Pith. sign in

REVIEW 3 major objections 4 minor 30 references

Computability of the Solutions to Navier-Stokes Equations via Recursive Approximation

T0 review · 3 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 For the 2D incompressible Navier-Stokes equation in a square, a strong local solution is uniformly computable from the initial velocity and forcing.

desk verdict The homogeneous nonlinear computability result is a real contribution, but the full Theorem 1 is not proven as written: Section 5.5 hand-waves the inhomogeneous forcing and the pressure. read the letter →

arxiv 1908.01226 v1 pith:ZJ2KX4GU submitted 2019-08-03 math.AP

Shu-Ming Sun , Ning Zhong , Martin Ziegler This is my paper · ORCID
classification math.AP MSC 03D8035Q3047D06
keywords Navier-StokesequationscomputableanalysisCauchyrepresentationStokesoperatorfixed-pointiterationfractionalSobolevspacesdivergence-freevectorfieldsuniformcomputability
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 proves that the 2D incompressible Navier-Stokes initial value problem, on the open square with Dirichlet boundary conditions, is computable in the local-time sense: from any square-integrable divergence-free initial velocity and any continuous forcing, a Turing machine can output the velocity and pressure at time t to any requested precision, for all t up to a computable positive time T(a,f). The computation is uniform in the data, meaning one algorithm works for all inputs as long as a name for the input is supplied. This addresses a long-standing open problem in computable analysis about whether classical nonlinear PDEs admit recursive solutions, and it does so by combining classical fixed-point existence proofs with explicit effective estimates. If correct, the result closes the question for this equation in the two-dimensional local setting and supplies certified error bounds for numerical approximation.

What carries the argument

The argument is carried by the mild (integral) formulation of the equation, $u(t)=e^{-tA}a+\int_0^t e^{-(t-s)A}g(s)\,ds-\int_0^t e^{-(t-s)A}Bu(s)\,ds$, with $A$ the Stokes operator and $Bu=\mathcal P(u\cdot\nabla)u$ the projected nonlinearity. The Helmholtz projection $\mathcal P$ removes the pressure and turns the system into an evolution equation; the Stokes semigroup $e^{-tA}$ is the linear propagator. The proof constructs a dense set of effective names for divergence-free vector fields from trimmed and mollified polynomials, shows the projection and semigroup are computable on those names, and then proves that the fixed-point iteration $v_0(t)=e^{-tA}a+\int_0^t e^{-(t-s)A}g(s)\,ds$, $v_{n+1}(t)=v_0(t)-\int_0^t e^{-(t-s)A}Bv_n(s)\,ds$ converges effectively. The convergence rate is controlled by explicit $\beta$-function bounds on fractional powers of $A$, giving $\|u_{m+1}(t)-u_m(t)\|_2 \le L\epsilon^{m-1}$ with $\epsilon<1$ and $T(a)$ computable.

What would settle it

Directly compute the constants $C$, $M$, and $C_\alpha$ for the square domain from their definitions in the cited classical sources and check whether they are rational or recursive reals with explicit bounds; alternatively, run the iteration on a simple divergence-free polynomial initial datum and see whether the $\beta$-function recurrence actually produces a valid $T(a)$. If any of those constants is not effectively computable, the claimed uniform computability fails.

Watch

Extended reading notes

Core claim

The central claim, Theorem 1, is that the solution operator of the Navier-Stokes initial value problem is effectively approximable. For every divergence-free initial velocity $a \in L^\sigma_{2,0}(\Omega)$ and every forcing $f \in C([0,\infty), L^\sigma_{2,0}(\Omega))$, there exists a computable positive time $T(a,f)$ such that the problem has a strong local solution $(u,P)$ on $[0,T(a,f)]$, and the map $(a,f,t) \mapsto (u,P)(t)$ is computable with respect to the natural Cauchy representations of these spaces. In particular, the unbounded Stokes operator need not itself be encoded: the solution semigroup is computed through its integral representation and effective convergence of the fixed-point iteration. The paper thus extends the known computability of linear evolution equations to a nonlinear equation whose global regularity is open.

Load-bearing premise

The proof assumes that the constants $C$, $M$, and $C_\alpha$ appearing in the classical Stokes-operator estimates are computable real numbers; their computability is deferred to a forthcoming paper, and the uniform algorithm's time bound and convergence certificate depend on them.

Editorial extensions

If this is right

  • The local solution operator of the 2D Navier-Stokes equation is a computable map in the representation-theoretic sense, for both velocity and pressure.
  • The time of existence $T(a,f)$ is computable from the initial data and forcing, so an algorithm can decide how long its own approximation is certified.
  • Effective convergence of the iteration yields explicit, machine-checkable error bounds for the numerical solution, not merely asymptotic convergence.
  • The result resolves, in the local 2D homogeneous and inhomogeneous case, the open problem of whether a classical nonlinear PDE admits recursive solutions in the sense of computable analysis.

Reading between the lines

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

  • The proof's restriction to dimension 2 and to times near 0 likely reflects the Sobolev embedding used to make the nonlinear term computable; whether the same uniform computability holds globally in time or in 3D is not addressed by this paper.
  • If the missing proof of computability of the constants $C$, $M$, and $C_\alpha$ fails, the theorem would still hold for individual data but the uniform-in-data algorithm would break, so the effective content of the theorem hinges on that deferred proof.
  • A natural testable extension is to implement the iteration for polynomial initial data and compare the certified $T(a)$ and convergence rate with the beta-function bounds; the paper does not report numerical experiments.
  • Because the pressure is recovered from the same computable velocity via a path integral, downstream quantities such as boundary fluxes may be computed with the same representation, though the paper does not pursue this.
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.

Referee Report

3 major / 4 minor

Summary. The paper develops a Type-2 Theory of Effectivity (TTE) framework for the two-dimensional incompressible Navier-Stokes initial value problem on the square Ω=(-1,1)^2. It constructs a computable representation of the divergence-free space Lσ_{2,0}(Ω), proves that the Helmholtz projection and the Stokes semigroup are computable, and then gives an effective fixed-point iteration for the nonlinear problem. The main theorem, Theorem 1, claims that the local strong solution (u,P) is uniformly computable from the initial datum a and forcing f, with a computable existence time T(a,f). The homogeneous nonlinear case (g≡0) is treated in detail in Sections 5.3 and 5.4; the inhomogeneous case and the pressure are addressed only in the final Subsection 5.5 by a brief assertion.

Significance. If Theorem 1 were fully established, the result would be a significant contribution to computable analysis: it would give a positive answer, in the local two-dimensional setting, to the Pour-El and Richards question about recursive treatment of the Navier-Stokes equation, and it would do so with a uniform, data-to-solution effective approximation scheme. The paper contains real and valuable technical work: an explicit dense set of smooth divergence-free polynomial codes, a computable Helmholtz projection via trigonometric series with effective truncation, a computable treatment of the Stokes semigroup using contour integrals, and a detailed effective convergence analysis of the nonlinear iteration in the homogeneous case. These parts are carefully developed and go substantially beyond a mere existence argument. The main reservations concern the gap between the homogeneous theorem, which is proven in depth, and the full statement of Theorem 1, which adds inhomogeneous forcing and pressure recovery with only a one-paragraph argument.

major comments (3)
  1. [§5.5 (Theorem 1; Eq. (5))] The proof of Theorem 1 in the inhomogeneous case is asserted rather than proved. Propositions 4 and 5 and Claims 1–4 are all stated and proved for g≡0, with u0(t)=e^{-tA}a and with T(a) computed from a alone. For a general forcing g∈C([0,∞),Lσ_{2,0}(Ω)), the base iterate is u0(t)=e^{-tA}a+∫_0^t e^{-(t-s)A}g(s)ds; the paper gives no analogue of Claim 1 for this term, no effective bounds on the forcing contribution, and no argument that a δ_{Lσ}-name of a together with a [ρ→δ_{Lσ}]-name of f yields a computable T(a,f). The sentence in §5.5, 'Similarly to (the proofs of) Propositions 5, 4, and [24, Lemma 3.7], this solution is seen to be computable,' states the desired conclusion rather than supplying the missing estimates. Since the inhomogeneous forcing and the dependence of T on f are explicit parts of Theorem 1, this is a load-bearing gap.
  2. [§5.5 (Eq. (6))] The pressure recovery is not justified. The paper defines P by the pointwise path integral ∫_0^x h(y)·dγ(y), where h=(I−P)[f+Δu−(u·∇)u] is, at the level of regularity established in the paper, only an L^2 or distributional field on Ω. Path integrals of such fields are not defined, and the observation that (I−P) maps onto 'conservative' or 'pure divergence' fields gives at most a weak/distributional gradient representation h=∇q; it does not imply that q is represented by a pointwise path-independent integral. No proof is provided that P∈L²(Ω), that the path integral is well defined, or that the map (u,f)↦P is computable. Thus the strong solution (u,P) claimed in Theorem 1 is not established by the written argument.
  3. [§5.2 (Fact 5, paragraph after the proof)] The effectiveness of the entire nonlinear construction depends on the constants C, M, and C_α in Fact 5 being computable, but this is only asserted in the paragraph following Fact 5 and deferred to a forthcoming paper. These constants enter the definition of ~C=c1 M B1 in the proof of Proposition 4, the recursive bounds in Claims 1–4, and the effective convergence rate of the iteration. Consequently the main theorem is, as written, conditional on an unproven premise: either the computability of these constants must be proved in this paper, or the theorem must be stated as an explicit assumption. This is a load-bearing issue, not a cosmetic one, because the claimed uniform computability of the solution operator is exactly what requires effective control of these constants.
minor comments (4)
  1. [§2 (after Eq. (10))] The phrase 'Qσ_0[R²] is “too big” to be used as a set of codes' could be misread: Qσ_0[R²] is not a subspace of Lσ_{2,0}, and the issue is that its closure in L² contains a proper superspace of the desired space. A more precise wording would help.
  2. [§3 (Proposition 2)] The proof says the argument is carried out on Ω=(0,1)² and then 'carries over' to Ω=(-1,1)² by scaling. It would be clearer to state the scaling explicitly, since the representation and the basis functions both change.
  3. [§5.1 (Lemma 4)] Lemma 4 is stated for s≥1, but the discussion immediately after it cites the Sobolev embedding H^s⊂C only for s>1. The proof of Lemma 4 itself does not use pointwise values, so the statement is fine, but the juxtaposition is momentarily confusing.
  4. [§5.4 (proof of Proposition 5)] In the estimate preceding the definition of η, the factor 2^{-2n} appears twice on the right-hand side of the chain of inequalities; the displayed estimate should be checked to ensure the powers of 2 are not counted twice.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular reduction; the effective iteration proof is not equivalent to its inputs, though two advertised ingredients are asserted rather than proved.

full rationale

The paper's derivation is not circular in the sense of assuming its conclusion. The linear Stokes semigroup is computed from a contour representation and explicit effective bounds in Proposition 3, and the nonlinear homogeneous case is proved by effectivizing the classical contraction estimates, with T(a) computed from the data and the stated constants and the solution u obtained as an effective limit of computable iterates. There is no fitted parameter renamed as a prediction and no uniqueness theorem imported from the authors' own prior work to force the choice. The two passages that warrant attention are proof gaps rather than circularity. First, after Fact 5 the paper records 'without going into the details, that the constants C, M, and Cα ... are in fact computable (some general discussions ... are forthcoming)', and Claim 1 later uses M and c1 'by assumption'; this is a deferred lemma, not a reduction of the conclusion to its inputs. Second, Section 5.5 states for the inhomogeneous case: 'Similarly to (the proofs of) Propositions 5, 4, and [24, Lemma 3.7], this solution is seen to be computable', and recovers pressure from a path integral of h=(I-P)[f+Δu-(u·∇)u] without proving the required regularity; this is an omitted proof of part of Theorem 1. The only self-citation involved, [24], is used as an analogy for an integration lemma and is not a load-bearing uniqueness or ansatz assumption. These gaps affect correctness, not circularity; hence the low score.

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

The central claim rests on standard semigroup and interpolation theory, on external existence theorems, and on a few unproved computability assertions, most notably the computability of constants C, M, C_alpha. No data-fitted parameters appear. No new physical or mathematical entities are postulated.

assumptions (5)
  • domain assumption Local existence and convergence of the iteration (5) for the inhomogeneous mild equation follows from Giga-Miyakawa [5, Theorem 2.3].
    Invoked in Section 5.5 to extend the homogeneous proof to general forcing f. The paper gives no independent proof for the inhomogeneous case and relies on the external theorem.
  • ad hoc to paper The constants C, M, and C_alpha in Fact 5 are computable.
    Section 5.2 states these constants are in fact computable with details forthcoming. The algorithms for T(a) and the convergence bounds depend on these constants, so this is a load-bearing unproved assertion.
  • standard math Sobolev embedding H^s_2(Omega) into C(Omega) for s > 1 holds with a computable constant C_s.
    Used in Lemma 5 for the computability of multiplication. The paper cites the classical embedding theorem and asserts the constant is computable without proof.
  • standard math The Stokes operator A is a positive self-adjoint operator generating an analytic semigroup, with interpolation domains D(A^alpha) related to Sobolev spaces for 2alpha < 3/2.
    Invoked throughout Sections 4 and 5 through Pazy [10], Giga [3], and Sohr [18]. This is standard semigroup and interpolation theory, not new to the paper.
  • domain assumption For a in P, the weighted Fourier sum sum (1+n^2+m^2)|a_{n,m}|^2 is computable from the code of a.
    Used in the proof of Proposition 3 to truncate the contour integral for e^{-tA}a. The paper cites Zhong [28] for this fact but does not prove it here.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Computability of the Solutions to Navier-Stokes Equations via Recursive Approximation." pith.science (2026). https://pith.science/paper/ZJ2KX4GU

@misc{pith2026190801226,
  author       = {Pith},
  title        = {Pith review of: Computability of the Solutions to Navier-Stokes Equations via Recursive Approximation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZJ2KX4GU}},
  note         = {Machine review of arXiv:1908.01226}
}
abstract

As one of the seven open problems in the addendum to their 1989 book "Computability in Analysis and Physics", Pour-El and Richards proposed ``... the recursion theoretic study of particular nonlinear problems of classical importance. Examples are the Navier-Stokes equation, the KdV equation, and the complex of problems associated with Feigenbaum's constant.'' In this paper, we approach the question of whether the Navier-Stokes Equation admits recursive solutions in the sense of Weihrauch's Type-2 Theory of Effectivity. A natural encoding (``representation'') is constructed for the space of divergence-free vector fields on 2-dimensional open square $\Omega = (-1, 1)^2$. This representation is shown to render first the mild solution to the Stokes Dirichlet problem and then a strong local solution to the nonlinear inhomogeneous incompressible Navier-Stokes initial value problem uniformly computable. Based on classical approaches, the proofs make use of many subtle and intricate estimates which are developed in the paper for establishing the computability results.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 30 canonical work pages

  1. [1]

    Philosophical Transactions of the Royal Soc iety A: Mathematical, Physical and Engineering Sciences 370, 3359–3384 (2012)

    Beggs, E., Costa, J.F., Tucker, J.V.: Axiomatising Physi cal Experiments as Oracles to Algorithms. Philosophical Transactions of the Royal Soc iety A: Mathematical, Physical and Engineering Sciences 370, 3359–3384 (2012)

    work page 2012

  2. [2]

    F., Fabrie, P.: Mathematical Tools for the Study of the Incompressible Navier-Stokes Equations and Related Models

    Boyer. F., Fabrie, P.: Mathematical Tools for the Study of the Incompressible Navier-Stokes Equations and Related Models. Spring Applie d Mathematical Sci- ences, (2013)

    work page 2013

  3. [3]

    Giga, Y.: Weak and Strong Solutions of the Navier-Stokes I nitial Value Problem. Publ. RIMS, Kyoto Univ. 19, 887–910 (1983)

    work page 1983

  4. [4]

    Giga, Y.: Time and spatial analyticity of solutions of the Navier-Stokes equations. Comm. Partial Differential Equations 8, 929–948 (1983)

    work page 1983

  5. [5]

    Archive for Rational Mechanics and Analysis 89 (3), 267–281 (1985)

    Giga, Y., Miyakawa, T.: Solutions in Lr of the Navier-Stokes initial value problem. Archive for Rational Mechanics and Analysis 89 (3), 267–281 (1985)

    work page 1985

  6. [6]

    A.: Finite Element Methods for Na vier-Stokes Equations

    Girault, V., Raviart, P. A.: Finite Element Methods for Na vier-Stokes Equations. Springer Series in Computational Mathematics 5. Springer, New York (1986)

    work page 1986

  7. [7]

    abstract in: Bulletin of Symbolic Logic 20(2), 231 (2014)

    Kawamura, A., Steinberg, F., Ziegler, M.: Complexity of L aplace’s and Poisson’s equation. abstract in: Bulletin of Symbolic Logic 20(2), 231 (2014). full version to appear in Mathem. Structures in Computer Science (2016)

    work page 2014

  8. [8]

    Landriani, G.S., Vandeven, H.: Polynomial approximatio n of divergence-free func- tions. Math. Comput. 52, 103–130 (1989)

    work page 1989

Show all 30 references
  1. [9]

    Cambridge University Press, London (2000)

    Mclean, W.: Strongly Elliptic Systems and Boundary Integ ral Equations. Cambridge University Press, London (2000)

    2000

  2. [10]

    Springer-Verlag, New York (1983)

    Pazy, A.: Semigroups of Linear Operators and Applicatio ns to Partial Differential Equations. Springer-Verlag, New York (1983)

    1983

  3. [11]

    Advances in Mathematics 39(4), 215–239 (1981)

    Pour-El, M.B., Richards, J.I.: The wave equation with co mputable initial data such that its unique solution is not computable. Advances in Mathematics 39(4), 215–239 (1981)

    1981

  4. [12]

    Springer, New York (1989)

    Pour-El, M.B., Richards, J.I.: Computability in Analys is and Physics. Springer, New York (1989)

    1989

  5. [13]

    Pour-El, M.B., Zhong, N.: The wave equation with computa ble initial data whose unique solution is nowhere computable. Math. Logic Quarter ly 43(4), 499–509 (1997). 30 S. M. Sun, N. Zhong, M. Ziegler

    1997

  6. [14]

    Internat ional Journal for Numerical Methods in Fluids 5(3), 225–244 (1985)

    Patel, M.K., Markatos, N.C., Cross, M.: A critical evalu ation of seven discretiza- tion schemes for convection-diffusion equations. Internat ional Journal for Numerical Methods in Fluids 5(3), 225–244 (1985)

    1985

  7. [15]

    Theoretical Computer Science 305, 43–76 (2003)

    Brattka, V., Presser, G.: Computability on subsets of me tric spaces. Theoretical Computer Science 305, 43–76 (2003)

    2003

  8. [16]

    N EC preprint (2003)

    Smith, W.D.: On the uncomputability of hydrodynamics. N EC preprint (2003)

    2003

  9. [17]

    Bulletin of S ymbolic Logic 2, 284–321 (1996)

    Soare, R.I.: Computability and recursion. Bulletin of S ymbolic Logic 2, 284–321 (1996)

    1996

  10. [18]

    Birkh¨ auser Advanced Texts

    Sohr, H.: The Navier-Stokes Equations: An Elementary Fu nctional Analytic Ap- proach. Birkh¨ auser Advanced Texts. Birkh¨ auser, New York(2001)

    2001

  11. [19]

    in: Proc

    Sun, S.M., Zhong, N., Ziegler, M.: Computability of Navi er-Stokes equation. in: Proc. 11th Conf. on Computability in Europe. Springer LNCS 9136. Springer, New York (2015)

    2015

  12. [20]

    Journal of the American Mathematical Society 29, 601-674 (2016)

    Tao, T.: Finite time blowup for an averaged three-dimens ional Navier-Stokes equa- tion. Journal of the American Mathematical Society 29, 601-674 (2016)

    2016

  13. [21]

    North- Holland Publishing Company, New York (1977)

    Temam, R.: Navier-Stokes Equations: Theory and Numeric al Analysis. North- Holland Publishing Company, New York (1977)

    1977

  14. [22]

    S pringer, New York (2000)

    Weihrauch, K.: Computable Analysis: an Introduction. S pringer, New York (2000)

    2000

  15. [23]

    Weihrauch, K., Zhong, N.: Is wave propagation computabl e or can wave comput- ers beat the Turing machine?. Proc. London Mathematical Soc iety 85(2), 312–332 (2002)

    2002

  16. [24]

    Theoreti cal Computer Science 332, 337–366 (2005)

    Weihrauch, K., Zhong, N.: Computing the solution of the K orteweg-de Vries equa- tion with arbitrary precision on Turing machines. Theoreti cal Computer Science 332, 337–366 (2005)

    2005

  17. [25]

    Journal of Complexity 22(6), 918–935 (2006)

    Weihrauch, K., Zhong, N.: Computing Schr¨ odinger propa gators on Type-2 Turing machines. Journal of Complexity 22(6), 918–935 (2006)

    2006

  18. [26]

    Mathematical Logic Quarterly 53, 511–531 (2007)

    Weihrauch, K., Zhong, N.: Computable analysis of the abs tract Cauchy problem in Banach spaces and its applications I. Mathematical Logic Quarterly 53, 511–531 (2007)

    2007

  19. [27]

    Jahres- bericht der Deutschen Mathematiker Vereinigung (DMV) 101(1), 1–25 (1999)

    Wiegner, M.: The Navier-Stokes equations – a never-endi ng challenge?. Jahres- bericht der Deutschen Mathematiker Vereinigung (DMV) 101(1), 1–25 (1999)

    1999

  20. [28]

    : Computability structure of the Sobolev space s and its applications

    Zhong, N. : Computability structure of the Sobolev space s and its applications. Theoretical Computer Science 219, 487–510 (1999)

    1999

  21. [29]

    Theoretical Computer Science 326, 187–211 (2004)

    Ziegler, M., Brattka, V.: Computability in linear algeb ra. Theoretical Computer Science 326, 187–211 (2004)

    2004

  22. [30]

    Applied Mathematics and Computation 215(4), 1431-1447 (2009)

    Ziegler, M.: Physically-relativized Church-Turing hy potheses: Physical foundations of computing and complexity theory of computational physic s. Applied Mathematics and Computation 215(4), 1431-1447 (2009). A Proof of Proposition 1 (a) For a divergence-free and boundary-free ...

    2009

Pith tools

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