{"id":"d0822898-9bce-470c-b075-8e246a5637ab","arxiv_id":"1908.03746","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For self-similar growth-fragmentations, the intrinsic area of small root-centered balls has a deterministic almost-sure rate with positive initial size and logarithmic fluctuations when the initial size is zero.","lead":"This paper derives sharp rates at which the intrinsic area of a tiny ball around the root vanishes in self-similar growth-fragmentation processes, a class of branching particle systems. It uses those rates to improve known bounds for the Brownian map and to extend a boundary-area result from Brownian disks to stable disks.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The main result is only as secure as Theorem 7 of the cited Arista–Rivero preprint [1], which supplies the small-time tail of the exponential functional I via Lemma 3; the paper does not reproduce or independently verify that theorem.","rationale":"I read the paper as a rigorous proof that the intrinsic area of small balls around the root has deterministic regularly varying asymptotics under assumptions (3) and (9), with applications to stable disks and maps. The proof structure is sound: Lemma 1 gives the Markov-branching decomposition of A; Lemma 2 connects E_1(A(t)) to the spinal lifetime I; Lemma 3 supplies the small-time tail of I from [1]; Lemma 6 identifies a martingale whose compensator S gives the dominant asymptotic; Lemmas 4, 5, and 7 control error terms; and the BDG argument in the proof of Theorem 1 handles fluctuations. I found no internal circularity or data-fitting: the constants are derived from model parameters, and the Brownian disk application explicitly accounts for the sqrt(8/3) change of cumulant. The weakest point is exactly the one identified by the reader: the small-time tail of the exponential functional I, and hence the leading constant in Theorem 1, is imported from Theorem 7 of the unpublished preprint [1] without reproducing its proof. This is a real verification gap, but it is a standard citation practice and does not, by itself, invalidate the paper. The proposed concrete check would settle the residual doubt; unless it fails, I do not see grounds to change the reader's ACCEPT verdict.","tokens_in":23196,"tokens_out":10493,"duration_ms":124294,"concrete_test":"Obtain the latest published or final arXiv version of Arista–Rivero [1] and mechanically re-derive Lemma 3: set γ = (ω− − ρ)/|α|, verify every hypothesis of Theorem 7 for the process |α|η^-, including the required tail-ratio condition, any non-lattice/spread-out condition, and finiteness of E_1^-(I^{-γ}), then recompute the leading constant (1 + γ)^{-1} E_1^-(I^{-γ}) and compare it with the constant in Lemma 3 and Theorem 1. If Theorem 7 is unavailable or has an unverified hypothesis that fails for this growth-fragmentation class, Lemma 3 and the central result lose their proof; if the constant differs, the stated normalization of Theorem 1 is incorrect.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 1 is proved by reducing the first moment of A(ǫ) to P_1^-(I ≤ ǫ) through Lemma 2, and then invoking Lemma 3. The proof of Lemma 3, in Section 4.1, consists of applying Theorem 7 of the unpublished preprint [1] to the Lévy process |α|η^-, checking the tail ratio Π_α^-((−∞,−x−y))/Π_α^-((−∞,−x)) → exp(−((ω− − ρ)/|α|) y), and then identifying the constant. This imported theorem supplies both the small-time rate and the explicit multiplicative constant that Theorem 1 inherits; Lemma 7 uses the same input for the a.s. asymptotics. The paper does verify that the Cramér-type condition |α|γ = ω− − ρ lies in (0, ω+ − ω−), which follows from (9), and the regular-variation ratio is consistent with (9). What is not verified is whether the full set of hypotheses of Theorem 7 in [1] holds for this class of processes, e.g. any non-lattice, spread-out, or additional moment condition on |α|η^-, and whether the constant stated in [1] is exactly as used. This is a genuine external dependency rather than an internal inconsistency: conditional on Lemma 3, the martingale argument in Section 4.2 and the envelope estimates in Section 5 are coherent. But if Theorem 7 of [1] is misstated, inapplicable, or if its constant differs, then Lemma 3, Lemma 7, and the normalization of Theorem 1 are not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the intrinsic area measure A on the leaf space of a self-similar growth-fragmentation with negative self-similarity index α and two positive Laplace/cumulant roots ω−<ω+. The main result, Theorem 1, claims an almost sure and L1 asymptotic for A(ε) as ε→0+ when the initial cell has positive size, with an explicit regular variation rate and an explicit constant expressed as an exponential-functional expectation under the tilted measure P1−. Propositions 2 and 3 give logarithmic upper and lower bounds for the rescaled area under the P0+ measure, and Section 6 applies these results to stable disks and stable/Brownian maps, retrieving Le Gall's boundary annulus result and improving known log-type bounds for balls in the Brownian map.","tokens_in":23455,"tokens_out":35167,"duration_ms":342875,"significance":"If the statement and the applications are corrected as discussed below, the paper gives a sharp first-order asymptotic with an explicit multiplicative constant for the intrinsic area near the origin, obtained by a martingale and regular-variation method rather than by fitting. The proof is detailed, and the connection to random planar maps makes the result valuable beyond the growth-fragmentation community. The use of a substantial external theorem from an unpublished preprint and several normalization inconsistencies currently prevent the result from being accepted as written.","major_comments":[{"comment":"The normalization in the displayed Theorem 1 is incompatible with the proof and with Lemma 3. Lemma 3 gives E1(A(ε)) ∼ const · ε^{1+ω−/|α|} Λ(ε^{1/|α|}), while Eq. (12) proves that (ε^{1+ω−/|α|} Λ(ε^{1/|α|}))^{-1} Mε → 0. Therefore the theorem should state ε^{-(1+ω−/|α|)} Λ(ε^{1/|α|})^{-1} A(ε) → ... , with Λ in the denominator, not in the numerator. As printed, the theorem is the reciprocal of the quantity proved in the paper, and it is also inconsistent with the application in Section 6, where the final ε^{-2}A(ε)→x for θ=3/2 requires the denominator form. This is a load-bearing error, though it appears to be a repairable typo.","section":"Section 3, Theorem 1; Section 4.2, Eq. (12)"},{"comment":"The notation Λ is used inconsistently. Assumption (9) defines Λ(x)=Λ((−∞,−x)), but Eq. (10) and the subsequent estimates, including Lemma 5, require Λ(x) to denote the small-jump left tail Λ((−x,0)); the tail at −∞ cannot be regularly varying with index −ρ at 0 for a Lévy measure. In Section 6 the formula Λ(ε) ∼ c−/θ ε^θ also has a sign error: combining Lemma 13 with Eq. (10) and ρ=θ gives Λ(ε) ∼ c−/θ ε^{-θ}. With the corrected Theorem 1 normalization from the previous comment, this sign is needed to recover the exponent (θ−1/2)/(θ−1) in the stable-disk application; with the sign as printed the application does not have the stated exponent.","section":"Section 2, Eq. (9); Section 6, after Eq. (24)"},{"comment":"The proof of Lemma 3 relies entirely on Theorem 7 of the unpublished arXiv preprint [1] for both the rate and the multiplicative constant in the small-time tail of I. The paper verifies the tail-ratio condition and the interval condition |α|γ∈(0,ω+−ω−), but it does not verify all hypotheses of Theorem 7 for |α|η−, nor does it reproduce or independently derive the needed statement. Since Lemma 3 feeds directly into Lemma 7 and hence into the normalization of Theorem 1, the central result is only as secure as this external theorem. The authors should either include a self-contained proof of the needed tail asymptotic or state the exact hypotheses of Theorem 7 and confirm them in detail for the processes considered.","section":"Section 4.1, Lemma 3"}],"minor_comments":[{"comment":"The sentence 'Provided that ε ↦→ Mε has regular variation at 0' is unclear, since M is a stochastic process and not a deterministic function; the argument only needs the a.s. comparison τ_ε ∼ ε and the supremum bound over [0,ε∧T] established later in the same proof.","section":"Section 4.2, paragraph before Eq. (12)"},{"comment":"The reconciliation with Le Gall's Theorem 3 is conceptually correct, but the reader must supply the calculation that replacing κ3/2 by √(8/3)·κ3/2 multiplies all distances by the stated constant and scales η− by the same factor; a few more lines here would improve clarity.","section":"Section 6, factor 3/8 discussion"},{"comment":"There are several typographical errors, e.g., 'Insitut' in the affiliation footnote, 'It holds that that' at the start of Lemma 3, and 'compare to' in Section 6; these should be corrected in a revision.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The most serious issue is the normalization error in Theorem 1: the displayed theorem multiplies by Λ while the proof and Eq. (12) divide by Λ. This appears fixable, but it is central rather than cosmetic and must be corrected before publication. I also want to flag to the editor that the dependence on Theorem 7 of [1] is a genuine external dependency; if that theorem is not yet published, the authors should be asked to add verification or a proof. With those changes, the paper would be a solid contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a serious, carefully written paper that delivers new exact asymptotics for intrinsic area in self-similar growth-fragmentations, and the random-map applications are real. The main risk is not internal—the martingale and envelope arguments check out—but external: Lemma 3 leans on Theorem 7 of an unpublished Arista–Rivero preprint, and the paper does not reproduce that proof. I would still send it to a referee, but the referee should be asked to verify that theorem or have the author supply the missing hypotheses.\n\nWhat is new: Theorem 1 gives an almost sure and L1 limit for epsilon^{-(1+omega_-/|alpha|)} Lambda(epsilon^{1/|alpha|}) A(epsilon) with an explicit constant, under regular variation of the Levy measure tail. Propositions 2 and 3 give logarithmic upper and lower bounds for the size-0 case. These are not in earlier work; [19] had absolute continuity and singularity but not rates, and [24] only handled the Brownian disk. The stable-disk extension and the Brownian-map bounds are honest applications, and the paper explicitly reconciles the 3/8 factor with Le Gall via the sqrt(8/3) scaling—good practice.\n\nThe proofs are detailed and coherent. Lemma 1's Markov-branching decomposition, the compensator computation in Lemma 7, and the envelope estimates in Section 5 are all substantial and appear correct. The self-citation [19] is used only for the absolute-continuity and regularity input in Lemma 2, which is legitimate.\n\nThe soft spot is exactly what the stress test says. Lemma 3 imports Theorem 7 of [1], which supplies both the rate and the multiplicative constant for the small-time tail of the exponential functional I. The paper checks the ratio condition and the Cramer-type condition, but it does not verify the full hypothesis set of [1] and does not prove the constant's exact form. If that theorem is wrong or misapplied, the normalization of Theorem 1 fails. This is a standard citation practice in the field, and the dependency is flagged clearly, so it lowers confidence rather than invalidating the paper. A referee should independently check [1] or ask for a self-contained proof of the needed tail estimate.\n\nFor whom: people working on growth-fragmentations, exponential functionals of Levy processes, and random maps, especially stable disks and the Brownian map. It deserves a serious referee; I would expect acceptance after the external theorem is confirmed.","headline":"Genuinely new exact rates for intrinsic area in growth-fragmentations, with the key caveat that the main tail input is imported from an unpublished preprint rather than proven here.","tokens_in":24078,"tokens_out":2124,"would_cite":true,"duration_ms":22980,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60J25","60G18","60G57"],"pacs":[],"model":"deepseek-v4-flash","headline":"Intrinsic area near the root has an explicit almost-sure law","keywords":["self-similar growth-fragmentations","intrinsic area","rate of growth","exponential functionals of Lévy processes","spinal decomposition","random planar maps","stable disks","Brownian map"],"falsifier":"Compute the small-time tail of the exponential functional $I$ for the hypergeometric stable-disk family (e.g., $\\theta=3/2$) directly from the explicit laws of $\\eta^-$, and compare the leading multiplicative constant with the one used in Lemma 3; any discrepancy refutes the constant in Theorem 1.","tokens_in":22919,"feed_emoji":"📏","tokens_out":18359,"duration_ms":157148,"temperature":0.7,"pith_summary":"Self-similar growth-fragmentations describe particle systems that grow, split, and die, and they encode the slicing of random surfaces such as stable disks and the Brownian map. This paper asks how the intrinsic area of a small ball centered at the root behaves as its radius tends to zero. The main theorem gives an explicit almost-sure and $L^1$ limit for the rescaled area when the founding cell has positive size: the limit is a deterministic constant times a regularly varying factor built from the small-jump tail of the driving Lévy process. When the founding cell is instead conditioned to grow indefinitely from size zero, the rescaled area is stationary in logarithmic time and fluctuates without a limit, but is controlled by powers of the logarithm. These results recover the boundary asymptotics for the free Brownian disk and extend them to the whole family of stable disks, while sharpening the known bounds for the Brownian map.","feed_headline":"Intrinsic area near the root has an explicit almost-sure law","feed_subtitle":"The formula recovers the Brownian disk result and sharpens the Brownian map bounds.","key_machinery":"The proof runs on two linked mechanisms. First, the Markov-branching decomposition of the area: $A(t)=\\sum_{s\\le t}|\\Delta^-\\chi_\\emptyset(s)|^{\\omega_-}A_s((t-s)|\\Delta^-\\chi_\\emptyset(s)|^\\alpha)$, with independent rescaled copies of $A$ attached to each child of the Eve cell, reduces the area of a ball to a sum over birth events. Second, the spinal decomposition $P^-$ selects a leaf with probability proportional to the intrinsic area, giving the identity $E_1(A(t))=P^-_1(I\\le t)$, where $I=\\int_0^\\infty\\exp(-\\alpha\\eta^-(t))\\,dt$ is the exponential functional (absorption time) of the tilted positive self-similar Markov process. The first-moment asymptotics follow from an imported small-time tail estimate for $I$, and a compensated-jump martingale $M_t$ then upgrades the expectation to the almost-sure statement of Theorem 1.","core_discovery":"The central claim is that the intrinsic area measure of a self-similar growth-fragmentation has a deterministic small-scale profile. For every positive initial size $x$, $P_x$-almost surely and in $L^1$,\n$$\\$varepsilon^{{-(1+\\omega_-/|\\alpha|)}}$\\Lambda(\\$varepsilon^{{1/|\\alpha|}}$)A(\\varepsilon) \\to \\frac{|\\$\\alpha$|\\rho}{(\\omega_- - \\rho)(\\omega_- + |\\$\\alpha$| - \\rho)}\\,E^-_1\\!\\left($I^{{\\frac{\\omega_- - \\rho}}${\\$\\alpha$}}\\right)$x^{{\\alpha+\\rho}}$$$\nas $\\varepsilon\\to0^+$. Under the same hypotheses the convergence also holds under the spine-tilted measure $P^+_x$. For a founding cell that grows indefinitely from size $0$, the paper proves that $t^{\\omega_-/\\alpha}A(t)$ is a stationary process in logarithmic time, has no almost-sure limit, yet obeys almost-sure logarithmic bounds, with an upper bound of order $|\\log t|^{1+\\delta}$ and a lower bound of order $|\\log t|^{-q}$ for $q$ sufficiently large.","pith_inferences":["The explicit prefactor in Theorem 1 offers a calibration route: fitting the $\\varepsilon$-scaling of measured intrinsic areas in simulated stable maps could estimate the tail index $\\rho$ and the constant, providing an independent test of the growth-fragmentation representation.","The imported small-time tail estimate is the only non-elementary input; if a direct proof of a tail bound for $I$ under weaker conditions were available, assumption (9) could likely be relaxed to a less restrictive regular-variation hypothesis.","Because Proposition 1 makes $t^{\\omega_-/\\alpha}A(t)$ stationary in $u=\\log t$, the logarithmic bounds of Propositions 2-3 could plausibly be sharpened into a law of the iterated logarithm for the fluctuations around the stationary mean.","The normalization factor $3/8$ relative to the Brownian disk constant makes explicit that the growth-fragmentation dictionary fixes the intrinsic metric only up to a scaling of the cumulant; any transfer between growth-fragmentation constants and map constants must fix this gauge first."],"forward_implications":["For any initial size $x>0$, the intrinsic area of the $\\varepsilon$-ball around the root is almost surely asymptotic to the explicit deterministic multiple of $\\varepsilon^{1+\\omega_-/|\\alpha|}\\Lambda(\\varepsilon^{1/|\\alpha|})$ given by Theorem 1.","When the founding cell is conditioned to start from size $0$ and grow forever, $t^{\\omega_-/\\alpha}A(t)$ is stationary in $\\log t$; it has no almost-sure limit, but almost surely $A(t)$ stays between $t^{\\omega_-/|\\alpha|}|\\log t|^{-q}$ and $t^{\\omega_-/|\\alpha|}|\\log t|^{1+\\delta}$ for small $t$.","For the stable-disk family $X_\\theta$, $\\theta\\in(1,3/2]$, Theorem 1 gives $\\varepsilon^{-(\\theta-1/2)/(\\theta-1)}A(\\varepsilon)\\to c\\,x$ almost surely, generalizing the free Brownian disk result ($\\theta=3/2$, $\\varepsilon^{-2}A(\\varepsilon)\\to x$ up to normalization).","The upper bound for the area of a small ball in the Brownian map is improved from $\\varepsilon^{4-\\delta}$ to $\\varepsilon^4|\\log\\varepsilon|^{1+\\delta}$, and a lower bound $\\varepsilon^4|\\log\\varepsilon|^{-q}$ for $q>6$ is obtained.","The same growth-fragmentation connection transfers these statements to stable maps for all $\\theta\\in(1,3/2]$."],"supporting_citations":[{"why":"Supplies the small-time tail $P^-_1(I\\le t)$ up to a multiplicative constant, used in Lemma 3 to derive the first-moment asymptotics and the constant in Theorem 1.","marker":"[1]"},{"why":"Introduces the spinal decompositions $P^-$ and $P^+$, the intrinsic-area martingale, and the growth-fragmentation connection to random planar maps.","marker":"[6]"},{"why":"Provides the identity $E_1(A(t))=P^-_1(I\\le t)$ (Lemma 2) and the absolute-continuity/singularity framework for the area measure.","marker":"[19]"},{"why":"Gives the Karamata theorems used repeatedly to pass from regular variation of $\\Lambda$ to asymptotics for integrals and tails.","marker":"[12]"},{"why":"The Brownian disk boundary-volume theorem $\\varepsilon^{-2}A(\\varepsilon)\\to x$ that Section 6 recovers and extends to stable disks.","marker":"[24]"},{"why":"Describes the lower envelope of positive self-similar Markov processes, used in the proof of Proposition 3.","marker":"[16]"},{"why":"Describes the upper envelope of positive self-similar Markov processes, used in the proof of Proposition 3.","marker":"[33]"}],"fun_headline_variants":["Intrinsic area near root: explicit almost-sure law","Growth-fragmentation area: exact law near origin","Stable disks: refined area bounds near boundary","Intrinsic area at origin: deterministic profile","New law for intrinsic area in growth-fragmentations"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The main theorem inherits its exact rate and constant from an imported, unreproduced small-time tail estimate for the exponential functional of a tilted Lévy process; if that estimate is wrong or inapplicable, the central result collapses.","fun_headline_variants_meta":{"raw":{"variants":["Intrinsic area near root: explicit almost-sure law","Growth-fragmentation area: exact law near origin","Stable disks: refined area bounds near boundary","Intrinsic area at origin: deterministic profile","New law for intrinsic area in growth-fragmentations"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001507,"raw_usage":{"total_tokens":6087,"prompt_tokens":1035,"completion_tokens":5052,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":651,"completion_tokens_details":{"reasoning_tokens":4978}},"tokens_in":651,"tokens_out":5052,"duration_ms":34503,"temperature":1.0,"reasoning_tokens":4978,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:05:10.684341+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the small-time tail of the exponential functional $I$ for the hypergeometric stable-disk family (e.g., $\\theta=3/2$) directly from the explicit laws of $\\eta^-$, and compare the leading multiplicative constant with the one used in Lemma 3; any discrepancy refutes the constant in Theorem 1.","supporting_citations":[{"cited_title":"Implicit renewal theory for exponential functionals of L\\'evy processes","cited_arxiv_id":"1510.01809","evidence_quote":"Supplies the small-time tail $P^-_1(I\\le t)$ up to a multiplicative constant, used in Lemma 3 to derive the first-moment asymptotics and the constant in Theorem 1."},{"cited_title":"Martingales in self- similar growth-fragmentations and their connections with random p lanar maps","cited_arxiv_id":null,"evidence_quote":"Introduces the spinal decompositions $P^-$ and $P^+$, the intrinsic-area martingale, and the growth-fragmentation connection to random planar maps."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the identity $E_1(A(t))=P^-_1(I\\le t)$ (Lemma 2) and the absolute-continuity/singularity framework for the area measure."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the Karamata theorems used repeatedly to pass from regular variation of $\\Lambda$ to asymptotics for integrals and tails."},{"cited_title":"Brownian disks and the Brownian snake","cited_arxiv_id":null,"evidence_quote":"The Brownian disk boundary-volume theorem $\\varepsilon^{-2}A(\\varepsilon)\\to x$ that Section 6 recovers and extends to stable disks."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Describes the lower envelope of positive self-similar Markov processes, used in the proof of Proposition 3."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Describes the upper envelope of positive self-similar Markov processes, used in the proof of Proposition 3."}],"review_version":1}