REVIEW 3 major objections 3 minor 5 references
A short note on Meyers' theorem
T0 review · 3 major / 3 minor · reviewed 2026-08-10 · deepseek-v4-flash
Pith's one-line read This note widens the dimension-free Meyers interval for gradient estimates of elliptic divergence-form equations, proving the gradient is \(L^p\)-bounded for every \(p\) between \((26K-16)/(13K-3)\) and \((26K-16)/(13K-13)\).
desk verdict A short note improving the Meyers interval by elementary interpolation, but the gain is conditional on an unproved Riesz-transform bound from an arXiv preprint. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing object is \(T=I+2(R\otimes R)\), where \(R=(R_1,\dots,R_d)\) is the vector Riesz transform and \(R\otimes R\) acts on vector-valued functions by \((R\otimes R F)_i=\sum_j R_iR_jF_j\). The earlier criterion [4] says that \(p\) lies in the Meyers interval whenever \(\|T\|_{p,p}<(K+1)/(K-1)\). The new work is Lemma 1, which bounds the left side by \(2.6(p^*-2)+1\). To get there, the paper uses the bound \(\|R\otimes R\|_{p,p}\le p^*-1\), interpolates between \($L^{2}$\) and \($L^{6}$\) with \(\|T\|_{6,6}\le 11\), and extends the bound to \(p<2\) by duality.
What would settle it
Compute the operator norm of \(R\otimes R\) on vector-valued \(L^p(\mathbb{R}^d)\) for a specific case, say \(d=2\) and \(p=6\), and check whether it is at most \(p^*-1=5\). A value above 5 would falsify Proposition 2 and with it the bound \(\|T\|_{6,6}\le 11\) that Lemma 1 relies on.
Extended reading notes
Core claim
The paper's central claim is Theorem 1: for every \(K>1\) and every dimension \(d\ge2\), the interval \(\left(\frac{26K-16}{13K-3},\frac{26K-16}{13K-13}\right)\) is contained in the set of exponents \(p\) for which \(\|\nabla u\|_p\le C\|F\|_p\) holds for weak solutions of \(\operatorname{div}(M\nabla u)=\operatorname{div}F\) with Dirichlet boundary conditions. The proof proceeds by improving the norm bound on the operator \(T=I+2(R\otimes R)\). Using the estimate \(\|R\otimes R\|_{p,p}\le p^*-1\), the triangle inequality, Riesz\textendash Thorin interpolation between \($L^{2}$\) and \($L^{6}$\), and duality, the paper obtains \(\|T\|_{p,p}\le 2.6(p^*-2)+1\). Combining this with the known criterion \(\|T\|_{p,p}<(K+1)/(K-1)\) yields the claimed interval.
Load-bearing premise
The proof leans on an unproved estimate for a certain Riesz-transform operator: its norm on \(L^p\) is at most \(p^*-1\); if that estimate is wrong, the interval collapses.
Editorial extensions
If this is right
- For every \(K>1\) and every \(d\ge2\), the gradient estimate holds on an interval strictly wider than the previous dimension-free interval from [4].
- As \(K\to1^+\), both endpoints tend to \(2\), matching the energy estimate; for large \(K\), the interval length grows roughly like \(20/(13K)\).
- Because the only operator bound used at \(p=6\) is \(\|T\|_{6,6}\le 11\), any sharper value there directly widens the interval; the note itself remarks that the constant \(2.6\) can be lowered to \(2.55\).
- The result transfers from \(\mathbb{R}^d\) to bounded \(C^2\) domains with Dirichlet conditions by standard localization, so the estimate holds in the original boundary-value setting.
Reading between the lines
- Beyond the paper: if the quoted bound \(\|R\otimes R\|_{p,p}\le p^*-1\) is independently verified, the same interpolation scheme could be run with an \(L^q\) bound at a different \(q>2\), producing a parametric family of intervals and possibly wider endpoints.
- Beyond the paper: because the proof is dimension-free, it does not exploit the special structure of \(d=2\), where the conjectured optimal interval is already known; deciding optimality in higher dimensions would require a dimension-dependent argument.
- Beyond the paper: a direct numerical or analytic test of the Riesz-transform bound would settle the paper's contribution without revisiting the elliptic PDE itself.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The note claims an improvement of the dimension-free Meyers interval for the L^p gradient estimate of weak solutions to div(M∇u)=div F. The main result, Theorem 1, states that the Meyers interval contains ((26K−16)/(13K−3), (26K−16)/(13K−13)). The proof proceeds by importing a norm estimate for the tensor product of Riesz transforms (Proposition 2, itself a restriction of Proposition 1 from Cassese's arXiv preprint), using it to bound ∥T∥_{6,6} ≤ 11, interpolating between L^2 and L^6 to obtain the bound ∥T∥_{p,p} ≤ 2.6(p*−2)+1 in Lemma 1, and finally combining this with the Iwaniec–Sbordone criterion (4). The proof of Lemma 1 is logically transparent: the Riesz–Thorin step, the reduction to a one-variable inequality, and the convexity argument are all clearly presented, conditional on the imported bound and on the numerical evaluations being made rigorous.
Significance. If the imported bound ∥R⊗R∥_{p,p} ≤ p*−1 is valid, the result would genuinely improve the best known dimension-free Meyers interval of Iwaniec and Sbordone. The argument has notable strengths: it introduces no free parameters, it is not circular (the theorem follows transparently from an explicitly stated external estimate plus an elementary interpolation argument), and the structure of the proof is easy to verify line by line. The claimed interval is also asymptotically consistent with existing bounds for large K. However, the central load-bearing estimate is not proved in the manuscript and is taken from an unreviewed arXiv preprint, and the numerical checks in Lemma 1 are stated with approximate values rather than rigorous inequalities. The significance of the contribution is therefore conditional on external validation and on tightening the proof of Lemma 1.
major comments (3)
- [Section 2, Proposition 2] The entire improvement over the Iwaniec–Sbordone interval rests on Proposition 2, but Proposition 2 is proved only as a restriction of Proposition 1 of the unreviewed arXiv preprint [3], and Proposition 1 itself is not proved in this note. This is load-bearing: in Lemma 1 the endpoint value ∥T∥_{6,6} ≤ 11 is exactly 1 + 2(6−1); if one instead used the previously available estimate ∥T∥_{6,6} ≤ 30 from [2], the Riesz–Thorin step would not give (7), and the interval of Theorem 1 would not follow. The note must either provide an independent proof of Proposition 1 (or a direct proof of Proposition 2) or cite a peer-reviewed source for it.
- [Section 2, Lemma 1] The proof that G(θ) ≥ 0 on [0,1] uses the numerical evaluations G(θ₀) ≈ 0.358 and G(θ₁) ≈ 0.1591, G′(θ₁) ≈ −0.0081, with no error bounds. These approximate values are used to conclude positivity of the chord and the tangent line, so they are part of the proof. The argument can likely be made rigorous with interval arithmetic or by exhibiting explicit rational or exponential bounds, but as written the proof is not complete. Please replace all approximate evaluations in this step by rigorous inequalities or certified interval computations.
- [Section 1, reduction to R^d] The abstract and Theorem 1 concern a bounded domain Ω with Dirichlet boundary conditions, but the proof is carried out on Ω = R^d. The sentence 'everything stated below easily extends to domains with an appropriately smooth boundary' is an assertion, not a proof. The Riesz-transform argument is global in nature, and the transfer to bounded domains with discontinuous coefficients requires a localization or extension argument. The manuscript should either supply that argument or explicitly restrict the theorem to R^d.
minor comments (3)
- [Throughout] The notation p* is used in Conjecture 2, Proposition 2, and Lemma 1 but is never defined. From context and from [4] the reader must infer that p* denotes max{p, p/(p−1)}; please define it explicitly.
- [References] Reference [3] is an arXiv preprint; please state its current publication status if available, and ensure that a result so central to the paper's main theorem is not left dependent on an unreviewed source.
- [Section 2, proof of Lemma 1] The sentence 'Since T is self-adjoint on L^2, duality yields ∥T∥_{p,p} = ∥T∥_{p′,p′}' is correct only if one also notes that the matrix operator T is symmetric in the sense that R_i and R_j commute; this is true for Riesz transforms, but the reason should be stated briefly.
Circularity Check
No significant circularity: Theorem 1 is an algebraic and interpolation consequence of a clearly stated external estimate, not of a fitted or self-referential input.
full rationale
The derivation chain is transparent. Proposition 2 (the vector Riesz product bound ||R⊗R||_{p,p} ≤ p*−1) is imported from Cassese's Proposition 4.1 via a valid restriction argument; Lemma 1 then combines this bound at p=6 with the L2 isometry of T and Riesz–Thorin interpolation to obtain the upper bound ||T||_{p,p} ≤ 2.6(p*−2)+1. Combining that bound with Iwaniec–Sbordone's sufficient condition (||T||_{p,p} < (K+1)/(K−1)) yields Theorem 1 by elementary algebra. Each step uses its stated inputs and does not assume its conclusion. There is no self-citation, no fitted parameter relabeled as a prediction, no uniqueness theorem imported from the authors' own prior work, and no renaming of a known result. The cited bound from the Cassese preprint is external and is the load-bearing assumption, but reliance on an unproved external estimate is a correctness or verification concern, not circularity; the note openly states the input, and the final interval is a direct consequence of the cited bound plus standard interpolation.
Assumptions & free parameters
assumptions (5)
- domain assumption M is measurable, symmetric, and uniformly elliptic with ratio K.
- domain assumption If ∥T∥_{p,p} < (K+1)/(K−1), then p belongs to the Meyers interval.
- ad hoc to paper ∥R⊗R∥_{p,p} ≤ p*−1 for all d≥2 and 1<p<∞.
- standard math Riesz-Thorin interpolation theorem.
- standard math T is self-adjoint on L^2, so ∥T∥_{p,p} = ∥T∥_{p',p'}.
Cite this review
Pith. "Pith review of A short note on Meyers' theorem." pith.science (2026). https://pith.science/paper/SH5UWEX6
@misc{pith2026260807030,
author = {Pith},
title = {Pith review of: A short note on Meyers' theorem},
year = {2026},
howpublished = {\url{https://pith.science/paper/SH5UWEX6}},
note = {Machine review of arXiv:2608.07030}
}
abstract
We consider weak solutions of the second-order elliptic equation $\operatorname{div} (\mathbf{M}\nabla u) = \operatorname{div}\mathbf{F}$ in $\Omega\subset {\mathbb R}^d$ with Dirichlet boundary conditions, where $\mathbf{M}$ is a uniformly elliptic real-valued symmetric matrix $\mathbf{M}:\Omega \rightarrow \mathbb{R}^{d\times d}$ such that $\frac{1}{K} |\xi|^2 \leq\langle \mathbf{M}\xi,\xi \rangle \leq K |\xi|^2, \forall \xi \in \mathbb{R}^d, K > 1$. We prove that $\|\nabla u\|_p \leq C\|\mathbf{F}\|_p$ for any $p\in \left(\frac{26K-16}{13K-3}, \frac{26K-16}{13K-13}\right)$.
Reference graph
Works this paper leans on
-
[3]
Cassese, G.: Korn’s inequality from the viewpoint of calculus of variations. arXiv:2603.22431. 3
-
[2]
Ba˜ nuelos, R., Lindeman II, A.: A Martingale Study of the Beurling-Ahlfors Transform inR n. J. Funct. Anal.,145, No. 1, 224–265 (1997). 2
work page 1997
- [1]
-
[4]
Iwaniec, T., Sbordone, C.: Quasiharmonic fields. Ann. I.H.Poincar´ e, Non Lin- ear Analysis,18, No. 5, 519-572 (2001) 2, 3 A SHORT NOTE ON MEYERS’ THEOREM 5
work page 2001
-
[5]
G.: AnL p-estimate for the gradient of solutions of second-order elliptic divergence equations
Meyers, N. G.: AnL p-estimate for the gradient of solutions of second-order elliptic divergence equations. Ann. Scuola Norm. Sup. Pisa,17, No. 3, 189-206 (1963). 1, 2 SCAD Soft Ltd., 3a Osvity Street, Kyiv, 03037, Ukraine Email address:mikeperelmuter@gmail.com
work page 1963
Reviewed August 10, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.