{"id":"955a8261-a06e-403d-a22e-1aeea24e5eae","arxiv_id":"1908.05766","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper proves that any finite-time blow-up of a smooth 3D Euler solution forces unbounded growth of the associated linearized Euler semigroup in every Lp norm.","lead":"A new theorem shows that if a smooth solution of the 3D Euler equations ever blows up in finite time, the linearized equations around that solution become increasingly unstable as the blow-up time approaches. This means direct numerical simulations of such a blow-up would lose predictability, because tiny computational errors get amplified without bound.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 2's pressure-boundary identity is false on curved boundaries, so the bounded-domain version of Theorem 1 (and the Luo–Hou boundary blow-up application) is unproved as written.","rationale":"The central inequality is plausible and the WKB construction is coherent; I do not see a reason to doubt Theorem 1 for R^3 or T^3. However, the most load-bearing weak point is not the C^2 regularity of the WKB data (which is guaranteed by s>9/2 on each [0,T]), but the pressure/boundary argument in Lemma 2. The claimed identity is demonstrably false for a steady rotational Euler solution in a ball. Since Lemma 2 underpins Proposition 2 for bounded domains, and the authors explicitly invoke the Luo–Hou boundary blow-up scenario, this must be corrected before the bounded-domain theorem is accepted. The reader's normalization gap in Proposition 2 is also genuine, but it is harmless because the initial Lp norm of the constructed data is 1+O(ε), so the limit is unaffected. I therefore agree with the CONDITIONAL verdict but not with the reader's stated weakest assumption.","tokens_in":10443,"tokens_out":34602,"duration_ms":341059,"concrete_test":"Analytic check on the unit ball: take u=v=(-x_2,x_1,0), a steady Euler solution with div u=0 and u·n=0 on ∂B. At x=(1,0,0), compute [(∂xu)v]·n = (u·∇)u·n = -1, directly falsifying the identity used in Lemma 2. Then re-prove Lemma 2 using the corrected Neumann condition ∂nP'=f·n+v·(u·∇n)+u·(v·∇n), and verify whether the stated bound ||∇P'||_{Lp} ≤ C(||u||_{H^s}||v||_{Lp}+||f||_{Lp}) still holds for all 1<p<∞, or whether boundary trace regularity restricts the admissible range of p.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the proof of Lemma 2 (Section 3), the authors assert that for u,v tangent to ∂Ω, [(∂xu)v]·n = [(vτ·∇τ)u]·n = (vτ·∇τ)(u·n) = 0, and similarly for (u·∇)v. The second equality is false on curved boundaries: differentiating a tangent vector field in a tangential direction can produce a normal component proportional to the second fundamental form. A concrete counterexample is solid-body rotation in the unit ball: take u=v=(-x_2,x_1,0), which satisfies div u=0 and u·n=0 on ∂B. At the boundary point x=(1,0,0), (u·∇)u=(-1,0,0), so [(∂xu)v]·n=-1, not 0. Consequently the derived pressure Neumann condition ∂nP'=f·n is missing the curvature terms v·(u·∇n)+u·(v·∇n); the correct condition follows from v·n=u·n=0 and is ∂nP'=f·n+v·(u·∇n)+u·(v·∇n). Lemma 2 is used crucially in Proposition 2 to control the error between the true linearized solution and the WKB approximation, and Theorem 1 explicitly covers bounded smooth domains, including the Luo–Hou boundary blow-up scenario. As written, the bounded-domain proof is invalid at this step; the result is plausibly repairable, but the manuscript needs a corrected elliptic estimate with the correct boundary data.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims a general instability mechanism for 3D Euler flows that blow up in finite time. For a smooth solution u on [0,T*) with initial vorticity ω0, Theorem 1 states that for every 1<p<∞ and every T<T*, the L^p growth γ_p(T) of the linearized Euler semigroup about u satisfies γ_p(T)^2 ≥ ||ω(T)||_{L∞}/||ω0||_{L∞}. Corollary 1 then asserts that if ||u(T)||_{H^s} is unbounded as T→T*, then ∫_0^{T*} γ_p(t)^2 dt = ∞ and limsup_{T→T*} γ_p(T) = ∞. The proof combines two propositions: Proposition 1 bounds the vorticity growth by the WKB amplitude β(T) through conservation laws for ω·ξ and det(b,btilde,ξ); Proposition 2 bounds β(T) below by γ_p(T) via an explicit WKB approximate solution of the linearized Euler equation with a highly oscillatory potential. The result is presented for R^3, T^3, and bounded smooth domains, with the Luo–Hou boundary blow-up scenario explicitly mentioned.","tokens_in":10732,"tokens_out":20005,"duration_ms":172890,"significance":"If the result is correct, it is a significant contribution: it shows that any finite-time H^s blow-up of 3D Euler forces the linearized semigroup to be unbounded in L^p, hence linear hydrodynamic instability, and it offers an explanation for the extreme difficulty of numerical blow-up detection. The argument is mostly self-contained, uses no fitted parameters, and is anchored in standard results (Beale–Kato–Majda, Inoue–Miyakawa, Krylov elliptic estimates). The conservation identities in Section 2 are elegant and check out. However, the paper currently contains load-bearing gaps in the bounded-domain case and in the WKB sign conventions; these are likely repairable, but as written the proof is not complete.","major_comments":[{"comment":"The boundary computation in Lemma 2 asserts that [(∂_x u)v]·n = [(v_τ·∇_τ)u]·n = (v_τ·∇_τ)(u·n) = 0, and similarly for [(∂_x v)u]·n. This second equality is false on curved boundaries: the tangential derivative of a vector field that is tangent to ∂Ω can have a nonzero normal component proportional to the second fundamental form. For example, for solid-body rotation u=v=(-x_2,x_1,0) in the unit ball, u·n=0 on ∂B, but at x=(1,0,0) one has [(∂_x u)v]·n = -1. The correct pressure Neumann condition must include the curvature terms v·(u·∇n)+u·(v·∇n). As written, the L^p elliptic estimate for ∇P' is not justified in bounded domains. Since Lemma 2 is used in Proposition 2 to control the error between the true linearized solution and the WKB approximation, and since Theorem 1 explicitly covers bounded smooth domains (including the Luo–Hou boundary blow-up scenario), the bounded-domain version of the theorem is unproved as written. This is repairable by deriving the corrected boundary condition and re-running the elliptic estimate, but the present proof is incomplete.","section":"Section 3, proof of Lemma 2"},{"comment":"The WKB construction has an internal sign inconsistency. Starting from the b-equation (5), the correct evolution for V_{ε,δ} = iφ b e^{iS/ε} is ∂_t V + (u·∇)V = -(V·∇)u + 2(ξ^T(∂_x u)V/|ξ|^2)ξ, not with a minus before the last term as displayed in the text. With the displayed definition q_{ε,δ} = -2iε ξ^T(∂_x u)V/|ξ|^2, one computes ∇q_{ε,δ} = 2(ξ^T(∂_x u)V/|ξ|^2)ξ + O(ε). Consequently the residual in the claimed equation ∂_t v_{ε,δ} + (u·∇)v_{ε,δ} + (v_{ε,δ}·∇)u + ∇q_{ε,δ} = R_{ε,δ} contains an O(1) term, and the bound ||R_{ε,δ}||_{L^p} ≤ C_{η,δ} ε does not follow from the displayed formulas. The signs can presumably be repaired (for example by taking q_{ε,δ} = +2iε ξ^T(∂_x u)V/|ξ|^2 and correcting the corresponding sign in the equation for V), but as written the proof of Proposition 2 does not produce an approximate solution of the linearized Euler equation. Therefore the inequality β(T) ≤ γ_p(T) is not established in the present text.","section":"Section 3, proof of Proposition 2 (equations near (14))"}],"minor_comments":[{"comment":"After constructing v_{ε,δ}, the initial datum v_{ε,δ}(0) is not shown to satisfy ||v_{ε,δ}(0)||_{L^p} ≤ 1 before applying the definition of γ_p(T). The argument should either normalize the initial data or explicitly note that ||v_{ε,δ}(0)||_{L^p} → 1 as ε → 0, so the lower bound for γ_p(T) follows with the same ε, η limits.","section":"Section 3, proof of Proposition 2"},{"comment":"The domain of the data is written as v0 ∈ H^1(Ω) × L^p(Ω), which appears to be a typo: the intended assumption is v0 ∈ H^1(Ω) and f ∈ L^2(0,T;H^1(Ω)) ∩ L^1(0,T;L^p(Ω)).","section":"Lemma 2 statement"},{"comment":"There are several typographical errors, including 'HYDRODYNAMICALL Y UNST ABLE' in the title, 'prop agation' and 'regu-larity' in the abstract, and 'fo instance' in the proof of Lemma 2. These do not affect the mathematics but should be corrected.","section":"Throughout"},{"comment":"In the computation of the determinant evolution, the notation '(Eij)kl = δkl' is confusing; it likely means the matrix E_{ij} with a 1 in entry (i,j). The trace computations are correct once this interpretation is adopted.","section":"Lemma 1 proof"}],"recommendation":"major_revision","confidential_remarks":"The paper's central idea is attractive and the conservation identities in Section 2 are solid. The two major issues I raise—the curved-boundary pressure condition in Lemma 2 and the sign inconsistencies in the WKB construction of Proposition 2—are specific and load-bearing, but both appear repairable within the manuscript's scope. The bounded-domain case needs a genuinely corrected elliptic estimate, not just a typo fix. I would encourage the authors to rewrite the relevant proofs carefully and, ideally, to make the residual estimates in Proposition 2 fully explicit. The paper would be a strong contribution after such a revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main thing you should know: this paper proves a genuinely new obstruction — for smooth 3D Euler, finite-time H^s blow-up forces the linearized semigroup to have infinite ∫γ_p^2 and unbounded lim sup γ_p. The idea is clean: vorticity growth is squared-controlled by the WKB b-amplitude, and b-amplitude is bounded by linearized Lp growth. I checked the algebraic conservation laws and the chain of inequalities; they hold. On R^3 and T^3 the proof looks solid.\n\nWhat's new and good: the statement itself — no one had the forward implication from blow-up to linear instability. The WKB machinery from Vishik's earlier work is used honestly; they prove the ODE properties they need instead of just citing. The conservation of (b×b̃)·ξ is elegant, and Proposition 1's bound on |ω| by β^2 is a neat piece of microlocal analysis.\n\nWhere it's soft: the bounded-domain case has a real gap. Lemma 2's proof asserts that for u,v tangent to ∂Ω, [(v·∇)u]·n = (v·∇)(u·n)=0. That's false on curved boundaries. Solid-body rotation in the ball is a counterexample: at the boundary point (1,0,0), (v·∇)u = (-1,0,0), so the normal component is -1. The pressure Neumann condition is missing curvature terms u·(v·∇n)+v·(u·∇n). The elliptic estimate in Lemma 2 probably survives — those extra terms are lower order and controlled by ‖u‖_{W^{1,∞}}‖v‖_{L^p} — but as written the proof of Lemma 2 is invalid for bounded domains, and Theorem 1's coverage of those domains, including the Luo–Hou boundary blow-up application, is unsupported until that's fixed. This is repairable, not fatal.\n\nAlso minor: in Proposition 2 the test data v_{ε,δ} is not explicitly normalized to L^p norm ≤ 1 before using γ_p(T). The norm is 1+O(ε), so dividing and letting ε→0 fixes it; the reader caught this and it's the kind of thing a referee should ask for.\n\nCitation pattern looks fine — BKM, Inoue–Miyakawa, Friedlander–Vishik, Vishik are the right anchors, and the new assertions come from the argument, not from fitting or circularity.\n\nWho it's for: people working on Euler singularity, hydrodynamic stability, and WKB methods. It deserves a serious referee. I'd send it out; with the boundary lemma fixed (or the domain restricted), it should be accepted.","headline":"New and likely correct in the whole-space/torus case; bounded-domain proof has a genuine gap in Lemma 2's boundary identity, but the result deserves refereeing.","tokens_in":11321,"tokens_out":4504,"would_cite":true,"duration_ms":42380,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["76B03","35B35","35B44"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that finite-time blow-up in smooth 3D Euler forces linearized instability: for every $T<T^*$ and $1<p<\\infty$, $\\gamma_p(T)^2 \\ge \\|\\omega(T)\\|_{L^\\infty}/\\|\\omega_0\\|_{L^\\infty}$.","keywords":["Euler equations","finite-time blow-up","hydrodynamic instability","linear stability","vorticity","WKB expansion","Lp semigroup","incompressible flow"],"falsifier":"Compute, for a proposed finite-time blow-up candidate of 3D Euler, the maximal $L^p$ growth $\\gamma_p(T)$ of the linearized equation on a sequence $T\\to T^*$. If at any such $T$ the product $\\gamma_p(T)^2\\|\\omega_0\\|_{L^\\infty}$ is strictly less than $\\|\\omega(T)\\|_{L^\\infty}$, the theorem's bound is violated; a rigorous contradiction would be a smooth solution that blows up in $H^s$ while the linearized semigroup stays bounded in some $L^p$, $1<p<\\infty$.","tokens_in":10208,"feed_emoji":"💥","tokens_out":20172,"duration_ms":172166,"temperature":0.7,"pith_summary":"The paper aims to prove that any smooth finite-time singularity of the incompressible 3D Euler equations must be accompanied by linear hydrodynamic instability. More precisely, it establishes that for every $1<p<\\infty$ and every time $T$ before the blow-up time $T^*$, the maximal $L^p$ growth $\\gamma_p(T)$ of the linearized Euler semigroup satisfies $\\gamma_p(T)^2 \\geq \\|\\omega(T)\\|_{L^\\infty}/\\|\\omega_0\\|_{L^\\infty}$, where $\\omega$ is the vorticity. Because the classical vorticity blow-up criterion says a loss of $H^s$ regularity forces vorticity to accumulate, the theorem yields $\\int_0^{T^*}\\gamma_p(t)^2\\,dt=\\infty$ and $\\limsup_{T\\to T^*}\\gamma_p(T)=\\infty$ for every $p$. In plain terms: close to a would-be blow-up, arbitrarily small initial perturbations of the velocity amplify without bound, so direct numerical simulation cannot reliably predict the blow-up. The result also distinguishes incompressible blow-up from stable shock formation in compressible models such as scalar conservation laws.","feed_headline":"3D Euler blow-ups must be linearly unstable","feed_subtitle":"Tiny perturbations grow without limit near a blow-up, so numerical blow-up predictions cannot be trusted.","key_machinery":"The engine is the WKB bicharacteristic-amplitude system describing high-frequency perturbations of the linearized Euler equation. Along the flow $\\dot\\gamma_t=u(t,\\gamma_t)$, the wave vector $\\xi_t$ solves $\\dot\\xi_t=-(\\nabla u)^T\\xi_t$, and the amplitude $b_t$ solves $\\dot b_t=-(\\nabla u)b_t+2\\xi_t^T(\\nabla u)b_t|\\xi_t|^{-2}\\xi_t$; this ODE turns the nonlocal pressure projection into a local algebraic term. The key conserved quantities, proved in Lemma 1, are the pairing $\\omega_t\\cdot\\xi_t$ between the transported vorticity and the wave vector, the orthogonality $b_t\\cdot\\xi_t=0$, and the determinant $(b_t\\times\\tilde b_t)\\cdot\\xi_t$; these give a local, pointwise version of helicity conservation. Proposition 1 uses this to bound $|\\omega(T,x)|^{1/2}$ by the amplitude growth $\\beta(T)$, and Proposition 2 constructs approximate solutions of the linearized equation from the WKB data, yielding $\\beta(T)\\le\\gamma_p(T)$. The chain $\\|\\omega(T)\\|_{L^\\infty}\\le\\beta(T)^2\\|\\omega_0\\|_{L^\\infty}\\le\\gamma_p(T)^2\\|\\omega_0\\|_{L^\\infty}$ is the whole theorem.","core_discovery":"The central discovery is a rigidity relation between vorticity growth and the linearized semigroup. For the incompressible Euler equation with initial data in $H^s$, $s>9/2$, on $\\mathbb{R}^3$, $\\mathbb{T}^3$, or a bounded domain with the impermeability condition, Theorem 1 asserts that for every $1<p<\\infty$ and every $T<T^*$, $\\gamma_p(T)^2 \\geq \\|\\omega(T)\\|_{L^\\infty}/\\|\\omega_0\\|_{L^\\infty}$. The proof runs through two comparisons: vorticity growth is bounded by the square of the WKB amplitude growth $\\beta(T)$, and $\\beta(T)$ is bounded by the $L^p$ semigroup growth $\\gamma_p(T)$. The direction is the reverse of the standard implication: regularity is known to imply stability, while the paper proves the converse in a strong quantified form. Consequently, any finite-time $H^s$ blow-up automatically makes the linearized evolution unbounded, so the blow-up is hydrodynamically unstable in every $L^p$, $1<p<\\infty$.","pith_inferences":["Beyond the paper, this inequality offers a numerical probe for any proposed blow-up candidate: compute $\\gamma_p(T)$ for the linearized equation and check that $\\gamma_p(T)^2\\|\\omega_0\\|_{L^\\infty}$ stays above $\\|\\omega(T)\\|_{L^\\infty}$ as $T\\to T^*$.","Beyond the paper, the WKB construction identifies the fastest-growing perturbations as high-frequency oscillatory modes carried by the flow; numerical schemes that dissipate high frequencies may suppress exactly the modes driving the instability, which could explain divergent results in computational blow-up studies.","Beyond the paper, the bound ties linearized growth to Lagrangian stretching of vorticity, suggesting that finite-time Lyapunov exponents of the flow could provide a computationally cheaper lower bound on $\\gamma_p(T)$ than solving the full linearized Euler equation."],"forward_implications":["If the $H^s$ norm of a smooth 3D Euler solution becomes unbounded at $T^*$, then for every $1<p<\\infty$ the linearized semigroup satisfies $\\int_0^{T^*}\\gamma_p(t)^2\\,dt=\\infty$ and $\\limsup_{T\\to T^*}\\gamma_p(T)=\\infty$; the blow-up is linearly unstable.","Numerical blow-up experiments are intrinsically unreliable: any numerical noise in the initial data is amplified without bound as the singularity approaches, so distinct runs cannot converge to a deterministic blow-up scenario.","The instability is quantitative: the growth of perturbations is bounded below by the square root of the relative growth of the sup norm of vorticity, not merely unbounded in an abstract sense.","The contrapositive gives a stability route to regularity: if the linearized semigroup remains bounded up to $T^*$ in some $L^p$, then no $H^s$ blow-up can occur before $T^*$.","Incompressible inviscid blow-up, if it exists, would behave unlike shock formation in scalar conservation laws, where the linearized evolution remains uniformly stable up to the shock."],"supporting_citations":[{"why":"Supplies the vorticity blow-up criterion connecting growth of $\\|\\omega\\|_{L^\\infty}$ to loss of $H^s$ regularity, the step that turns Theorem 1 into Corollary 1.","marker":"[2]"},{"why":"Establishes well-posedness of the linearized Euler equation with $H^1$ initial data, which defines the semigroup and justifies the approximation argument in Proposition 2.","marker":"[9]"},{"why":"Introduces the bicharacteristic-amplitude WKB system for small oscillations of an ideal fluid, the core ODE machinery that converts nonlocal pressure into local amplitude dynamics.","marker":"[14]"},{"why":"Extends the vorticity blow-up criterion to bounded domains, covering the boundary case in Theorem 1 and Corollary 1.","marker":"[4]"},{"why":"Provides the elliptic regularity estimate for the pressure used in the $L^p$ energy estimate of Lemma 2.","marker":"[12]"}],"fun_headline_variants":["Blow-up forces linear instability in 3D Euler","3D Euler blow-up implies unbounded linear growth","Blow-up makes Euler linearly unstable in all Lp","Finite-time blow-up forces linear instability in Euler","Euler blow-up automatically makes linearized flow unstable"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument requires the flow, frequency, and amplitude of the WKB system to stay twice differentiable with respect to their initial data up to every $T<T^*$; the paper gets this from $s>9/2$, so a blow-up that first made $\\nabla\\nabla u$ discontinuous would fall outside the proof.","fun_headline_variants_meta":{"raw":{"variants":["Blow-up forces linear instability in 3D Euler","3D Euler blow-up implies unbounded linear growth","Blow-up makes Euler linearly unstable in all Lp","Finite-time blow-up forces linear instability in Euler","Euler blow-up automatically makes linearized flow unstable"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000503,"raw_usage":{"total_tokens":2399,"prompt_tokens":830,"completion_tokens":1569,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":446,"completion_tokens_details":{"reasoning_tokens":1491}},"tokens_in":446,"tokens_out":1569,"duration_ms":11340,"temperature":1.0,"reasoning_tokens":1491,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:06:57.999388+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for a proposed finite-time blow-up candidate of 3D Euler, the maximal $L^p$ growth $\\gamma_p(T)$ of the linearized equation on a sequence $T\\to T^*$. If at any such $T$ the product $\\gamma_p(T)^2\\|\\omega_0\\|_{L^\\infty}$ is strictly less than $\\|\\omega(T)\\|_{L^\\infty}$, the theorem's bound is violated; a rigorous contradiction would be a smooth solution that blows up in $H^s$ while the linearized semigroup stays bounded in some $L^p$, $1<p<\\infty$.","supporting_citations":[{"cited_title":"Inoue and T","cited_arxiv_id":null,"evidence_quote":"Establishes well-posedness of the linearized Euler equation with $H^1$ initial data, which defines the semigroup and justifies the approximation argument in Proposition 2."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the bicharacteristic-amplitude WKB system for small oscillations of an ideal fluid, the core ODE machinery that converts nonlocal pressure into local amplitude dynamics."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Extends the vorticity blow-up criterion to bounded domains, covering the boundary case in Theorem 1 and Corollary 1."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the elliptic regularity estimate for the pressure used in the $L^p$ energy estimate of Lemma 2."}],"review_version":1}