D* by identifying which points are glued — only trivial identifications in the looptree or zero-distance — and by showing geodesics between typical points are unique and can be compared via a two-source construction. En route the paper establishes fine quantitative facts: local minima of Z avoid the skeleton of the looptree; the two-point function satisfies N(sup Z>1)=α(α−1)/2; the volume of root-centered balls has stretched-exponential tails; geodesics to the root are simple, the cut locus is totally disconnected, and the maximal number of geodesics from a point is 2.","pith_inferences":["Not proved in the paper: the same convergence should hold if the non-genericity hypothesis is weakened to regular variation or if bipartiteness is dropped; the paper only sketches a re-rooting route toward these extensions.","Not proved in the paper: if the conjectural link to γ-LQG is correct, S_α should coincide with the chemical metric inside a conformal loop ensemble, giving these spaces a conformal-geometric characterization.","Not proved in the paper: as α→2, the construction should degenerate to the Brownian-sphere regime, interpolating between the stable and Brownian universality classes.","Not proved in the paper: the exact value N(sup Z>1)=α(α−1)/2 yields an explicit candidate for the two-point distance distribution, testable by simulation of the continuum process or of large maps."],"forward_implications":["Every non-generic critical Boltzmann map with exponent α belongs to a single universality class: its scaling limit is S_α, independent of all other details of the face-weight sequence q.","The Hausdorff dimension 2α of S_α gives a new exact exponent for distances in these maps, refining the earlier tightness exponent.","The dichotomy at α=3/2 is sharp: for α∈[3/2,2) the limit is almost surely homeomorphic to the Sierpinski carpet, while for α∈(1,3/2) its faces touch, so the topology changes at the dilute/dense transition.","In any such limit, the geodesics toward a typical root point are simple, the cut locus is totally disconnected, and at most two distinct geodesics can start from the same point.","The volume of balls in S_α has stretched-exponential tail bounds, providing a robust local estimate usable in further metric-geometric analysis."],"supporting_citations":[{"why":"Establishes tightness of rescaled non-generic Boltzmann maps and the coding-process convergence that this paper builds on.","marker":"[97]"},{"why":"Introduces the α-stable looptree coded by a stable Lévy excursion, the continuum skeleton underlying the limiting faces.","marker":"[47]"},{"why":"Supplies the labeled-tree bijection encoding distances in Boltzmann maps, the discrete input to the whole proof.","marker":"[32]"},{"why":"Constructs the resistance metric and Brownian motion on stable looptrees, used to realize the label process Z as a Gaussian process.","marker":"[9]"},{"why":"Provides the two-source bi-pointed construction with delays, adapted here to build and control typical geodesics in the limit.","marker":"[111]"},{"why":"Gives the analogue of simple geodesics and their uniqueness on the Brownian sphere, the pattern followed for the root-geodesic classification.","marker":"[90]"},{"why":"Introduces the strategy of identifying exact equivalence classes to prove uniqueness of the Brownian map limit, adapted here to the stable setting.","marker":"[89]"}],"fun_headline_variants":["Carpet-like or dense: one fractal limit per alpha","Large-face maps rescale to a new fractal per alpha","Fractal limit for large-face maps: explicit and two-phase","Scaled maps with big faces have explicit fractal limits"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof depends on the non-genericity condition (1.2), which requires the root-face degree tail to be asymptotic to a specific constant times $k^{{-α}}$ and to involve the same s_q appearing in the rescaling; if that fine-tuned condition fails, the limit is not claimed to be S_α.","fun_headline_variants_meta":{"raw":{"variants":["Carpet-like or dense: one fractal limit per alpha","Large-face maps rescale to a new fractal per alpha","Fractal limit for large-face maps: explicit and two-phase","Scaled maps with big faces have explicit fractal limits"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001776,"raw_usage":{"total_tokens":6965,"prompt_tokens":870,"completion_tokens":6095,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":486,"completion_tokens_details":{"reasoning_tokens":6026}},"tokens_in":486,"tokens_out":6095,"duration_ms":46087,"temperature":1.0,"reasoning_tokens":6026,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T22:57:23.277705+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take two non-generic weight sequences with the same α and simulate large conditioned Boltzmann maps; compute the rescaled Gromov–Hausdorff–Prokhorov limits (for example via distance matrices between a growing cloud of sampled vertices) and the Hausdorff dimension of the limiting space. If the two limits differ or the dimension is not 2α, Theorem 1.1 is false. A smaller-scale check: evaluate N(sup Z>1) by Monte Carlo on a discretized stable looptree with independent Brownian bridges on each loop; the paper's two-point function identities imply the exact value α(α−1)/2, so any discrepancy would trace the failure to the continuum construction.","supporting_citations":[{"cited_title":"Le Gall and G","cited_arxiv_id":null,"evidence_quote":"Establishes tightness of rescaled non-generic Boltzmann maps and the coding-process convergence that this paper builds on."},{"cited_title":"Miermont","cited_arxiv_id":null,"evidence_quote":"Provides the two-source bi-pointed construction with delays, adapted here to build and control typical geodesics in the limit."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the analogue of simple geodesics and their uniqueness on the Brownian sphere, the pattern followed for the root-geodesic classification."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the strategy of identifying exact equivalence classes to prove uniqueness of the Brownian map limit, adapted here to the stable setting."}],"review_version":1}