Pith. sign in

REVIEW 2 major objections 4 minor 32 references

On the sharpness of bounds on the rate of growth of Lebesgue norms of the velocity in Navier-Stokes flows

T0 review · 2 major / 4 minor · reviewed 2026-07-12 · grok-4.5

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

Pith's one-line read A classic upper bound on how fast Navier-Stokes velocity norms can grow is sharp up to a numerical constant.

desk verdict Solid numerical evidence that the Robinson–Sadowski Lq-growth bound is saturated in the exponent by concrete local maximizers; the local-vs-global caveat is real but does not erase the result. read the letter →

arxiv 2607.02739 v1 pith:VPGM6IBV submitted 2026-07-02 math.AP physics.flu-dyn

Fabian Bleitner , Bartosz Protas This is my paper · ORCID
classification math.APphysics.flu-dyn MSC 35Q3049M4165N3576D05
keywords Navier-StokesequationsLadyzhenskaya-Prodi-SerrinconditionsLebesguenormsaprioriboundssharpnessRiemannianconjugategradientsingularityformation
topics Navier-Stokes Regularity
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 asks whether a standard a priori estimate that controls the instantaneous growth of Lq norms of the velocity in three-dimensional Navier-Stokes flows can be improved. That estimate is intimately tied to the Ladyzhenskaya-Prodi-Serrin conditions that guarantee smoothness. By solving a constrained variational problem that maximises the growth rate for fixed Lq norm, the authors construct families of velocity fields whose growth realises the same power of the norm that appears on the right-hand side of the bound. The conclusion is that the bound cannot be strengthened in its scaling; only the prefactor might still be refined. The same computations also indicate that the growth rate becomes unbounded as the exponent q approaches the critical value 3.

What carries the argument

A Riemannian conjugate-gradient method on the Hilbert manifold of divergence-free, zero-mean fields with fixed Lq norm, maximising the objective functional Rq that expresses the instantaneous growth rate after elimination of pressure via the Poisson equation.

What would settle it

An independent maximisation of the same objective, started from a qualitatively different family of initial fields or run at substantially higher resolution, that produces a strictly slower growth rate whose scaling is weaker than the predicted power of the Lq norm.

Watch

Extended reading notes

Core claim

Numerical maximisers of the instantaneous rate of growth of the Lq norm of velocity, obtained for several q>3 and for norms spanning several orders of magnitude, saturate the power-law upper bound d/dt ||u||_q^q ≤ C ||u||_q^{q(q-1)/(q-3)} as the norm tends to infinity. The bound is therefore sharp up to a numerical prefactor and cannot be fundamentally improved.

Load-bearing premise

That the local maximisers found by continuing from the small-data ABC flow remain representative of the true global maximum growth rate at large norms, and are not artefacts of finite resolution or of the particular geometric operations used on the constraint manifold.

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.

Referee Report

2 major / 4 minor

Summary. The paper investigates the sharpness of the a priori bound (5) on the instantaneous growth rate of the L^q norm of velocity for 3D Navier-Stokes flows (q>3), which is closely tied to the Ladyzhenskaya-Prodi-Serrin conditions. An objective functional R_q(u) for (d/dt)||u||_q^q is derived (Eq. (15)), and maximizers subject to fixed ||u||_q=B, divergence-free and zero-mean constraints are sought via a Riemannian conjugate-gradient method on the Hilbert manifold X_B subset H^{3/2-1/q} (Problem 3). Branches of local maximizers are continued from the small-data ABC eigenfunction (Section 3, Algorithm 1). Numerical results for q=4,5,6,9 show that R_q(eu_B) saturates the power q(q-1)/(q-3) as B o∞ (Table 3, Figures 4–5), up to a numerical prefactor, while the q o3 limit appears ill-posed. The optimizers become increasingly localized and approach the edge of the ambient Sobolev space.

Significance. If the numerical evidence is accepted as representative, the work supplies concrete computational support that bound (5) cannot be improved in the exponent, complementing earlier variational studies of enstrophy growth (Problem 1) and completing the picture summarized in Table 1. The derivation of R_q, the small-data analysis, and the Riemannian CG formulation are clean and reusable. The observation that the same states do not simultaneously saturate the enstrophy bound, and that the q=3 problem appears ill-posed, are of independent interest for the conditional-regularity program. Strengths include transparent appendices (A–C), explicit power-law fits, and careful spectral monitoring.

major comments (2)
  1. The central claim that bound (5) “cannot be fundamentally improved” rests on local maximizers obtained exclusively by continuation from the ABC flow (Algorithm 1, Section 5). Problem 3 is non-convex; nothing rules out other branches (different topology or localization) that could produce a strictly larger growth rate or a higher exponent. The paper should either (i) attempt restarts from qualitatively different initial data at large B, or (ii) rephrase the claim as “sharp along the computed branches / lower bound on the true supremum.” Without this, the global-sharpness language in the abstract and Section 6 overreaches the evidence.
  2. Figures 4–5 and Table 3 report power-law saturation, yet the maximizers approach the edge of H^{3/2-1/q} (Fig. 2b) and become highly localized while resolution is capped at N=1024^{3}. No systematic resolution study or extrapolation of the fitted exponents eα versus N is provided. A short convergence check (e.g., recompute a large-B point at two resolutions and report the change in R_q and eα) is needed to confirm that the observed scaling is not an artifact of under-resolution or of the particular filter length ℓ=0.1.
minor comments (4)
  1. Table 1 and the abstract state that the bound is sharp “up to a numerical prefactor,” yet the measured prefactors eC in Table 3 are extremely small (10^{-15}–10^{-4}). A brief remark on the practical size of the constant would help readers assess how close the bound is to being saturated in absolute terms.
  2. Section 5.4 (q o3) reports diverging exponents with “relatively large uncertainties.” The fitting procedure and the range of B used for those fits should be stated more precisely so that the claimed divergence can be reproduced.
  3. Typographical slips: “formatioon” (keywords), “ealier” (Section 6), and occasional missing spaces around math operators. A light copy-edit pass would suffice.
  4. Figure 6 captions refer to “normalized” fields but do not specify the normalization; a short clarification would improve readability.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: independent analytic bound is compared post-hoc to unconstrained numerical maximizers of the exact growth functional.

full rationale

The a priori bound (5) is derived independently in Appendix A via Hölder/Young/Gagliardo-Nirenberg estimates applied to the exact expression (15) for (1/q) d/dt ||u||_q^q. The optimization Problems 2/3 maximize that same exact functional R_q(u) (obtained by testing the NSE with |u|^{q-2}u and eliminating pressure) subject only to the L^q-norm constraint, divergence-free and zero-mean conditions; the bound itself is never inserted into the objective, the Riemannian CG iteration, the retraction, or the continuation from the ABC eigenfunction. After the maximizers eu_B are obtained, their achieved R_q values are simply plotted against B and fitted to a power law whose exponent is then compared with the analytic exponent q(q-1)/(q-3). Self-citations (to [7,17,14,27,32] etc.) supply only the methodological template for the Riemannian CG solver and earlier enstrophy results; none of those citations is used to justify the sharpness claim for bound (5). The observed saturation is therefore an independent numerical observation, not a tautology. Minor residual self-citation of the authors' own prior optimization framework does not load-bear the central claim, yielding a score of 1.

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

The central claim rests on the classical Navier-Stokes equations, standard Sobolev embeddings, the Riesz representation theorem used to construct Hilbert-Sobolev gradients, and a handful of numerical hyperparameters that control the Riemannian optimizer and the continuation. No new physical entities are postulated; the free parameters are purely algorithmic.

free parameters (4)
  • Sobolev filter length ℓ = 0.1
    Set by hand to 0.1 to improve convergence of the Riemannian iterations (eq. 45, Table 2); changes the effective gradient but is not fitted to the target scaling.
  • constraint increment δB = 10^{1/8}-1
    Continuation step size 10^{1/8}-1 chosen for numerical stability (Algorithm 1, Table 2).
  • momentum reset interval = 25
    β_k reset to zero every 25 iterations to improve convergence; pure algorithmic choice.
  • prefactors ĒC in power-law fits = 2.9e-15 … 7.0e-4
    Least-squares constants reported in Table 3; they quantify the gap to the analytic C but are not used to claim sharpness of the exponent.
assumptions (4)
  • domain assumption Navier-Stokes equations on the unit torus with u=1 and periodic boundary conditions
    Stated in Section 1; all subsequent analysis and numerics are performed for this system.
  • standard math Sobolev embedding H^{3/2-1/q} o W^{1,3q/(q+1)} for q≥3
    Used to justify the Hilbert-space formulation (eq. 18) and the well-definedness of the objective functional.
  • standard math Riesz representation theorem in L^{2} and H^s allowing construction of Hilbert-Sobolev gradients from the L^{2} gradient
    Invoked in Section 4.2 to obtain abla_H R_q from abla_L R_q via Fourier multipliers.
  • ad hoc to paper Local maximizers obtained by Riemannian CG from the ABC initial guess capture the asymptotic scaling of the global supremum
    Implicit throughout Section 5; no global optimality certificate is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On the sharpness of bounds on the rate of growth of Lebesgue norms of the velocity in Navier-Stokes flows." pith.science (2026). https://pith.science/paper/VPGM6IBV

@misc{pith2026260702739,
  author       = {Pith},
  title        = {Pith review of: On the sharpness of bounds on the rate of growth of Lebesgue norms of the velocity in Navier-Stokes flows},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VPGM6IBV}},
  note         = {Machine review of arXiv:2607.02739}
}
abstract

In this paper we consider solutions $\boldsymbol{u}$ of the three-dimensional Navier-Stokes system and investigate sharpness of the a priori bound \begin{align*} \frac{d}{dt}\|\boldsymbol{u}\|_q^q \leq C\|\boldsymbol{u}\|_q^{q\frac{q-1}{q-3}}, \qquad q > 3. \end{align*} This bound is closely related to the Ladyzhenskaya-Prodi-Serrin conditions characterizing classical solutions of the Navier-Stokes system. Velocity fields maximizing the rate of growth $(d/dt)\|\boldsymbol{u}\|_q^q$ under certain constraints are found as solutions of a suitable optimization problem which is solved numerically using a Riemannian conjugate gradient approach. The results obtained for different $q$ and increasing values of $\|\boldsymbol{u}\|_q$ indicate that the bound is indeed sharp, up to a numerical prefactor, and therefore cannot be fundamentally improved. Additionally, the results also suggest that the rate of growth $(d/dt)\|\boldsymbol{u}\|_q^q$ diverges as $q\to 3$.

Figures

Figures reproduced from arXiv: 2607.02739 by the authors.

Figure 1
Figure 1. Schematic illustration of a single iteration of the Riemannian conjugate gradient [PITH_FULL_IMAGE:figures/full_fig_p021_1.png] view at source ↗
Figure 2
Figure 2. (a) The objective functional R5(uk) versus iterations k in the solution of Problem 3 with q = 5 and B = 109/4 ≈ 177.8. (b) The corresponding energy spectra E(s,uk) (see (48)) of the approximations uk of the maximizer ue B obtained at different iterations k; the slanting red line represents the relation Cs−28/5 for some C > 0 which is the asymptotic form (as |ξ| → ∞) of the energy spectrum for functions immediately o… view at source ↗
Figure 3
Figure 3. Time evolution of the L 5 norm of the solution of system (1)–(2) with the initial condition given by the optimal field ue B obtained by solving Problem 3 for q = 5 and B = 109/4 ≈ 177.8. of order q in u, dominates the pressure term, which is of order q + 1 in u. This observation is also consistent with our analysis of solutions to Problem 3 in the limit B → 0 in Section 3, cf. (27). On the other hand, for B > B¯ the… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Dependence of the maximum values of the objective functional [PITH_FULL_IMAGE:figures/full_fig_p029_4.png]
Figure 5
Figure 5. Figure 5: Dependence of the maximum values of the objective functional compensated [PITH_FULL_IMAGE:figures/full_fig_p030_5.png]
Figure 6
Figure 6. Figure 6: Normalized (a)–(e) velocity ue B (x) and (f)–(j) vorticity (∇ × ue B )(x) fields in the solutions of Problem 3 with q = 5 and indicated values of the constraint parameter. The different colors represent the Cartesian components of the fields: (red) x1, (blue) x2 and (g…
Figure 7
Figure 7. Figure 7: we show ue B obtained for q = 5 and B = 109/4 ≈ 177.8, which is the representative case already discussed in Subsection 5.1. In the figure we also plot the integrand expression of the objective functional (15) evaluated at ue B , i.e., r(ue B ) := −ν|ue B | q−2 |∇ue B …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 6 canonical work pages

  1. [1]

    Optimization Algorithms on Matrix Manifolds

    Absil, P.A., Mahony, R., Sepulchre, R., 2008. Optimization Algorithms on Matrix Manifolds. Princeton University Press

    2008

  2. [2]

    Sobolev Spaces

    Adams, R.A., Fournier, J.F., 2005. Sobolev Spaces. Elsevier, Amster- dam

    2005

  3. [3]

    Sharp bounds on enstrophy growth for viscous scalar conservation laws

    Albritton, D., De Nitti, N., 2023. Sharp bounds on enstrophy growth for viscous scalar conservation laws. Nonlinearity 36, 7142. doi:10.1088/ 1361-6544/ad073f

    2023

  4. [4]

    Maximum palinstrophy amplification in the two-dimensional navier–stokes equations

    Ayala, D., Doering, C.R., Simon, T.M., 2018. Maximum palinstrophy amplification in the two-dimensional navier–stokes equations. Journal of Fluid Mechanics 837, 839–857. doi:10.1017/jfm.2017.874

    doi:10.1017/jfm.2017.874 2018

  5. [5]

    On maximum enstrophy growth in a hy- drodynamic system

    Ayala, D., Protas, B., 2011. On maximum enstrophy growth in a hy- drodynamic system. Physica D: Nonlinear Phenomena 240, 1553–1563. doi:10.1016/j.physd.2011.07.003

    work page doi:10.1016/j.physd.2011.07.003 2011

  6. [6]

    Maximum palinstrophy growth in 2d in- compressible flows

    Ayala, D., Protas, B., 2014. Maximum palinstrophy growth in 2d in- compressible flows. Journal of Fluid Mechanics 742, 340–367. doi:10. 1017/jfm.2013.685

    2014

  7. [7]

    Extreme vortex states and the growth of enstrophy in three-dimensional incompressible flows

    Ayala, D., Protas, B., 2017. Extreme vortex states and the growth of enstrophy in three-dimensional incompressible flows. Journal of Fluid Mechanics 818, 772–806. doi:10.1017/jfm.2017.136

    work page doi:10.1017/jfm.2017.136 2017

  8. [8]

    Nonlinearity and Functional Analysis

    Berger, M.S., 1977. Nonlinearity and Functional Analysis. Academic Press

    1977

Show all 32 references
  1. [9]

    Spectral Methods

    Canuto, C., Hussaini, M.Y., Quarteroni, A., Zang, T.A., 2006. Spectral Methods. Springer. doi:10.1007/978-3-540-30726-6

    2006 doi

  2. [10]

    L3,∞-solutions of the Navier-Stokes equations and backward uniqueness

    Escauriaza, L., Seregin, G.A., Sverak, V., 2003. L3,∞-solutions of the Navier-Stokes equations and backward uniqueness. Russian Mathemat- ical Surveys 58, 211–250. doi:10.1070/rm2003v058n02abeh000609

    2003 doi

  3. [11]

    Existence and smoothness of the Navier–Stokes equation

    Fefferman, C.L., 2000. Existence and smoothness of the Navier–Stokes equation. URL:https://www.claymath.org/wp-content/uploads/ 2022/06/navierstokes.pdf. 39

    2000

  4. [12]

    Gevrey class regularity for the solutions of the navier-stokes equations

    Foias, C., Temam, R., 1989. Gevrey class regularity for the solutions of the navier-stokes equations. Journal of Functional Analysis 87, 359–369. doi:10.1016/0022-1236(89)90015-3

    1989 doi

  5. [13]

    Potentially singular behavior of the 3D Navier–Stokes equations

    Hou, T., 2022. Potentially singular behavior of the 3D Navier–Stokes equations. Foundations of Computational Mathematics 23, 2251–2299. doi:10.1007/s10208-022-09578-4

    2022 doi

  6. [14]

    Searching for singularities in navier–stokes flows based on the ladyzhenskaya–prodi–serrin conditions

    Kang, D., Protas, B., 2022. Searching for singularities in navier–stokes flows based on the ladyzhenskaya–prodi–serrin conditions. Journal of Nonlinear Science 32, 1–35. doi:10.1007/s00332-022-09832-7

    2022 doi

  7. [15]

    Maximum amplification of enstro- phy in three-dimensional Navier-Stokes flows

    Kang, D., Yun, D., Protas, B., 2020. Maximum amplification of enstro- phy in three-dimensional Navier-Stokes flows. Journal of Fluid Mechan- ics 893, A22. doi:10.1017/jfm.2020.204

    2020 doi

  8. [16]

    On the existence and unique- ness of the solution of the nonstationary problem for a viscous, incom- pressible fluid

    Ladyzhenskaya, O.A., Kiselev, A.A., 1957. On the existence and unique- ness of the solution of the nonstationary problem for a viscous, incom- pressible fluid. Izvestiya Akademii Nauk SSSR. Seriya Matematich- eskaya 21, 655–680

    1957

  9. [17]

    Limits on enstrophy growth for solutions of the three-dimensional Navier-Stokes equations

    Lu, L., Doering, C.R., 2008. Limits on enstrophy growth for solutions of the three-dimensional Navier-Stokes equations. Indiana University Mathematics Journal 57, 2693–2728. doi:10.1512/iumj.2008.57.3716

    2008 doi

  10. [18]

    Vorticity and Incompressible Flow

    Majda, A.J., Bertozzi, A.L., 2002. Vorticity and Incompressible Flow. Cambridge University Press, Cambridge

    2002

  11. [19]

    Sobolev Gradients and Differential Equations

    Neuberger, J.W., 2010. Sobolev Gradients and Differential Equations. 2 ed., Springer. doi:10.1007/978-3-642-04041-2

    2010 doi

  12. [20]

    Numerical Optimization

    Nocedal, J., Wright, S.J., 1999. Numerical Optimization. Springer, New York

    1999

  13. [21]

    NumericalOptimization

    Nocedal, J., Wright, S.J., 2006. NumericalOptimization. 2ed., Springer. doi:10.1007/978-0-387-40065-5

    2006 doi

  14. [22]

    Numerical Recipes

    Press, W.H., Teukolsky, S.A., Vetterling, W.T., Flannery, B.P., 2007. Numerical Recipes. 3 ed., Cambridge University Press. 40

    2007

  15. [23]

    Un teorema di unicità per le equazioni di navier-stokes

    Prodi, G., 1959. Un teorema di unicità per le equazioni di navier-stokes. Annali di Matematica Pura ed Applicata 48, 173–182. doi:10.1007/ BF02410664

    1959

  16. [24]

    Adjoint-based optimization of PDE systems with al- ternative gradients

    Protas, B., 2008. Adjoint-based optimization of PDE systems with al- ternative gradients. Journal of Computational Physics 227, 6490–6510. doi:10.1016/j.jcp.2008.03.013

    2008 doi

  17. [25]

    Systematic search for extreme and singular behaviour in some fundamental models of fluid mechanics

    Protas, B., 2022. Systematic search for extreme and singular behaviour in some fundamental models of fluid mechanics. Philosophical Transac- tions of the Royal Society A: Mathematical, Physical and Engineering Sciences 380, 20210035. doi:10.1098/rsta.2021.0035

    2022 doi

  18. [26]

    A computational framework for the regularization of adjoint analysis in multiscale pde systems

    Protas, B., Bewley, T.R., Hagen, G., 2004. A computational framework for the regularization of adjoint analysis in multiscale pde systems. Jour- nalofComputationalPhysics195, 49–89. doi:10.1016/j.jcp.2003.08. 031

    2004 doi

  19. [27]

    Systematic search for extreme behavior in 3D Navier- Stokes flows based on the Ladyzhenskaya-Prodi-Serrin conditions

    Ramírez, E., 2025. Systematic search for extreme behavior in 3D Navier- Stokes flows based on the Ladyzhenskaya-Prodi-Serrin conditions. Ph.D. thesis. URL:http://hdl.handle.net/11375/32607

    2025

  20. [28]

    The ladyzhenskaya-prodi-serrin condi- tions and the search for extreme behavior in 3d navier-stokes flows

    Ramírez, E., Protas, B., 2026. The ladyzhenskaya-prodi-serrin condi- tions and the search for extreme behavior in 3d navier-stokes flows. arXiv:2604.13338

    2026 arXiv

  21. [29]

    A local smoothness criterion for solutions of the 3d navier-stokes equations

    Robinson, J.C., Sadowski, W., 2014. A local smoothness criterion for solutions of the 3d navier-stokes equations. Rendiconti del Seminario Matematico della Università di Padova 131, 159–178. URL:http:// www.numdam.org/item/RSMUP_2014__131__159_0/

    2014

  22. [30]

    On the interior regularity of weak solutions of the navier-stokes equations

    Serrin, J., 1962. On the interior regularity of weak solutions of the navier-stokes equations. Archive for Rational Mechanics and Analysis 9, 187–195. doi:10.1007/BF00253344

    1962 doi

  23. [31]

    Quantitative bounds for critically bounded solutions to the navier-stokes equations.arXiv:1908.04958

    Tao, T., 2020. Quantitative bounds for critically bounded solutions to the navier-stokes equations.arXiv:1908.04958

    2020 arXiv

  24. [32]

    Systematic search for singularities in 3d euler flows

    Zhao, X., Protas, B., 2023. Systematic search for singularities in 3d euler flows. Journal of Nonlinear Science 33, 1–33. doi:10.1007/ s00332-023-09977-z. 41

    2023

Pith tools

Reviewed July 12, 2026 · model on record in the stance chip above.