{"id":"a64f5a41-5aed-4b4c-b230-a0d8d7673da0","arxiv_id":"2505.01248","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Almost all small solutions of the 1D Kirchhoff equation keep their Fourier actions almost frozen for ε^{-r} and sub-exponential times.","lead":"For the 1D Kirchhoff string equation, this paper proves that almost every small initial datum stays stable for time scales of order ε^{-r} for any r in Sobolev spaces and for super-polynomial times in Gevrey spaces. It introduces a rational normal form for reversible vector fields, bypassing the equation's lack of external parameters.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Norm mismatch in Lemma 6.1: U_N^γ ⊂ ℓ^2_s is µ-null under the Gaussian measure on ℓ^2_{s-1}, invalidating the measure estimate as written.","rationale":"The reader correctly identified the small-divisor measure estimate and its preservation under the normal-form process as the load-bearing step. My stress test agrees with that choice but locates a more specific internal inconsistency: the Gaussian measure used for the 'almost every initial datum' conclusion lives on a strictly larger space than the one in which U_N^γ is defined. Section 3 places U_N^γ inside ℓ^2_s with conditions involving ‖z‖_s^2, while the measure (6.1) has support in ℓ^2_{s−1}; with coordinate variances ~ m^{−2s}, the norm ‖z‖_s is infinite almost surely. Consequently, the set εz ∈ U_N^γ is µ-null, and Lemma 6.1 cannot be true as stated. This is not merely a missing estimate: it is an index mismatch that breaks the measure-theoretic conclusion of Theorem 1.1. The rest of the normal-form construction may be sound in the abstract ℓ^2_s setting, and the mismatch might be repairable by shifting all indices by one and adjusting Lemma 6.2–6.3 accordingly, but as written the central claim is not established. The concrete test of computing µ(ℓ^2_s) would settle whether the concern lands exactly as described.","tokens_in":83377,"tokens_out":25177,"duration_ms":218061,"concrete_test":"Compute µ(ℓ^2_s) under (6.1): for independent Gaussian z_m with E|z_m|^2 = (2m^{2s})^{-1}, E∑_{m=1}^M m^{2s}|z_m|^2 = (1/2)H_M → ∞, so by the three-series theorem ∑ m^{2s}|z_m|^2 = ∞ a.s. This makes any U_N^γ ⊂ ℓ^2_s µ-null, so Lemma 6.1 is false as stated. A decisive repair check: re-run Lemma 6.1 with U_N^γ redefined using ‖z‖_{s−1} and κ_j^{−2(s−1)}; if the probability bound (6.4) still holds with λ depending on r,s, then the central theorem can be recovered modulo this index shift.","verdict_should_be":"REJECT","load_bearing_attack":"Section 3 defines U_N^γ as a subset of ℓ^2_s, with conditions (3.10)–(3.11) involving ‖z‖_s^2 and κ_j^{-2s}. Lemma 6.1 then asserts µ(εz ∈ U_N^γ) ≥ 1 − λγ for the Gaussian measure (6.1), whose density is e^{−∑ m^{2s}|z_m|^2} normalized on the ball ∑ m^{2s−2}|z_m|^2 ≤ 1/2. This measure is supported on ℓ^2_{s−1}, not ℓ^2_s: the coordinates have variances ~ m^{−2s}, so E∑_{m≤M} m^{2s}|z_m|^2 = (1/2)∑_{m≤M} 1/m → ∞. Hence ∑ m^{2s}|z_m|^2 = +∞ almost surely, and ℓ^2_s has µ-measure zero. Since U_N^γ ⊆ ℓ^2_s, the set {εz ∈ U_N^γ} is µ-null, contradicting (6.4). The proof of Lemma 6.1 only bounds the probability of |Ω_j^{(2)}(I)| ≤ (γ/2)N^{−4l−2}κ_j^{−2s} (Lemma 6.2); this controls the complement only if the small-divisor conditions used ‖z‖_{s−1}^2 and κ_j^{−2(s−1)} in place of ‖z‖_s^2 and κ_j^{−2s}. As written, the threshold in (3.10)–(3.11) involves the a.s. infinite quantity ‖z‖_s^2, so the non-resonance set is empty modulo µ. The 'almost every' statement in Theorem 1.1 therefore does not follow from the supplied proof.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the one-dimensional Kirchhoff equation with Dirichlet boundary conditions and small initial data. It claims almost global existence and stability for most small data: arbitrarily long polynomial lifespan ε^{-r} in Sobolev spaces for every integer r ≥ 4, and sub-exponential lifespan in Gevrey and analytic spaces. The method is a rational normal form theory for infinite-dimensional reversible vector fields without external parameters, building on the authors' earlier framework. The proof proceeds through a resonant normal form theorem (Section 2), a calculus of rational vector fields with global small-divisor control (Sections 3–4), rational normal form theorems (Section 5), a Gaussian measure estimate for the non-resonance set (Section 6), and a Gevrey/analytic analogue (Section 7).","tokens_in":83760,"tokens_out":15480,"duration_ms":149944,"significance":"If the main result is correct, it is a substantial advance: it gives arbitrarily long polynomial lifespan for a quasi-linear PDE without external parameters, and the Gevrey/analytic extension gives sub-exponential time. The paper's strengths are the explicit rational normal form construction for reversible vector fields, including the lower-order terms generated by homological equations, the explicit constant bookkeeping, and the clear separation of the small-divisor control conditions. The cited papers [LX24a,b] provide the framework but do not contain the lifespan results proved here, so I do not see a circularity problem. The obstacle to acceptance is the Gaussian measure estimate for the non-resonance set, which as written does not support the 'almost every' conclusion.","major_comments":[{"comment":"The measure estimate is internally inconsistent. The Gaussian measure (6.1) has finite-dimensional marginals with E|z_m|² ~ m^{-2s}, so Σ_{m≤M} m^{2s}|z_m|² has expectation of order log M and diverges almost surely as M→∞; the measure is therefore supported on ℓ²_{s-1}, not on ℓ²_s, and µ(ℓ²_s)=0. Since U_N^γ is defined only for z∈ℓ²_s and both conditions (3.10) and (3.11) contain ‖z‖_s² on the right, the event {εz∈U_N^γ} is µ-null, contradicting (6.4) and invalidating the 'almost every' content of Theorem 1.1 and (1.8). The proof of Lemma 6.1 does not close this gap: after (6.21) it proves inequalities with ‖z‖_{s-1}² in the thresholds, not with the ‖z‖_s² appearing in (3.10)–(3.11), and Lemmas 6.2–6.3 are stated with κ_j^{-2s} thresholds. These are different sets, and the cancellation in Lemma 3.3 at (3.40)–(3.41) depends on the exact ‖z‖_s² normalization. A consistent reindexing replacing s by s-1 throughout Sections 3–7 would likely repair the argument, but the manuscript as written does not supply it.","section":"§6.1, Lemma 6.1, Eqs. (6.1)–(6.4), vs. §3.1, Eqs. (3.10)–(3.11)"}],"minor_comments":[{"comment":"The notation µ(ε(u,v)∈V_{r,s}) is ambiguous because V_{r,s} is itself defined as a union over scales ε in (6.49); please clarify whether this is meant as a conditional measure on B_s(ε) and state the normalization explicitly.","section":"§1, Eq. (1.8), and §6.2, Eq. (6.49)"},{"comment":"The measure (6.1) is written with dz d\\bar z but is really a Gaussian on ℓ²_{s-1}; the integration space and the role of z as both the random variable and the initial datum should be disentangled.","section":"§6.1, Eq. (6.1)"},{"comment":"The displayed formula contains the typographical fragment '1k!' where 1/k! is evidently intended.","section":"§5.1, Eq. (5.25)"},{"comment":"The resonant computation of K5 is summarized as a direct calculation; given that this explicit formula drives the later rational normal form steps, a short derivation or a pointer to the details would improve verifiability.","section":"§2.2, Eqs. (2.43)–(2.44)"}],"recommendation":"major_revision","confidential_remarks":"The main technical problem is the norm mismatch in Lemma 6.1. The rest of the normal form machinery appears coherent, and if the measure estimate can be repaired by a consistent weight shift, the paper would be publishable. I would not reject on novelty grounds."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: substantial paper with a real gap in the measure estimate. The normal form machinery is serious and the lifespan bounds would be a genuine advance, but as written Theorem 1.1 does not follow.\n\nWhat is new: prior best for 1D Kirchhoff was ε^{-4} for all data and ε^{-6} for non-resonant data (Baldi–Haus). Here they claim ε^{-r} for arbitrary r on almost all small Sobolev data, and sub-exponential in Gevrey/analytic. To do this they build rational normal form for infinite-dimensional reversible vector fields. The homological equation is genuinely different from the Hamiltonian case, generating lower-order non-integrable terms; their treatment of that is the main technical novelty. The iterative estimates are explicit, and the structure is coherent. Credit where due: this is a real contribution if it works.\n\nThe load-bearing problem is Lemma 6.1. U_N^γ is defined in Section 3 as a subset of ℓ^2_s, with thresholds involving ||z||_s^2 and κ_j^{-2s}. The Gaussian measure (6.1) has E|z_m|^2 ~ m^{-2s}, so ∑ m^{2s}|z_m|^2 = ∞ almost surely; the measure lives on ℓ^2_{s-1} and gives zero mass to ℓ^2_s. The proof of Lemma 6.1 actually uses ||z||_{s-1}^2 in the thresholds. That is the right space, but it is not the set defined earlier. So the measure estimate (6.4) contradicts the definition of U_N^γ as written. This looks repairable by reindexing (use s-1 consistently in U_N^γ and the small-divisor exponents, or change the Gaussian weight), but it is not a cosmetic slip: the 'almost every' statement in Theorem 1.1 is not proven as submitted.\n\nMinor soft spots: several central computations are summarized as 'direct calculation'—most notably the K5 resonant part—which makes verification slow. The paper leans heavily on the authors' own earlier work, but the central lifespan results and reversible-vector-field homological equations are not in those papers, so self-citation is not the issue.\n\nWho is it for: people working on long-time existence for quasilinear PDEs. It deserves a serious referee: the claim is important, the machinery is substantial, and the gap is plausibly fixable. I wouldn't cite it in its current form.","headline":"A substantial normal-form paper with a real, load-bearing gap in the measure estimate: the 'almost every' statement in Theorem 1.1 does not follow as written, though the flaw looks repairable.","tokens_in":84324,"tokens_out":7203,"would_cite":false,"duration_ms":65784,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35L70","35B35","37K55","35B20"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper establishes that for almost every small initial datum, Kirchhoff string solutions exist and remain stable for a polynomial lifespan in Sobolev spaces and a sub-exponential lifespan in Gevrey/analytic spaces.","keywords":["Kirchhoff equation","almost global solution","rational normal form","reversible vector fields","small divisors","long-time stability","Sobolev spaces","Gevrey and analytic spaces"],"falsifier":"Take $r=4$, $N=2(r+1)$, $s=s_0=O(r^2)$, draw actions $\\{I_a\\}$ from the Gaussian law used in Lemma 6.1, and compute, for every irreducible resonant multi-index $j$ of length at most 4, whether $|\\Omega^{(2)}_j(\\varepsilon^2 I)|>\\gamma\\varepsilon^2 N^{-4l-2}\\kappa_j^{-2s}$ and the corresponding $\\Omega^{(4)}$ bound fail on a set of Gaussian measure at least $\\varepsilon^{1/14}$; a violation would refute the 'almost every initial datum' assertion of Theorem 1.1. A second check is to test Lemma 3.4 on a single pair of reversible and anti-reversible rational vector fields: if their commutator does not satisfy the control condition $\\prod_m\\kappa_{h_m}\\le\\prod_m j^*_m$, the iterative normal form collapses.","tokens_in":83170,"feed_emoji":"🎻","tokens_out":11478,"duration_ms":110952,"temperature":0.7,"pith_summary":"This paper proves that the one-dimensional Kirchhoff equation for a clamped string—the quasilinear wave equation $\\partial_{tt}u-(1+\\int_0^\\pi|\\partial_xu|^2dx)\\partial_{xx}u=0$ with Dirichlet boundary conditions—has almost global solutions for almost all sufficiently small initial data. In Sobolev spaces, for any fixed integer $r\\ge4$, the lifespan and stability of the solution are at least of order $\\varepsilon^{-r}$, and the set of initial data of size $\\varepsilon$ for which this fails has Gaussian measure at most $\\varepsilon^{1/14}$. In Gevrey and analytic spaces, the time becomes $\\varepsilon^{-|\\ln\\varepsilon|/(c\\ln|\\ln\\varepsilon|)}$ for a constant $c>0$, with exceptional measure at most $\\varepsilon^{1/15}$. The technical engine is a rational normal form theorem for infinite-dimensional reversible vector fields without external parameters, which is genuinely different from the Hamiltonian rational normal form because the homological equation generates lower-order non-integrable terms that do not arise from Poisson brackets.","feed_headline":"Almost all small data: Kirchhoff strings survive polynomial time","feed_subtitle":"Gives ε^{-r} lifespan in Sobolev spaces and sub-exponential lifespan in Gevrey/analytic spaces.","key_machinery":"The engine of the proof is the rational normal form for reversible rational vector fields, defined as vector fields whose monomials are of the form $z_a\\zeta_j f_{h,k,n}(I)$ (and their conjugates) with denominators built from small divisors $\\Omega^{(2)}_{h_m}(I)$, $\\Omega^{(4)}_{h_m}(I)$ and $\\Omega^{(4)}_{k_m}(I)$, together with the coefficient symmetry $\\tilde X_{(b,c,d,h)}=\\tilde X_{(c,b,d,h)}$ and the global control condition $\\prod_m\\kappa_{h_m}\\le\\prod_m j^*_m$ (condition (3.21)). The argument is carried by three homological lemmas: one associated with the cubic integrable field $Z_3^{\\le N}$ eliminating the quintic term, one associated solely with $Z_3^{\\le N}$ eliminating the septic term, and one associated with $Z_3^{\\le N}+Z_5^{\\le N}$ eliminating higher-order terms. Unlike the Hamiltonian case, solving the vector-field homological equation produces extra lower-order non-integrable terms $\\tilde Z_{2l+1}$ and $\\tilde Z_{2l-1}$ that involve the quantity $D I_a[\\chi]=\\bar z_a\\chi(z_a)+z_a\\overline{\\chi(z_a)}$; the paper handles this with a modified solution $M_3$ for which $D I_a[M_3]=0$, so that the first two steps close after exactly two sub-steps. The global control condition is chosen so that it is preserved by commutators of reversible and anti-reversible rational vector fields (Lemma 3.4), which keeps the number of small divisors under control throughout the iteration.","core_discovery":"The central claim is that the Kirchhoff equation, after removing unbounded off-diagonal terms and rescaling time, is best studied as a reversible vector field rather than a Hamiltonian system, and that this structure is enough to run a rational normal form iteration. The paper proves three normal form theorems: a resonant normal form that produces an integrable cubic field $Z_3$ and a quintic resonant field $K_5$; a two-step rational normal form that eliminates the non-integrable part of $K_5$ and the non-normal part of $K_7$ using $Z_3$ alone; and an arbitrary finite-step rational normal form that eliminates the higher-order terms using $Z_3+Z_5$. Along the way it introduces a new class of rational vector fields whose monomials have small-divisor denominators, a global control condition $\\prod_m\\kappa_{h_m}\\le\\prod_m j^*_m$ that is preserved by commutators, and a modified solution $M_3$ of the quintic homological equation satisfying $D I_a[M_3]=0$, which stops the otherwise infinite regression of regenerated quintic terms. With this normal form in hand, the paper obtains Theorem 1.1 and Theorem 1.2 by estimating the Gaussian measure of the non-resonant set $U^N_\\gamma$ and then applying the normal form transformation to a bootstrap argument for the Sobolev/Gevrey norm and the actions $I_a$.","pith_inferences":["A direct corollary the paper does not spell out is that the per-mode action bound rules out any transfer of mass to high modes of amplitude larger than $\\varepsilon^3$ over the $\\varepsilon^{-r}$ time scale; this could in principle be tested numerically on finite-$N$ truncations.","The two-step elimination of the regenerated quintic term suggests that reversible vector-field normal forms have a one-step memory structure, so an analogous two-step pre-normalization may be needed for other reversible quasi-linear equations such as derivative nonlinear Schr\\\"odinger or Benjamin\\textendash{}Ono in reversible form.","The time scale obtained here has the same shape as the conjectured-optimal scale for the Schr\\\"odinger\\textendash{}Poisson equation; if that optimality is believed, the sub-exponential bound may be close to the true generic lifespan for Kirchhoff solutions, though the paper does not make that claim."],"forward_implications":["For every fixed $r\\ge4$ and Sobolev index $s$ of size $O(r^2)$, a set of initial data of Gaussian measure at least $1-\\varepsilon^{1/14}$ yields solutions that exist and stay in the $4\\varepsilon^2$ ball for $|t|\\le\\varepsilon^{-r}$.","The same set of solutions satisfies the per-mode stability bound $\\sup_a a^{2s}|I_a(t)-I_a(0)|\\le\\varepsilon^3$, so not only the norm but each individual action is almost conserved over the long time scale.","In Gevrey and analytic spaces, the existence and stability time is $\\varepsilon^{-|\\ln\\varepsilon|/(15800(1+2/\\theta)\\ln|\\ln\\varepsilon|)}$ with exceptional data of measure at most $\\varepsilon^{1/15}$; the larger $\\theta$ (closer to analytic) gives the longer time.","The generalized Kirchhoff equation with $\\varphi(\\int|\\partial_xu|^2dx)$, where $\\varphi(0)>0$ and $\\varphi'(0)\\ne0$, enjoys the same conclusions.","The rational normal form is not tied to Hamiltonian structure: the transformed system is a reversible vector field, so the four key points of the construction are claimed to apply to more general reversible systems without external parameters."],"supporting_citations":[{"why":"Supplies the bounded change of variables that removes the unbounded off-diagonal part of the Kirchhoff vector field, producing the system (2.8) from which the normal form analysis starts.","marker":"[BH20]"},{"why":"Gives the second-step resonant normal form of the Kirchhoff equation on the torus, the base case that Theorem 2.1 extends to arbitrary order r.","marker":"[BH21]"},{"why":"Provides the abstract Birkhoff normal form framework whose commutator estimates and iterative scheme are adapted here to polynomial vector fields.","marker":"[BG06]"},{"why":"Introduced rational normal forms for Hamiltonian PDEs using normal-form terms to modulate frequencies; the present paper adapts that idea to reversible vector fields.","marker":"[BFG20b]"},{"why":"Established exact global control of small divisors in rational normal forms for Hamiltonian functions, the predecessor whose two conditions are replaced by the single reversible control condition (3.21).","marker":"[LX24a]"},{"why":"Justifies countable additivity of the Gaussian measure on the Hilbert phase space used in the measure estimates (1.4)-(1.5) and (6.1).","marker":"[Kuk19]"}],"fun_headline_variants":["Almost global solutions for Kirchhoff equation with small data","Reversible vector fields yield almost global Kirchhoff stability","Long-time existence for Kirchhoff with small data via normal form","Kirchhoff equation: almost global lifespan for almost every small data","Rational normal form for reversible vector fields extends Kirchhoff lifespan"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the non-resonance frequency-separation lower bounds hold for almost every small initial datum and survive every normalization step; if the Gaussian measure of the bad set were not $O(\\gamma)$, the claimed $\\varepsilon^{-r}$ lifespan would not hold for a set of initial data of measure $1-\\varepsilon^{1/14}$.","fun_headline_variants_meta":{"raw":{"variants":["Almost global solutions for Kirchhoff equation with small data","Reversible vector fields yield almost global Kirchhoff stability","Long-time existence for Kirchhoff with small data via normal form","Kirchhoff equation: almost global lifespan for almost every small data","Rational normal form for reversible vector fields extends Kirchhoff lifespan"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00093,"raw_usage":{"total_tokens":4020,"prompt_tokens":1021,"completion_tokens":2999,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":637,"completion_tokens_details":{"reasoning_tokens":2917}},"tokens_in":637,"tokens_out":2999,"duration_ms":22830,"temperature":1.0,"reasoning_tokens":2917,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:22:50.193102+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take $r=4$, $N=2(r+1)$, $s=s_0=O(r^2)$, draw actions $\\{I_a\\}$ from the Gaussian law used in Lemma 6.1, and compute, for every irreducible resonant multi-index $j$ of length at most 4, whether $|\\Omega^{(2)}_j(\\varepsilon^2 I)|>\\gamma\\varepsilon^2 N^{-4l-2}\\kappa_j^{-2s}$ and the corresponding $\\Omega^{(4)}$ bound fail on a set of Gaussian measure at least $\\varepsilon^{1/14}$; a violation would refute the 'almost every initial datum' assertion of Theorem 1.1. A second check is to test Lemma 3.4 on a single pair of reversible and anti-reversible rational vector fields: if their commutator does not satisfy the control condition $\\prod_m\\kappa_{h_m}\\le\\prod_m j^*_m$, the iterative normal form collapses.","supporting_citations":[],"review_version":1}