{"id":"71b94fba-cd36-44c3-a7c2-c0d7f1ea2f83","arxiv_id":"2505.22478","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The defocusing quintic NLS on the real line has an invariant infinite-volume Gibbs measure for p between 3 and 5.","lead":"The paper proves that the defocusing quintic nonlinear Schrödinger equation on the real line has an infinite-volume Gibbs measure that is invariant under the flow, extending Bourgain's cubic result. It adds a stochastic quantization proof of the required growth estimate for the measure, a technique that may transfer to other infinite-volume equations.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.1 for p=5 is conditional on an unstated finite-volume invariance result: the proof uses Law(u^L(t))=μ^L as a black box for the periodic quintic NLS.","rationale":"The reader's weakest assumption coincides with the point I would stress: the finite-volume dynamics and invariance of μ^L under the periodic quintic NLS are taken as a black box. I read the rest of the proof carefully. Theorem 1.3's stochastic-quantization proof of the measure growth estimate is plausible and the main new ingredient; the coupling construction via quantitative Skorokhod and the Wasserstein estimate is detailed; the difference estimates and dyadic iteration leading to (1.6) are coherent. The only place where the argument would collapse is if the periodic p=5 invariance is not available: then Propositions 4.1 and 4.2 cannot upgrade initial-data estimates to all times, and the convergence-to-a-limit step in Theorem 1.1 has no foundation. Since the paper does not state this as a hypothesis or give a reference for p=5, the theorem as written is conditional. This is not a claim of error in the internal estimates; it is a demand to make the external input explicit and verified. If the finite-volume p=5 invariance is indeed a known consequence of Bourgain's periodic theory, then the concern is resolved by a citation and the verdict should be ACCEPT. Pending that check, CONDITIONAL is the honest recommendation.","tokens_in":42646,"tokens_out":64000,"duration_ms":697846,"concrete_test":"Check Bourgain's periodic works [Bou94, Bou96] and the subsequent literature for a theorem explicitly asserting that the 2πL-periodic Gibbs measure μ^L is invariant under the 1D quintic NLS for each L; equivalently, re-derive Proposition 4.2's step 'Law(u^L(t))=μ^L' for p=5 from the finite-dimensional truncated dynamics. If such a theorem exists, cite it in Theorem 1.1 and Propositions 4.1–4.2; if not, the range p=5 must be restricted or a proof of finite-volume invariance must be added.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central estimate chain — Theorem 1.3 via stochastic quantization, the coupling in Section 3.4, and the difference estimates in Sections 5–6 — is internally coherent. The load-bearing external input is finite-volume dynamics. Theorem 1.1 begins with 'let u^L be the unique global solution of (1.2) with initial data u^L(0)=φ^L', and Propositions 4.1 and 4.2 use the invariance of μ^L under (1.2) to replace u^L(t) with φ^L in moment estimates, e.g. around (4.8)–(4.9). For the cubic case this is Bourgain's periodic theorem, but no citation or proof is supplied for the periodic 1D quintic case p=5. If finite-volume invariance for p=5 is not already established, the almost-sure convergence and invariance statements have no starting point. The paper's original contribution (the infinite-volume measure growth estimate) does not by itself provide this input, and the text never isolates it as an assumption.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves the invariance of the infinite-volume Gibbs measure for the one-dimensional defocusing nonlinear Schrödinger equation (1.2) for exponents 3 ≤ p ≤ 5, extending Bourgain's infinite-volume result for the cubic case. The new ingredient is a uniform growth estimate (Theorem 1.3) for the supremum of the finite-volume Φ_1^{p+1} Gibbs measures on intervals of length R, obtained by the Hairer–Steele stochastic quantization method. Using this estimate together with a new quantitative coupling of the finite-volume measures and a localized mass comparison argument, the authors show that the finite-volume solutions u^L converge almost surely in C^0_t C^α_x on compact sets to a limit u, that u solves the equation in the sense of distributions, and that the law of u(t) is the infinite-volume Gibbs measure for every t.","tokens_in":42847,"tokens_out":13642,"duration_ms":138997,"significance":"If the main theorem is correct, this is a substantive advance: it removes the cubic restriction that had stood since Bourgain's infinite-volume work and shows that the stochastic quantization method can be used to control the relevant measure tails in a one-dimensional setting. The paper also introduces a quantitative coupling construction via a quantitative Skorokhod theorem and simplifies parts of Bourgain's argument by avoiding estimates for the kernel of P_{≤N}e^{itΔ}. The measure estimate Theorem 1.3 is proved in detail from the Langevin dynamics with a maximum-principle argument, and the difference estimates in Sections 5–6 are carefully structured. The main concerns are two load-bearing points: the finite-volume invariance of μ^L is used as an unstated external input for p=5, and the proof of the one-point density estimate in Lemma 3.13 contains a parameter choice that does not close the estimate as written. Both points appear fixable within the scope of the manuscript.","major_comments":[{"comment":"The proofs of Propositions 4.1 and 4.2 rely on the statement that μ^L is invariant under the periodic flow (1.2), and this invariance is used in a load-bearing way when replacing u^L(t) by the initial data φ^L in moment estimates such as (4.8)–(4.9). The introduction only attributes to [Bou94, Bou96] the cases d=1 and (d,p)=(2,3), and no explicit reference or proof is supplied for the finite-volume invariance in the quintic case p=5 covered by Theorem 1.1. The manuscript should state precisely which periodic theorem establishes existence, uniqueness, and invariance of μ^L under (1.2) for all 3≤p≤5, or should prove that finite-volume statement. Without this input, Theorem 1.1 for p=5 is conditional on an unstated external result.","section":"Section 4, Propositions 4.1–4.2 and Eq. (4.8)–(4.9)"},{"comment":"In the proof of Lemma 3.13, the choice t∼δ^{4p(1−κ)} does not prove the stated density estimate (3.35). The first term in (3.36) is bounded by Ct^{−1/4}δ, which with this choice equals δ^{1−p(1−κ)}. For p>1 and κ<3/4, this is not bounded by Cδ^κ as δ→0; indeed for p=5 and κ=3/4 this exponent is negative. Replacing the choice by t∼δ^{4(1−κ)} appears to balance the Gaussian term against δ^κ while making the exponential term in the nonlinear remainder acceptable. The parameter choice and the subsequent conclusion should be corrected, since Lemma 3.13 is one of the three inputs used to construct the coupling in Proposition 3.11 and hence to verify assumption (1.5) of Theorem 1.1.","section":"Section 3.4, proof of Lemma 3.13 after Eq. (3.36)"}],"minor_comments":[{"comment":"The estimate (3.48) is asserted to follow easily from Corollary 3.6 and (3.43), but the details are omitted; since this estimate is the exponential decay that produces (3.38), a short derivation or a precise reference would improve the exposition.","section":"Section 3.4, proof of Lemma 3.14"},{"comment":"The proof of the high-frequency component (5.8) of the good event is omitted with the comment that it is similar to (5.7). Given that (5.8) feeds into the difference estimates, a brief sketch of the logarithmic weight and dyadic summation would make the argument easier to verify.","section":"Section 5, proof of Lemma 5.2"},{"comment":"The sentence 'Bourgain [Bou94, Bou96] proved the invariance of the Gibbs measure under (1.1) for d=1 and (d,p)=(2,3)' is ambiguous: it should specify which periodic cases are covered and which reference covers p=5 in dimension one.","section":"Introduction, paragraph on Bourgain's periodic results"},{"comment":"The good event G_{L,T,R} is said to depend on a constant δ chosen depending on α from (1.6), but α is not otherwise present in Definition 5.1; the dependence should be stated explicitly so that the later choice of δ in the proof of Theorem 1.1 is transparent.","section":"Definition 5.1"}],"recommendation":"major_revision","confidential_remarks":"The paper is close to the standard for acceptance, and both major issues appear addressable by revision. The finite-volume invariance for p=5 may well already follow from Bourgain's periodic work, in which case the fix is a precise citation and a clarification of the statement in the introduction; if not, the main theorem for p=5 is conditional and needs a proof or a restriction of the statement. The Lemma 3.13 parameter-choice error looks like a typographical exponent error rather than a fatal flaw, but it must be corrected because it blocks the coupling construction."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a real extension. Bourgain's infinite-volume result was cubic, and this paper reaches the quintic endpoint, p=5, via a new uniform exponential-tail estimate for the infinite-volume Phi_1^{p+1} measures. The stochastic-quantization proof of Theorem 1.3 using the Hairer-Steele maximum principle is clean and well matched to the setting, and it has independent value beyond the dynamics application. The paper also simplifies Bourgain's technical apparatus, removing the kernel estimates for P_{<=N} e^{itDelta}. The quantitative coupling construction in Section 3.4 is a genuine improvement over the qualitative Skorokhod step in [Bou00]. The main line of the proof is coherent, and I do not see circularity: the measure estimate comes from Langevin dynamics independent of the NLS flow.\n\nThe soft spot is exactly the one flagged in the stress test. The proof uses Law(u^L(t)) = mu^L as a black box, and this finite-volume invariance for the periodic quintic equation is not stated as an assumption and no reference is supplied for p=5. For p=3 you can point to Bourgain's periodic theory, but the paper does not make that citation explicit even there, and for p=5 it is not clear that the cited periodic results include the endpoint. This is a load-bearing external input, not a circular step: the measure estimate does not depend on it. But the authors need to either state it as a hypothesis or prove it in an appendix. If Bourgain's method covers p=5 with minor modifications, they should say so and indicate where; if it does not, a short periodic-side argument is required.\n\nMinor issues: a few sub-estimates, notably Lemma 5.2 and the final step of Lemma 3.14, are sketched or deferred to the literature. In a paper of this length that is defensible, but a referee should ask for the missing details in (3.48). The paper is long and notation-heavy, but the structure is clear and the central argument is readable. The citation pattern looks normal; the relevant prior work is cited and the new contribution is distinct.\n\nThis paper is for dispersive PDE and stochastic quantization readers. It deserves a serious referee: the result closes a natural gap and Theorem 1.3 is a useful standalone estimate. Send it to review, with the referee asked to check the finite-volume invariance input carefully. My own verdict would depend on whether that input is already in the literature or can be added cleanly, but the paper is clearly worth refereeing.","headline":"First infinite-volume invariant Gibbs measure for the 1D quintic NLS, backed by a genuinely new measure estimate; one unmarked finite-volume invariance input needs to be stated or proved.","tokens_in":765,"tokens_out":869,"would_cite":true,"duration_ms":76713,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q55","60H15","37L40","35R60"],"pacs":[],"model":"deepseek-v4-flash","headline":"Defocusing quintic NLS on the real line has an invariant Gibbs measure, proved via stochastic quantization.","keywords":["Gibbs measure","nonlinear Schrödinger equation","infinite volume","quintic nonlinearity","stochastic quantization","invariant measure","almost sure convergence","large deviations"],"falsifier":"Test the endpoint p=5 by direct simulation on large tori: draw samples from μ^L, evolve them under (1.2), and check whether solutions starting from L and L/2 data obey the convergence bound (1.6) on fixed compact sets for radii R up to a small power of L; a systematic breakdown of $C^{0}$_t C^α_x convergence as L→∞ would falsify the theorem. Alternatively, compute the L∞ tail of μ^L over [-R,R] at p=5: the proof predicts probability at most C $e^{{-cλ}}$ for threshold (log R + λ)^{2/8}; observing a slower Gaussian-type tail of order (log R)^{1/2} would invalidate Theorem 1.3.","tokens_in":42455,"feed_emoji":"🌊","tokens_out":7068,"duration_ms":68562,"temperature":0.7,"pith_summary":"This paper proves that the defocusing quintic nonlinear Schrödinger equation on the real line preserves the infinite-volume Gibbs measure. It extends the previously known cubic case to the range 3≤p≤5. The proof treats 2πL-periodic approximations drawn from finite-volume Gibbs measures μ^L and shows these solutions converge almost surely to a limit u in $C^{0}$_t C^α_x on compact sets, that u solves the equation in the sense of distributions, and that the law of u(t) is the same infinite-volume measure μ for every time t. The new engine is a tail estimate for the $Φ^{{p+1}}$_1 measures: with high probability the L∞ norm over [-R,R] grows at most like (log R)^{2/(p+3)}, proved by stochastic quantization rather than Gaussian estimates. This faster decay of tails is what lowers the Gronwall exponent enough for p≤5.","feed_headline":"Quintic NLS on the line has an invariant Gibbs measure","feed_subtitle":"A stochastic-quantization tail bound extends the known cubic result to every nonlinearity p with 3≤p≤5.","key_machinery":"The central object is the frequency-truncated, exponentially localized mass M_R(w)=∫ |P_{≤R}w|^2 $e^{{-|x|/R}}$dx, where P_{≤R} is a Littlewood-Paley low-frequency projection and w=u^L-$u^{{L/2}}$ is the difference of two periodic solutions at scales L and L/2. On a high-probability 'good event' the time derivative of M_R obeys a Gronwall inequality whose growth rate is log(R)^{2(p-1)/(p+3)}; keeping M_R small on long time intervals requires 2(p-1)/(p+3)≤1, i.e. p≤5. The rate comes from splitting frequencies: low frequencies are controlled by a log-concave comparison inequality between μ^L and the Gaussian free field, while the sharper log exponent comes from a stochastic-quantization argument controlling exponential moments of ||φ||_{$L^{{p+1}}$([-1,1])}; a quantitative Skorokhod representation theorem converts the measure estimates into an explicit coupling of μ^L and μ. Iterating the mass estimate with a sequence of radii R_j = R_{j+1}^2 produces the final almost-sure convergence.","core_discovery":"Theorems 1.1 and 1.3 together assert that for 3≤p≤5 the defocusing NLS i∂_t u + ∂$_x^{2}$ u = |u|^{p-1}u on R admits an invariant Gibbs measure in a strong sense. If (φ^L) are couplings with Law(φ^L)=μ^L and Law(φ)=μ satisfying the quantitative coupling condition (1.5), then the associated global solutions u^L converge P-a.s. to u in $C^{0}$_t C^α_x([-T,T]×I) for all α<1/2, T≥1, and compact intervals I; the limit solves the equation in the space-time distribution sense and Law(u(t))=μ for all real t. The paper also establishes that the infinite-volume $Φ^{{p+1}}$_1 measure has large-deviation tails of order (log R + λ)^{2/(p+3)} for the L∞ norm over [-R,R], uniformly in the period L. The p=5 endpoint is the first nonlinearity beyond cubic for which such an infinite-volume invariance statement is proved.","pith_inferences":["A natural testable extension is p>5: here the Gronwall exponent 2(p-1)/(p+3) exceeds 1, so the present argument cannot close; whether the invariance statement itself fails or merely needs a different estimate is left open by the paper.","The same coupling-and-difference scheme could be applied to other translation-invariant infinite-volume Gibbs measures whose tails are known via stochastic quantization, such as higher-dimensional Φ models, provided finite-volume invariance is available.","The quantitative Skorokhod step suggests that a polynomial Wasserstein rate plus a local density bound may be enough to upgrade weak convergence of measures to almost-sure convergence of solutions; one testable consequence is whether the exponential rate in (1.13) is optimal at p=5.","Because finite-volume invariance is taken as a black box, a proof of periodic invariance for p=5 would make the theorem's conclusion self-contained; conversely, any failure of finite-volume invariance at p=5 would also invalidate the infinite-volume claim."],"forward_implications":["For every 3≤p≤5, the defocusing NLS on the real line has a solution dynamics on a full-measure set of Gibbs-sampled initial data, with the Gibbs law preserved for all times.","The limits of the periodic approximants are distributional solutions with C^0_t C^α_x regularity for every α<1/2 on compact space-time sets, and the solutions do not decay at infinity.","The infinite-volume Φ^{p+1}_1 measure samples grow at most like (log R)^{2/(p+3)} on [-R,R] with sub-Gaussian tail probabilities, uniformly in the period.","The quantitative coupling estimate (1.6) gives explicit polynomial-in-L control of the convergence rate of u^L to u, not merely convergence in law.","The proof removes the need for kernel estimates of P_{≤N}e^{itΔ} used in the cubic case, simplifying the frequency analysis."],"supporting_citations":[{"why":"Supplies the infinite-volume cubic case this work extends, including the local-mass and Gronwall strategy for controlling differences of periodic solutions.","marker":"[Bou00]"},{"why":"Provides the periodic invariance theory for Gibbs measures under NLS that the finite-volume dynamics are assumed to have.","marker":"[Bou94]"},{"why":"Extends the periodic invariance method to defocusing nonlinearities and underlies the transfer of initial-data bounds to all times via measure invariance.","marker":"[Bou96]"},{"why":"Supplies the log-concave Brascamp-Lieb inequality used to compare μ^L with the Gaussian free field and to control high-frequency tails and regularity.","marker":"[BL76]"},{"why":"Provides the stochastic-quantization argument for exponential moments and sub-Gaussian tails that is adapted here to prove the Φ^{p+1}_1 growth estimate.","marker":"[HS22]"},{"why":"Supplies the maximum-principle estimate for a nonlinear stochastic heat equation that controls the Langevin solution uniformly in L.","marker":"[MW20a]"},{"why":"Provides the Kolmogorov continuity theorem used to turn moment bounds into Hölder regularity in time for the solutions.","marker":"[Str93]"},{"why":"Provides the classical Skorokhod representation theorem that the quantitative version in Appendix A refines for the coupling of the Gibbs measures.","marker":"[Bil99]"}],"fun_headline_variants":["Invariant Gibbs measure proven for quintic NLS on R","Bourgain's cubic result extended to quintic NLS on line","Stochastic quantization proves invariant Gibbs measure for quintic NLS","Defocusing quintic NLS on real line admits invariant Gibbs measure"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes as a black box that the finite-volume Gibbs measures μ^L are invariant under the 2πL-periodic nonlinear Schrödinger flow and that the periodic solutions u^L exist uniquely and globally; if that finite-volume invariance were not already available, the transfer of estimates from initial data to all times has no starting point.","fun_headline_variants_meta":{"raw":{"variants":["Invariant Gibbs measure proven for quintic NLS on R","Bourgain's cubic result extended to quintic NLS on line","Stochastic quantization proves invariant Gibbs measure for quintic NLS","Defocusing quintic NLS on real line admits invariant Gibbs measure"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000385,"raw_usage":{"total_tokens":1981,"prompt_tokens":835,"completion_tokens":1146,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":451,"completion_tokens_details":{"reasoning_tokens":1071}},"tokens_in":451,"tokens_out":1146,"duration_ms":9235,"temperature":1.0,"reasoning_tokens":1071,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T13:07:38.566761+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Test the endpoint p=5 by direct simulation on large tori: draw samples from μ^L, evolve them under (1.2), and check whether solutions starting from L and L/2 data obey the convergence bound (1.6) on fixed compact sets for radii R up to a small power of L; a systematic breakdown of $C^{0}$_t C^α_x convergence as L→∞ would falsify the theorem. Alternatively, compute the L∞ tail of μ^L over [-R,R] at p=5: the proof predicts probability at most C $e^{{-cλ}}$ for threshold (log R + λ)^{2/8}; observing a slower Gaussian-type tail of order (log R)^{1/2} would invalidate Theorem 1.3.","supporting_citations":[],"review_version":1}