Blow-up solutions to 3D Euler are hydrodynamically unstable

Alexis Vasseur; Misha Vishik

REVIEW 2 major objections 4 minor 14 references

Blow-up solutions to 3D Euler are hydrodynamically unstable

T0 review · 2 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper proves that finite-time blow-up in smooth 3D Euler forces linearized instability: for every $T<T^*$ and $1<p<\infty$, $\gamma_p(T)^2 \ge \|\omega(T)\|_{L^\infty}/\|\omega_0\|_{L^\infty}$.

desk verdict 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. read the letter →

arxiv 1908.05766 v1 pith:IUSYJYQN submitted 2019-08-15 math.AP

Alexis Vasseur , Misha Vishik This is my paper · ORCID

classification math.AP MSC 76B0335B3535B44

keywords Eulerequationsfinite-timeblow-uphydrodynamicinstabilitylinearstabilityvorticityWKBexpansionLpsemigroupincompressibleflow

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

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

What carries the argument

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.

What would settle it

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 $L^p$, $1<p<\infty$.

Watch

Extended reading notes

Core claim

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 $L^p$, $1<p<\infty$.

Load-bearing premise

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.

Editorial extensions

If this is right

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.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

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.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the 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.

What'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.

Where 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.

Also 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.

Citation 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.

Who 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.

Referee Report

2 major / 4 minor

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.

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 (2)

[Section 3, proof of Lemma 2] 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 3, proof of Proposition 2 (equations near (14))] 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.

minor comments (4)

[Section 3, proof of Proposition 2] 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.
[Lemma 2 statement] 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)).
[Throughout] 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.
[Lemma 1 proof] 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.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: Theorem 1 is derived from directly-proved WKB estimates and a self-contained error estimate; the only self-citation is non-load-bearing.

full rationale

The paper's central claim, Theorem 1, is obtained by combining Proposition 1 (vorticity growth bounded by the WKB amplitude growth beta(T)) and Proposition 2 (beta(T) bounded by the linearized semigroup growth gamma_p(T)). Proposition 1 is proved in Section 2: Lemma 1 establishes the bicharacteristic-amplitude ODE properties (existence, uniqueness, C^2 dependence on initial data, and the conserved quantities) directly from Cauchy-Lipschitz and determinant computations, without invoking the target instability statement. Proposition 2 is proved in Section 3 by explicitly constructing high-frequency test solutions v_{epsilon,delta} from the WKB data, verifying that they solve the linearized Euler equation up to O(epsilon), and then using Lemma 2's elliptic and Gronwall estimates to control the difference between the true linearized solution and the WKB approximation. The estimate in Lemma 2 relies on external results (Inoue-Miyakawa for well-posedness and Krylov for elliptic regularity) and on the assumed H^s regularity of u, not on the conclusion of Theorem 1. No parameter is fitted to the predicted quantity, and the proof never assumes gamma_p(T) is large or that blow-up occurs. The only self-citation is to Vishik [14] for the origin of the WKB expansion method, but the paper re-proves the required ODE and error estimates itself, so the citation is contextual rather than load-bearing. A potential mathematical issue in Lemma 2 concerning a boundary identity on curved domains would be a correctness defect, not a circularity defect, and under the circularity-only mandate it does not alter the score. Overall, the derivation chain is self-contained and non-circular; score 1 reflects only the minor, non-load-bearing self-citation.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The proof is self-contained modulo standard results: local well-posedness in H^s, the Beale-Kato-Majda criterion, Inoue-Miyakawa linear theory, and Krylov elliptic estimates. It introduces no free parameters and no new physical entities. The WKB framework is taken from prior work, but the needed identities are proved directly in Section 2.

assumptions (5)

standard math Local well-posedness for the 3D Euler equation in H^s, s>9/2, with Sobolev embedding giving ∇u and ∇∇u in C^1([0,T]×Ω).
Invoked in Lemma 1 and throughout to ensure flows, frequencies, and amplitudes are Lipschitz in time and C^2 in initial data.
standard math Beale-Kato-Majda criterion: if ∫_0^{T*} ||ω||_∞ dt is finite, the smooth solution extends beyond T*.
Used in Corollary 1 to convert unbounded H^s norm into divergence of the vorticity integral, which combined with Theorem 1 gives ∫γ_p^2 = ∞.
standard math Inoue-Miyakawa existence and uniqueness for the linearized Euler equation with H^1 initial data.
Justifies the definition of γ_p and the use of H^1 initial data in Lemma 2 and Proposition 2.
standard math Krylov elliptic regularity estimates for the pressure Poisson equation with Neumann boundary data.
Used in Lemma 2 to bound ∇P' in Lp in terms of u, v, and f.
domain assumption The domain Ω is either R3, T3, or a bounded smooth domain with impermeability, and the flow map γ_t maps Ω into itself.
The proof uses flow invariance of the domain and boundary to keep WKB localized perturbations compactly supported and to justify that γ_t is a diffeomorphism.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Blow-up solutions to 3D Euler are hydrodynamically unstable." pith.science (2026). https://pith.science/paper/IUSYJYQN

@misc{pith2026190805766,
  author       = {Pith},
  title        = {Pith review of: Blow-up solutions to 3D Euler are hydrodynamically unstable},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IUSYJYQN}},
  note         = {Machine review of arXiv:1908.05766}
}

read the original abstract

We study the interaction between the stability, and the propagation of regularity, for solutions to the incompressible 3D Euler equation. It is still unknown whether a solution with smooth initial data can develop a singularity in finite time. This article explains why the prediction of such a blow-up, via direct numerical experiments, is so difficult. It is described how, in such a scenario, the solution becomes unstable as time approaches the blow-up time.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

14 extracted references · 13 canonical work pages

[1]

V. Arnold. Sur la g´ eom´ etrie diﬀ´ erentielle des groupes de Lie de dimension inﬁnie et ses applications ` a l’hydrodynamique des ﬂuides parfaits. Ann. Inst. Fourier (Grenoble) , 16(fasc. 1):319–361, 1966

work page 1966
[2]

J. T. Beale, T. Kato, and A. Majda. Remarks on the breakdow n of smooth solutions for the 3-D Euler equations. Comm. Math. Phys. , 94(1):61–66, 1984

1984
[3]

Constantin

P. Constantin. Euler equations, Navier-Stokes equatio ns and turbulence. In Mathematical foundation of turbulent viscous ﬂows , volume 1871 of Lecture Notes in Math. , pages 1–43. Springer, Berlin, 2006

work page 2006
[4]

A. B. Ferrari. On the blow-up of solutions of the 3-D Euler equations in a bounded domain. Comm. Math. Phys. , 155(2):277–294, 1993

work page 1993
[5]

Friedlander, W

S. Friedlander, W. Strauss, and M. Vishik. Nonlinear ins tability in an ideal ﬂuid. Ann. Inst. H. Poincar´ e Anal. Non Lin´ eaire, 14(2):187–209, 1997

work page 1997
[6]

Friedlander and M

S. Friedlander and M. Vishik. Dynamo theory, vorticity g eneration, and exponential stretching. Chaos, 1(2):198–205, 1991

work page 1991
[7]

Friedlander and M

S. Friedlander and M. Vishik. Instability criteria for t he ﬂow of an inviscid incompressible ﬂuid. Phys. Rev. Lett., 66(17):2204–2206, 1991

work page 1991
[8]

Hou and R

Th. Hou and R. Li. Dynamic depletion of vortex stretching and non-blowup of the 3-D incompressible Euler equations. J. Nonlinear Sci. , 16(6):639–664, 2006

work page 2006

Show all 14 references

[9]

Inoue and T

A. Inoue and T. Miyakawa. On the existence of solutions fo r linearized Euler’s equation. Proc. Japan Acad. Ser. A Math. Sci. , 55(8):282–285, 1979

1979
[10]

R. Kerr. Bounds for Euler from vorticity moments and lin e divergence. J. Fluid Mech. , 729:R2, 13, 2013

2013
[11]

Kruzkhov

S.N. Kruzkhov. Generalized solutions of the cauchy pro blem in the large for nonlinear equations of ﬁrst order. Dokl. Akad. Nauk. SSSR , 187:29–32, 1969

1969
[12]

N. V. Krylov. Lectures on elliptic and parabolic equations in Sobolev spa ces, volume 96 of Graduate Studies in Mathematics . American Mathematical Society, Providence, RI, 2008

2008
[13]

Luo and Th

G. Luo and Th. Hou. Toward the ﬁnite-time blowup of the 3D axisymmetric Euler equations: a numerical investigation. Multiscale Model. Simul. , 12(4):1722–1776, 2014

2014
[14]

M. Vishik. Spectrum of small oscillations of an ideal ﬂu id and Lyapunov exponents. J. Math. Pures Appl. (9) , 75(6):531–557, 1996. 12 V ASSEUR AND VISHIK (Alexis F. Vasseur) Department of Mathematics, The University of Texas at Austin, Austin, TX 78712, USA E-mail address : v...

1996

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

V. Arnold. Sur la g´ eom´ etrie diﬀ´ erentielle des groupes de Lie de dimension inﬁnie et ses applications ` a l’hydrodynamique des ﬂuides parfaits. Ann. Inst. Fourier (Grenoble) , 16(fasc. 1):319–361, 1966

work page 1966

[2] [2]

J. T. Beale, T. Kato, and A. Majda. Remarks on the breakdow n of smooth solutions for the 3-D Euler equations. Comm. Math. Phys. , 94(1):61–66, 1984

1984

[3] [3]

Constantin

P. Constantin. Euler equations, Navier-Stokes equatio ns and turbulence. In Mathematical foundation of turbulent viscous ﬂows , volume 1871 of Lecture Notes in Math. , pages 1–43. Springer, Berlin, 2006

work page 2006

[4] [4]

A. B. Ferrari. On the blow-up of solutions of the 3-D Euler equations in a bounded domain. Comm. Math. Phys. , 155(2):277–294, 1993

work page 1993

[5] [5]

Friedlander, W

S. Friedlander, W. Strauss, and M. Vishik. Nonlinear ins tability in an ideal ﬂuid. Ann. Inst. H. Poincar´ e Anal. Non Lin´ eaire, 14(2):187–209, 1997

work page 1997

[6] [6]

Friedlander and M

S. Friedlander and M. Vishik. Dynamo theory, vorticity g eneration, and exponential stretching. Chaos, 1(2):198–205, 1991

work page 1991

[7] [7]

Friedlander and M

S. Friedlander and M. Vishik. Instability criteria for t he ﬂow of an inviscid incompressible ﬂuid. Phys. Rev. Lett., 66(17):2204–2206, 1991

work page 1991

[8] [8]

Hou and R

Th. Hou and R. Li. Dynamic depletion of vortex stretching and non-blowup of the 3-D incompressible Euler equations. J. Nonlinear Sci. , 16(6):639–664, 2006

work page 2006