{"id":"39afde25-b05c-4016-9f03-1f1c0e357b3d","arxiv_id":"2607.22146","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"For Gaussian-localized data, vorticity retains a Gaussian spatial bound up to the solution's maximal lifespan, while velocity satisfies an explicit expansion in derivatives of the Laplacian fundamental solution with O(|x|^{-2d-1}) remainder.","lead":"This mathematics paper shows that the swirl (vorticity) of a fluid keeps a Gaussian-shaped bell as long as the solution exists, while the velocity spreads out and decays like a power law. It also gives a precise high-order formula for that velocity tail, which sharpens earlier first-order results and could improve how far-field behavior of Navier-Stokes solutions is described.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.2 is internally inconsistent: the claimed O(|x|^{-2d-1}) remainder requires the |α|=d term, which the displayed sum excludes.","rationale":"The central claim is the expansion. The paper's own Proposition 3.7 with N=d yields a term |α|=d whose decay equals the claimed remainder; the theorem's display omits it. This is not merely a proof gap: as stated, the theorem is false for generic data, although the intended version with N=d appears correct. The reader's weaker assumption (gradient estimate sketch) is reasonable but less decisive: it concerns a routine iteration that is described as verbatim. My check would settle whether the |α|=d coefficient is actually nonzero and whether the theorem statement must be corrected. Therefore the CONDITIONAL verdict remains appropriate; the authors should adjust the summation index or the remainder exponent, and include the Proposition 3.5 details.","tokens_in":39694,"tokens_out":12521,"duration_ms":132382,"concrete_test":"For d=2, take u0=∇^\\perp G_{\\nu\\sigma}, which is divergence-free, bounded, and Gaussian-localized. Compute the first Picard iterate u_1(s) and evaluate C(t)=∫_0^t∫ y_1^2 u_1^i u_1^j dy ds for a nonzero component (e.g. i=j=1). If C(t)≠0 for some small t, then the |α|=2 term in Proposition 3.7 with N=2 has size |x|^{-5} with coefficient C(t), while Theorem 1.2's sum stops at |α|=1 and claims remainder O(|x|^{-5}); contradiction. Alternatively, re-derive Theorem 1.2 by substituting p=d+1,N=d in Proposition 3.7 and compare terms; the |α|=d term must either be added to the sum or absorbed into a remainder of the same order, which is impossible for nonzero coefficient.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Proposition 3.7 is the engine of the paper. For Gaussian data, p=d+1 and the admissible range is 1≤N<d+1, so N=d is allowed. If one chooses N=d−1, the R1 remainder is O(|x|^{-(d+(d−1)+1)})=O(|x|^{-2d}), not O(|x|^{-2d-1}); the exponential Gaussian term and R2 are smaller. To get O(|x|^{-2d-1}) one must choose N=d, but then the expansion contains the extra term −∇∂^α K^{ij}(x)∫_0^t M^{ij}_α(s)ds for |α|=d. This term decays like |x|^{-d-|α|-1}=|x|^{-2d-1}, i.e. at exactly the claimed remainder order, and its coefficient ∫_0^t∫ y^α u^i u^j dy ds is not identically zero for generic Gaussian-localized data (incompressibility imposes no such d-th moment vanishing). Hence Theorem 1.2 as displayed cannot follow from Proposition 3.7. Remark 4 explicitly says to take N=d, so the intended theorem should sum to |α|=d (or should weaken the remainder to O(|x|^{-2d}) if the sum stops at d−1). This is the most direct load-bearing issue; the reader's flagged gap in Proposition 3.5 is secondary because the gradient bound is only sketched, not contradicted.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies spatial decay of strong solutions to the incompressible Navier-Stokes equations on R^d. The first main result, Theorem 1.1, states that if the initial vorticity has a Gaussian bound, then this Gaussian localization is inherited by the vorticity up to the maximal lifespan, in dimensions two and three. The second main result, Theorem 1.2, claims a high-order asymptotic expansion at spatial infinity for the velocity field generated by Gaussian-localized L^∞ initial data: the leading part is a finite sum of derivatives of the Laplacian fundamental solution with coefficients given by moments of u⊗u, and the remainder is O(|x|^{-2d-1}). The proof is built on a vorticity Gaussian-inheritance argument, polynomial decay estimates for u and ∇u, a Poisson-potential representation of the bilinear term, and a Taylor-expansion argument for the singular convolution. The key internal engine is Proposition 3.7, which gives a parameterized version of the expansion for solutions satisfying |u|+√(νt)|∇u| ≤ h(t)⟨x⟩^{-p}.","tokens_in":40035,"tokens_out":7208,"duration_ms":77426,"significance":"If the statement of Theorem 1.2 is corrected, the paper constitutes a substantial contribution: it improves the first-order spatial asymptotics of Brandolese-Vigneron to higher order and extends the validity up to the maximal lifespan. The vorticity Gaussian-inheritance result appears to be new and is a clean separation between the velocity and vorticity formulations. The proof strategy is self-contained and does not fit parameters to force the expansion; the coefficients are moments of the solution itself, and the key cancellation in Lemma 3.11 is elegant and correctly isolates the leading convolution term. The polynomial-decay framework of Section 3 is also a useful tool beyond the Gaussian case.","major_comments":[{"comment":"The displayed theorem is inconsistent with the engine of the paper. Theorem 1.2 sums |α|=0,...,d-1 and claims a remainder O(|x|^{-2d-1}). However, Proposition 3.7 gives R_1=O(|x|^{-d-N-1}). For Gaussian data one has p=d+1, so the admissible range is N<d+1; choosing N=d-1 gives R_1=O(|x|^{-2d}), not O(|x|^{-2d-1}). To obtain O(|x|^{-2d-1}) one must choose N=d, but then the expansion contains the |α|=d term -∇∂^α K^{ij}(x)∫_0^t M^{ij}_α(s)ds, which decays exactly like |x|^{-2d-1}; its coefficient is not identically zero for generic Gaussian-localized divergence-free data. Thus Theorem 1.2 as stated does not follow from Proposition 3.7. The theorem should either sum to |α|=d and state the remainder as O(|x|^{-2d-1}) (with the caveat that this is an approximation, not an asymptotic expansion in the strict o(last-term) sense), or keep the sum to d-1 and weaken the remainder to O(|x|^{-2d}). R","section":"Theorem 1.2, Remark 4, Proposition 3.7"},{"comment":"The gradient decay estimate (3.24) is load-bearing: it is used in the uniform bound (3.39), in the remainder estimates (3.41)-(3.42), and in Lemma 3.10. Yet the proof is not written: the text says it is 'an almost verbatim repeat of the proof of Proposition 3.4' and leaves it to the reader. Given that the entire expansion depends on this bound, the proof should be supplied in full, or at least the necessary modifications should be detailed in an appendix. This is not merely a cosmetic omission.","section":"Section 3.3, Proposition 3.5"}],"minor_comments":[{"comment":"The proof of Proposition 3.3 is headed 'Proof of Proposition 3.2'; the label should be corrected.","section":"Section 3.2"},{"comment":"There is a typo: 'don not' should be 'do not'.","section":"Abstract"},{"comment":"In the computation of the moments M_β(τ), the summation bounds after interchanging the β-sum and the γ-sum are written in a way that is ambiguous for even N'. Please clarify the integer range, e.g. with explicit floor notation, to avoid a possible indexing error.","section":"Section 3.4, Lemma 3.11 proof"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the main expansion theorem is misstated as printed. The sum runs to |α|=d−1, but the claimed remainder O(|x|^{-2d−1}) only follows from Proposition 3.7 by taking N=d, which includes the |α|=d term. That term decays exactly like |x|^{-2d−1} and its coefficients are generic moments of the solution, so it cannot be absorbed into the remainder. This is a genuine off-by-one in the statement, not a subtle gap—the proof, including Remark 4, actually establishes the version with the sum to d. Second, the paper is still worth engaging with: the explicit d-th order expansion via the Poisson-potential formulation and the cancellation in Lemma 3.11 is a real step beyond Brandolese–Vigneron’s first-order profile, and the vorticity Gaussian-inheritance result is new and nicely proved.\n\nWhat is genuinely good: the expansion is written in terms of moment coefficients, holds up to the maximal lifespan, and the remainder analysis in Section 3.4 is mostly careful. The Gaussian bound for vorticity (Theorem 1.1) is a clean iterative argument and stands on its own. There is no circularity and no fitting—the coefficients are moments of the solution itself.\n\nSoft spots. The indexing error is load-bearing and must be fixed. The reader’s flagged concern about Proposition 3.5 is also legitimate: the pointwise gradient decay estimate is needed for the far-field remainder estimates, and its proof is only sketched as “almost verbatim repeat” of Proposition 3.4. Given that the theorem statement already needs correction, the authors should use the revision to write out that gradient estimate. The novelty discussion relative to McOwen–Topalov and Topalov is also thinner than it should be: those papers are cited, but the text does not crisply say what the new expansion adds over their spherical-harmonic expansions. The likely answer—explicit moment coefficients and maximal-lifespan validity—needs to be stated.\n\nWho this is for: specialists in spatial asymptotics for Navier–Stokes, especially the Brandolese–Vigneron line. The vorticity result has broader interest. I would send it to peer review rather than desk-reject, because the core is sound and the fix is straightforward, but not in its current form. The referee should insist on a corrected theorem statement and a real proof of Proposition 3.5.","headline":"Theorem 1.2 as printed has an off-by-one indexing error: the claimed O(|x|^{-2d-1}) remainder requires including the |α|=d term, which the displayed sum omits; the proof actually establishes the corrected version with the sum to d.","tokens_in":40523,"tokens_out":4509,"would_cite":false,"duration_ms":45152,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q30","35B40","76D05"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that velocity and vorticity inherit different spatial decay from Gaussian data; velocity develops an explicit rational expansion at infinity.","keywords":["Navier-Stokes equations","vorticity","Gaussian bounds","asymptotic expansion","pointwise decay","fundamental solution","maximal lifespan","strong solutions"],"falsifier":"Take a smooth compactly supported initial velocity in R³ and a time t just before any possible blow-up; if |x|^{2d+1}|u(x,t) + Σ_{|α|=0}^{d-1} (−1)^{|α|}/α! ∇∂^α K^{i,j}(x) ∫_0^t M^{i,j}_α(s) ds| does not go to 0 uniformly as |x|→∞, the expansion is false. More directly, exhibit a strong solution with |∇u(x,t)| decaying no faster than ⟨x⟩^{-p} for p=d/2+1 at some t; the remainder estimate then breaks and the expansion order cannot be attained.","tokens_in":39582,"feed_emoji":"🌊","tokens_out":4934,"duration_ms":51365,"temperature":0.7,"pith_summary":"The paper draws a sharp line between the velocity and vorticity equations for incompressible Navier-Stokes in the whole space. It proves that if the initial vorticity has a Gaussian bound, the vorticity keeps a Gaussian spatial bound for every time up to the maximal lifespan. The velocity, by contrast, cannot keep that bound: starting from well-localized (Gaussian or polynomial) initial data, the velocity develops rational, non-Gaussian tails and has an explicit expansion at infinity as a finite sum of derivatives of the fundamental solution of the Laplacian, up to order d−1, with coefficients that are moments of the product u⊗u. This is shown to hold for the whole maximal lifespan, with a remainder decaying like |x|^{-(2d+1)}. The paper also derives a necessary and sufficient condition, in terms of isotropy of the space-time L² correlations, for a localized strong solution to decay faster than the critical rate.","feed_headline":"Vorticity keeps Gaussian decay; velocity breaks it","feed_subtitle":"A d-term expansion for localized Navier-Stokes solutions predicts |x|^{-(d+1)} tails unless space-time moments align.","key_machinery":"The argument uses a mild-solution representation that replaces the usual Riesz-transform kernels with second derivatives of the Poisson potential, obtained through integration by parts where the singularity of the Poisson kernel contributes a trace term. The convolution is split into close and far regions: a Taylor expansion of the kernel acts in a region where the solution's pointwise polynomial decay is controlled, while the remaining regions are shown to contribute only higher-order decay. A final convolution lemma (Lemma 3.11) reduces to expanding ∇G_{νt}∗∂^α K^{i,j}; because K^{i,j} is harmonic away from the origin, all higher-order terms cancel and only the leading term survives. The w","core_discovery":"The central claim is that velocity and vorticity inherit different spatial localization from Gaussian initial data. For d=2,3, a strong vorticity solution keeps a Gaussian bound up to the maximal lifespan. For d≥2, a strong velocity solution with Gaussian-localized L∞ data satisfies an explicit asymptotic expansion: u(x,t) = −Σ_{|α|=0}^{d-1} (−1)^{|α|}/α! ∇∂^α K^{i,j}(x) ∫_0^t M^{i,j}_α(s) ds + R(x,t), with sup_{t∈[0,T]}|R(x,t)| = O(|x|^{−2d−1}) as |x|→∞, uniformly on any compact time interval inside the maximal lifespan. Here K^{i,j}=∂²_{i,j}Γ is the second derivative of the fundamental solution of the Laplacian and M^{i,j}_α are moments of u^i u^j. The expansion holds up to maximal lifespa","pith_inferences":["One could test the theorem numerically in 2D or 3D by computing the moment integrals from a simulation and comparing the predicted far field to the computed velocity; a mismatch at any time would pinpoint a failure of the pointwise gradient decay estimate.","The same Poisson-potential decomposition may apply to other incompressible models with a pressure-induced nonlocal coupling, such as the Boussinesq or magnetohydrodynamic systems, to derive analogous far-field expansions.","The necessary and sufficient condition suggests a possible measurement strategy: observing the angular dependence of the far field at the critical decay rate reveals whether the solution's space-time correlations are isotropic.","If the gradient decay estimate (Prop. 3.5) were false, the remainder order would drop; hence the expansion itself is a subtle test of gradient localization, and the gap in the written proof is worth closing."],"forward_implications":["The far-field profile of any well-localized strong solution is fully determined by the moment integrals of u⊗u spent over time; no other feature of the solution enters the leading term.","The orthogonality criterion gives a concrete, checkable condition for faster-than-critical decay: the matrix of time-integrated cross-correlations must be a scalar multiple of the identity.","Vorticity-based simulations or estimates can safely assume Gaussian localization for all positive times, while velocity-based estimates cannot.","The expansion is uniform up to the maximal lifespan, so it remains meaningful even if the solution blows up at the end: the spatial profile at any time before blow-up still has this explicit form.","The method extends beyond Gaussian data to the larger class of initial values with polynomial decay ⟨x⟩^{-p}, p≤d+1."],"fun_headline_variants":["Vorticity keeps Gaussian decay; velocity expands with power tails","Gaussian bound holds for vorticity, fails for velocity","Velocity loses Gaussian localization; vorticity keeps it","Expansion at infinity: velocity tails beyond Gaussian bound","Vorticity stays Gaussian; velocity develops power-law tails"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The expansion rests on the pointwise bound (3.32), especially the gradient part |∇u(x,t)| ≤ h(t)⟨x⟩^{-p}/√(νt) for some p>d/2+1 up to the maximal lifespan; the proof of the gradient part is only sketched as a repetition of the line-by-line argument for the velocity, so if that bound fails the O(|x|^{-2d-1}) remainder could fail as well.","fun_headline_variants_meta":{"raw":{"variants":["Vorticity keeps Gaussian decay; velocity expands with power tails","Gaussian bound holds for vorticity, fails for velocity","Velocity loses Gaussian localization; vorticity keeps it","Expansion at infinity: velocity tails beyond Gaussian bound","Vorticity stays Gaussian; velocity develops power-law tails"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000935,"raw_usage":{"total_tokens":3904,"prompt_tokens":876,"completion_tokens":3028,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":620,"completion_tokens_details":{"reasoning_tokens":2959}},"tokens_in":620,"tokens_out":3028,"duration_ms":20335,"temperature":1.0,"reasoning_tokens":2959,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T05:39:59.146849+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a smooth compactly supported initial velocity in R³ and a time t just before any possible blow-up; if |x|^{2d+1}|u(x,t) + Σ_{|α|=0}^{d-1} (−1)^{|α|}/α! ∇∂^α K^{i,j}(x) ∫_0^t M^{i,j}_α(s) ds| does not go to 0 uniformly as |x|→∞, the expansion is false. More directly, exhibit a strong solution with |∇u(x,t)| decaying no faster than ⟨x⟩^{-p} for p=d/2+1 at some t; the remainder estimate then breaks and the expansion order cannot be attained.","supporting_citations":[],"review_version":1}