Intermittent singular solutions of the stationary 2D Navier-Stokes equations in sharp Sobolev spaces

Ataleshvara Bhargava; Estepan Ashkarian; Matthew Novack; Nicholas Gismondi

REVIEW 2 major objections 4 minor 1 cited by

Intermittent singular solutions of the stationary 2D Navier-Stokes equations in sharp Sobolev spaces

T0 review · 2 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The stationary 2D Navier-Stokes equations on the torus admit nonzero weak solutions at every Sobolev regularity strictly below $L^2$.

desk verdict A genuinely new stationary-2D-Navier-Stokes construction that likely proves the claimed sharp regularity, but with one load-bearing estimate in the diagonal error term that is not justified as written and needs a routine fix. read the letter →

arxiv 2506.00841 v1 pith:XJ4IE2VI submitted 2025-06-01 math.AP

Estepan Ashkarian , Ataleshvara Bhargava , Nicholas Gismondi , Matthew Novack This is my paper

classification math.AP MSC 35Q3076D0535D30

keywords stationaryNavier-Stokes2DtorusweakparaproductsolutionsintermittentMikadoflowssharpSobolevregularityNashiterationL^{2-epsilon}convexintegration

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

On the two-dimensional torus, smooth stationary solutions of the incompressible Navier-Stokes equations must vanish, and the same rigidity was already known for solutions in the Lorentz space $L^{2,1}$. This paper establishes the opposite behavior at every regularity index strictly below $L^2$: there exist nonzero, non-constant stationary weak solutions in $\bigcap_{\epsilon\in(0,1)} L^{2-\epsilon}(\mathbb{T}^2)\cap \dot{H}^{-\epsilon}(\mathbb{T}^2)$. Because the velocity field is not square integrable, the paper defines the product $u\otimes u$ through an absolutely summable paraproduct in a negative Sobolev space. The proof is a Nash iteration using intermittent Mikado flows, and it shows that stationary Navier-Stokes solutions become flexible precisely where the energy identity loses its force. If correct, the result is essentially sharp, leaving only the endpoint $u\in L^2$ open.

What carries the argument

The central object is the intermittent Mikado flow $W_{q+1}^{k\perp}$: a divergence-free, mean-zero oscillation in the direction $k^\perp$, built from a periodic profile compressed into $\lambda_{q+1}^{\epsilon_\gamma}$ thin parallelograms of width $\lambda_{q+1}^{-1}$. Its $L^p$ norm scales like $\lambda_{q+1}^{(\epsilon_\gamma-1)(1/p-1/2)}$, and it is used to produce velocity increments with a large frequency gap of order $\lambda_{q+1}^{6}$ between the principal and corrector components. The paraproduct definition of $u\otimes u$ (Definition 1.1), the geometric reconstruction of symmetric tensors (Lemma 2.5), and the super-exponential separation of frequency shells $\lambda_{q+1}>2^{100}\lambda_q$ together carry the induction.

What would settle it

Compute directly the $L^2$ norm of $P_{>\lambda/2}W_{q+1}^{k\perp}$ for the Mikado flow of Lemma 4.1. If this norm fails to decay as $\lambda\to\infty$, then estimates (4.28)-(4.29) and hence item 3 of the induction collapse, and the proof needs a different diagonal argument. A Fourier-side calculation on the explicit $\rho_{q+1}$ profile would settle it.

Watch

Extended reading notes

Core claim

The paper's central claim is Theorem 1.3: for every $\epsilon\in(0,1)$ there is a nontrivial, mean-zero weak solution $u$ of the stationary 2D Navier-Stokes equations on $\mathbb{T}^2$ that belongs to $L^{2-\epsilon}$ and to $\dot{H}^{-\epsilon}$, with the product $u\otimes u$ interpreted through an absolutely convergent paraproduct sum. The constructed solution is a limit $u=\sum_q w_q$ of smooth iterates, where each increment $w_{q+1}$ is a high-frequency oscillation built from intermittent Mikado flows at frequency scale $\lambda_{q+1}^{6}$. The Reynolds stress error $R_q$ is driven to zero only in $\dot{H}^{-2}$, and the convergence of $u_q\otimes u_q$ is justified by the paraproduct definition rather than by $L^2$ summability. The paper also notes that any smooth divergence-free vector field can be approximated to within $\epsilon$ in $\dot{H}^{-\epsilon}\cap L^{2-\epsilon}$ by such solutions.

Load-bearing premise

The induction's Reynolds-stress bound rests on the unproved assertion that the high-frequency projection $P_{>\lambda/2}W_{q+1}^{k\perp}$ of each Mikado flow tends to zero in $L^2$ as $\lambda\to\infty$; the paper proves only the weaker $L^1$ decay of $P_{>\lambda/2}(W\otimes W)$.

Editorial extensions

If this is right

Mean-zero stationary solutions in $L^{2,1}(\mathbb{T}^2)$ are trivial, so the new solutions occupy exactly the largest $L^p$ scale that can support nontriviality: every exponent $2-\epsilon$, but not $L^2$.
The approximating sequence $u_q$ converges in $L^{2-\epsilon}\cap\dot{H}^{-\epsilon}$ for every $\epsilon\in(0,1)$, and the paraproduct sum representing $u\otimes u$ is absolutely convergent in $\dot{H}^{-2}$.
By Remark 1.4, the construction perturbs any prescribed smooth divergence-free vector field into a nontrivial stationary solution within $\epsilon$ in $\dot{H}^{-\epsilon}\cap L^{2-\epsilon}$.
The solutions lie in $\bigcap_{\epsilon>0} B^{-1/2-\epsilon}_{\infty,\infty}$, so their Besov regularity is fixed at $-1/2-\epsilon$ even though their $L^p$ integrability ranges over all $p<2$.
The Reynolds stress $R_q$ tends to zero in $\dot{H}^{-2}$ but not necessarily in $L^1$; this weaker convergence is what allows the diagonal terms of $u\otimes u$ to be controlled.

Reading between the lines

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

A natural test of the construction is to compute directly the $L^2$ norm of $P_{>\lambda/2}W_{q+1}^{k\perp}$ for the Mikado flows in Lemma 4.1; if this quantity does not decay, the diagonal Reynolds-stress estimate could instead be attempted through the $L^1$ decay of $P_{>\lambda/2}(W\otimes W)$ stated in the lemma.
The method is torus-specific because both the paraproduct and the Mikado flows rely on periodicity and on the absence of boundaries; a version on a bounded domain would require new boundary conditions for the renormalized product.
The huge frequency gap suggests the same iteration could work in settings where the Reynolds stress is controlled only in negative norms, for instance forced stationary flows or higher-dimensional analogues, although the intermittency exponents would need to be rebalanced.
One numerical check would be to truncate the construction to finitely many shells and compare the partial sums' Fourier energy decay with the rates predicted by membership in $L^{2-\epsilon}$ and $B^{-1/2-\epsilon}_{\infty,\infty}$.

Share X Bluesky LinkedIn Reddit HN

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

Colleague —

Here’s my read of Ashkarian–Bhargava–Gismondi–Novack, the stationary 2D Navier–Stokes intermittency paper.

What’s real: the paper constructs nontrivial stationary solutions in the intersection over ε of L^{2−ε} ∩ H^{−ε}, a genuine extension of Lemarié-Rieusset’s H^{-1} ∩ BMO^{-1} result. The new mechanism is the use of intermittent Mikado flows with a very high-frequency modulation (λ^6), which makes successive iterates decorrelated; the proof of convergence of u_q⊗u_q to the paraproduct is inductive and does not rely on any hidden fitting. The exposition is clear, and the frequency-support argument in §4.5 is clean.

The soft spot is in §4.4, the diagonal error estimate. In (4.28) they reduce ||w^{(p,2)}⊗w^{(p,3)}|| to a sum over ||P_{>λ²}W^{k⊥}||_{L²}, then cite Lemma 4.1(7) to get smallness. Lemma 4.1(7) is an L¹ statement about P_{>λ²}(W⊗W), not an L² statement about P_{>λ²}W. Those are different objects; the cited item does not imply the bound they need. Since (4.29) feeds into (4.30), and (4.30) gives the Reynolds-stress decay in item 3 and the summability in item 5, the induction as written rests on an unproven estimate. That said, I agree with the stress-test note: this is very likely fixable with a Schwartz-tail argument, because W has frequency scale λ, so P_{>λ²}W decays faster than any power of λ. The fix is routine, but it has to be written down.

Two smaller things. Lemma 2.7 is only outlined; the last term of (2.3) needs a sentence or two on why L¹ embeds into H̋^{-2} and the assumption handles it. Minor. And the paraproduct notion of solution is an interpretive choice, but it is explicit and the convergence of the product is proved, so I don’t see circularity.

Verdict: the paper deserves a serious referee. I would send it out with a request that the authors supply the missing L² estimate and expand Lemma 2.7. If they do that, the result stands. I’d bring it to reading group, and I’d cite it once the gap is patched.

Referee Report

2 major / 4 minor

Summary. The paper constructs nontrivial mean-zero weak solutions to the stationary 2D Navier-Stokes equations on the torus, belonging to the intersection of L^{2-epsilon} and dot-H^{-epsilon} for every epsilon in (0,1), with the nonlinear term interpreted through an absolutely summable paraproduct (Definitions 1.1 and 1.2). The proof is a Nash iteration: starting from smooth solutions of the Euler-Reynolds system, each velocity increment is built from intermittent Mikado flows (Lemma 4.1 and Definition 4.2), and Proposition 3.1 establishes the inductive estimates on L^p convergence, frequency separation, Reynolds stress decay in dot-H^{-2}, and summability of diagonal and off-diagonal products. Theorem 1.3 is then obtained by passing to the limit in the weak formulation. The paper also includes a remark that the endpoint case u in L^2 is not treated and would require a different method.

Significance. If correct, the result is a sharp extension of Lemarié-Rieusset's construction in dot-H^{-1} cap BMO^{-1}: it gives solutions at every regularity index strictly below L^2, with an explicit paraproduct interpretation of the nonlinear term. The proof is self-contained, the induction is carefully organized, and there is no fitting of parameters: the constants C and lambda_{q+1} are chosen to satisfy explicit inequalities. The authors also state honestly what their method does not reach, namely the L^2 endpoint. The main technical gap identified below is local and can be repaired within the manuscript's scope.

major comments (2)

[§4.4, Eq. (4.28)–(4.29)] The estimate for w^{(p,2)} otimes w^{(p,3)} is reduced to a sum of terms of the form ||P_{>lambda^2}(W^{k⊥})||_{L^2} and then concluded by citing Lemma 4.1 item 7. Item 7, however, only asserts the L^1 decay of P_{>lambda^2}(W^{k⊥} otimes W^{k⊥}); it does not by itself imply the L^2 decay of P_{>lambda^2}(W^{k⊥}) that is used in (4.28). This step is load-bearing because (4.29) is used to obtain (4.30), which controls the decay of the Reynolds stress in item 3 and thus the convergence of u_q otimes u_q in the proof of Theorem 1.3. The desired estimate is very likely true and can be obtained from Lemma 4.1 item 5 with one derivative, or from the compact frequency support implicit in the construction of W; the authors should supply this argument at the cited spot rather than referring to item 7.
[§2, Lemma 2.7] The proof of Lemma 2.7 is only outlined, and the estimate for the second term in the decomposition (2.3) is asserted without the displayed inequality. Since this lemma is used for the oscillatory term (4.23), a complete proof should be included. In addition, the hypotheses on beta_lambda (the uniform L^1 bound and the limit ||P_{>lambda^2} beta_lambda||_{L^1} -> 0) should be verified explicitly in the application beta_lambda = W^{k⊥} otimes W^{k⊥}; this verification is not written in the paper.

minor comments (4)

[§2, Lemma 2.6] In the estimate of the low-frequency term, the Fourier support is described as |k| gtrsim lambda^2, but from the displayed decomposition the relevant lower cutoff is |k| gtrsim lambda; the convergence argument is unchanged, but the displayed cutoff should be corrected.
[§4.1, Lemma 4.1] The construction of W^{k⊥} would be clearer if it stated explicitly that the Fourier support of W^{k⊥} is contained in an annulus of radius comparable to lambda_{q+1}; this would make item 7 immediate and would also clarify the estimate in (4.28).
[§4.4, Eq. (4.24)] The cross-term estimate is attributed to [4, Lemma 5.2] and [5, Remark 5.4] in a footnote, but the text should state the precise support-size and orthogonality argument being used, since these properties are not listed in Lemma 4.1.
[§4.6, Eq. (4.36)] The equality between ||w_{m+1} otimes w_{m+1}||_{dot-H^{-2}} and ||P_{neq 0}(w_{m+1} otimes w_{m+1})||_{dot-H^{-2}} is used without comment; it holds because w_{m+1} is mean-zero, but this should be stated.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the construction is a self-contained Nash iteration with no fitted parameters, and the paraproduct weak-solution convention is an interpretive definition rather than an input that forces existence.

full rationale

This paper constructs solutions by an explicit Nash iteration. No quantity is fitted to the target: \lambda_{q+1} is merely required to be a sufficiently large power of 2, and all estimates are convergence statements with constants independent of \lambda. The paraproduct notion of solution (Definitions 1.1-1.2) is a convention for interpreting u\otimes u for u outside L^2; it does not assert existence and therefore is not a self-definitional victory. The proof establishes convergence of the paraproduct series via item 5 of the induction, not by assuming the conclusion. The sharpness comparison to Lemari\'e-Rieusset is external context. The only citation involving a current author is [5] for the support-size bound at (4.24), but the same bound is also cited to the independent paper [4] and is a technical, checkable estimate rather than the central claim. The manuscript's (4.28)-(4.29) step may be missing justification (deducing L2 smallness of P_{>\lambda^2}W from Lemma 4.1(7)'s L1 smallness of P_{>\lambda^2}(W\otimes W)), but that is a proof gap, not an equation reducing to its own input, and so is not circularity under the stated rules.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

No new particles, forces, dimensions, or physical entities are introduced. The only non-classical object is the paraproduct definition of the product, which is a mathematical convention rather than an invented entity.

free parameters (4)

Initial amplitude A = sufficiently small positive constant
Chosen so that the base velocity u0 and Reynolds stress R0 satisfy item 3; the value is arbitrary and not fitted to external data.
Inductive constant C = sufficiently large positive constant
Chosen so that the lower bound in item 2 and the upper bounds in item 5 hold; it is a bookkeeping parameter.
Intermittency exponent epsilon_gamma = any number in (0,1) with lambda^{epsilon_gamma} integer and epsilon_gamma below the reconstruction threshold
Controls the concentration of the Mikado flows; it is a design parameter, not determined by data.
Superlarge frequency lambda_{q+1} = large power of 2 satisfying constraints (4.9)-(4.35)
Chosen at each induction step to make all error terms small; the existence of such a choice is part of the induction claim.

assumptions (4)

standard math Littlewood-Paley theory, Bernstein inequalities, Sobolev embedding and duality on the torus.
Used throughout Sections 2 and 3; quoted from [3,18].
domain assumption Definition 1.2 is the correct notion of weak stationary solution when u tensor u is not locally integrable.
The result is stated relative to this definition; an external notion of solution could change the theorem's meaning.
standard math Reconstruction of symmetric tensors (Lemma 2.5) and existence of intermittent Mikado flows (Lemma 4.1).
Taken from [9,15,8] and adapted; assumed as background for the iteration.
ad hoc to paper The paraproduct product P_{not 0}(fg) is defined by absolute summability of dyadic products (Definition 1.1).
This is the central reinterpretation used to make sense of u tensor u; it is introduced by the authors.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Intermittent singular solutions of the stationary 2D Navier-Stokes equations in sharp Sobolev spaces." pith.science (2026). https://pith.science/paper/XJ4IE2VI

@misc{pith2026250600841,
  author       = {Pith},
  title        = {Pith review of: Intermittent singular solutions of the stationary 2D Navier-Stokes equations in sharp Sobolev spaces},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XJ4IE2VI}},
  note         = {Machine review of arXiv:2506.00841}
}

abstract

In this paper we construct non-trivial solutions to the stationary Navier-Stokes equations on the two dimensional torus which lie in $\bigcap_{\epsilon \in (0,1)} L^{2-\epsilon}(\mathbb{T}^2) \cap \dot H^{-\epsilon}(\mathbb{T}^2)$. Due to the fact that our solutions are not square integrable, we must redefine the notion of solution. Our result gives a sharp extension of recent work of Lemari\'e-Rieusset, who proved a similar result in the space $\dot{H}^{-1} \cap {BMO}^{-1}$. The main new ingredient is the incorporation of intermittency into the construction of the solutions.

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

On non-uniqueness of mild solutions and stationary singular solutions to the Navier-Stokes equations
math.AP 2026-03 conditional novelty 8.0 of 10

Unconditional uniqueness of mild Navier-Stokes solutions fails in every Besov space with negative regularity index, via non-L^2 stationary singular solutions built by convex integration.

Reference graph

Works this paper leans on

22 extracted references · 21 canonical work pages · cited by 1 Pith paper

[1]

Albritton, E

D. Albritton, E. Brué, and M. Colombo. Non-uniqueness of Leray solutions of the forced Navier-Stokes equations.Annals of Mathematics, 196:1, 2022

work page 2022
[2]

Albritton and M

D. Albritton and M. Colombo. Non-uniqueness of Leray Solutions to the Hypodissipative Navier–Stokes Equations in Two Dimensions.Communications in Mathematical Physics, 402, 2023. 17

work page 2023
[3]

Bahouri, J.-Y

H. Bahouri, J.-Y. Chemin, and R. Danchin.Fourier analysis and nonlinear partial differen- tial equations. Grundlehren der mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences], volume 343. Springer, Heidelberg, 2011

work page 2011
[4]

Beekie, T

R. Beekie, T. Buckmaster, V. Vicol. Weak Solutions of Ideal MHD Which Do Not Conserve Magnetic Helicity Ann. of PDE, 6, 1, (2020)

work page 2020
[5]

Beekie, M

R. Beekie, M. Novack. Non-conservative solutions of the Euler-α equations J. Math. Fluid Mech., 25, 22 (2023)

work page 2023
[6]

Buckmaster, M

T. Buckmaster, M. Colombo, V. Vicol. Wild solutions of the Navier–Stokes equations whose singular sets in time have Hausdorff dimension strictly less than 1. JEMS, 24, no. 9, pg. 3333–3378

work page
[7]

Buckmaster and V

T. Buckmaster and V. Vicol. Nonuniqueness of weak solutions to the Navier-Stokes equation. Ann. of Math., 189(1):101–144, 2019

work page 2019
[8]

Buckmaster and V

T. Buckmaster and V. Vicol. Convex integration and phenomenologies in turbulenceEMS Surveys in Mathematical Sciences., Volume 6, Pages 173-263, July 2020

work page 2020

Show all 22 references

[9]

Shkoller, and V

T.Buckmaster, S. Shkoller, and V. Vicol. Nonuniqueness of weak solutions to the SQG equa- tion. Communications in Pure and Applied Mathematics, Volume 72, Issue 9, September 2019

2019
[10]

Cheskidov and X

A. Cheskidov and X. Luo. Sharp non-uniqueness for the Navier-Stokes equations. Invent. Math., 229, pg. 987–1054 (2022)

2022
[11]

Cheskidov and X

A. Cheskidov and X. Luo. Critical Nonuniqueness for the 2D Navier-Stokes Equations.Ann. PDE, 9:13 (2023)

2023
[12]

M. Christ. Nonuniqueness of weak solutions of the nonlinear Schrödinger equation. Preprint available at https://arxiv.org/pdf/math/0503366

arXiv
[13]

M. Christ. Nonuniqueness of generalized solutions of the Navier-Stokes equations. Preprint available https://math.berkeley.edu/ mchrist/Papers/stokes.pdf
[14]

Non-UniquenessofSmoothSolutionsoftheNavier-StokesEquations from Critical Data.arXiv prepint

M.Coiculescu, S.Palasek. Non-UniquenessofSmoothSolutionsoftheNavier-StokesEquations from Critical Data.arXiv prepint
[15]

Daneri and L

S. Daneri and L. Székelyhidi. Non-uniqueness and h-Principle for Hölder-Continuous Weak Solutions of the Euler EquationsArchive for Rational Mechanics and Analysis., Volume 224, Issue 2, Pages 471-514, May 2017

2017
[16]

De Lellis and L

C. De Lellis and L. Székelyhidi, Jr. The Euler equations as a differential inclusion.Ann. of Math., (2), 170 (2009), no. 3, 1417-1436

2009
[17]

De Lellis and L

C. De Lellis and L. Székelyhidi. Dissipative continuous Euler flows.Invent. Math., 193(2):377– 407, 2013

2013
[18]

Grafakos

L. Grafakos. Modern Fourier Analysis. Graduate Texts in Mathematics, Edition 2. Springer, New York, 2009

2009
[19]

P. G. Lemarié-Rieusset. Highly singular (frequentially sparse) steady solutions for the 2D Navier–Stokes equations on the torus.Journal of Functional Analysis, Volume 288, Issue 4, February 2025. 18

2025
[20]

X. Luo. Stationary solutions and nonuniqueness of weak solutions for the Navier-Stokes equa- tions in high dimensions.Arch. Rat. Mech. Anal., 233(2):701-747, 2019

2019
[21]

Scheffer

V. Scheffer. An inviscid flow with compact support in space-time.J. Geom. Anal., 3(4):343– 401, 1993

1993
[22]

Shnirelman

A. Shnirelman. Weak solutions with decreasing energy of incompressible Euler equations. Comm. Math. Phys., 210(3):541–603, 2000. Department of Mathematics, Purdue University, West Lafayette, IN, USA. Email address: bharga37@purdue.edu. Email address: eashkari@purdue.edu. Email...

2000

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Albritton, E

D. Albritton, E. Brué, and M. Colombo. Non-uniqueness of Leray solutions of the forced Navier-Stokes equations.Annals of Mathematics, 196:1, 2022

work page 2022

[2] [2]

Albritton and M

D. Albritton and M. Colombo. Non-uniqueness of Leray Solutions to the Hypodissipative Navier–Stokes Equations in Two Dimensions.Communications in Mathematical Physics, 402, 2023. 17

work page 2023

[3] [3]

Bahouri, J.-Y

H. Bahouri, J.-Y. Chemin, and R. Danchin.Fourier analysis and nonlinear partial differen- tial equations. Grundlehren der mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences], volume 343. Springer, Heidelberg, 2011

work page 2011

[4] [4]

Beekie, T

R. Beekie, T. Buckmaster, V. Vicol. Weak Solutions of Ideal MHD Which Do Not Conserve Magnetic Helicity Ann. of PDE, 6, 1, (2020)

work page 2020

[5] [5]

Beekie, M

R. Beekie, M. Novack. Non-conservative solutions of the Euler-α equations J. Math. Fluid Mech., 25, 22 (2023)

work page 2023

[6] [6]

Buckmaster, M

T. Buckmaster, M. Colombo, V. Vicol. Wild solutions of the Navier–Stokes equations whose singular sets in time have Hausdorff dimension strictly less than 1. JEMS, 24, no. 9, pg. 3333–3378

work page

[7] [7]

Buckmaster and V

T. Buckmaster and V. Vicol. Nonuniqueness of weak solutions to the Navier-Stokes equation. Ann. of Math., 189(1):101–144, 2019

work page 2019

[8] [8]

Buckmaster and V

T. Buckmaster and V. Vicol. Convex integration and phenomenologies in turbulenceEMS Surveys in Mathematical Sciences., Volume 6, Pages 173-263, July 2020

work page 2020