New Kakeya estimates using Gromov's algebraic lemma

Joshua Zahl

REVIEW 1 major objections 3 minor 27 references

New Kakeya estimates using Gromov's algebraic lemma

T0 review · 1 major / 3 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 A nested-variety tube-counting inequality yields Kakeya maximal function estimates at dimension at least (2−√2)n, new for n≥5 except n=6.

desk verdict First-rate Kakeya paper with a genuine but likely repairable gap in Lemma 2.6; referee should require a Sard-type argument before accepting. read the letter →

arxiv 1908.05314 v4 pith:Z7ADHWX5 submitted 2019-08-14 math.CA

Joshua Zahl This is my paper · ORCID

classification math.CA MSC 42B2528A7814P10

keywords KakeyamaximalfunctionBesicovitchsetsdirection-separatedtubespolynomialWolffaxiomssemi-algebraicYomdin-GromovalgebraiclemmamultilinearHausdorffdimension

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 new quantitative overlap bounds for thin tubes in $\mathbb{R}^n$ with separated directions. Its main result is a Kakeya maximal function estimate at dimension $d(n)=\max_{2\le k\le n}\min(n-k+2,(n^2+k^2+n-k)/(2n))$, which in particular satisfies $d(n)\ge(2-\sqrt{2})n$; for $n\ge5$, $n\ne6$, these exponents are better than any previously known bound. The key new ingredient is a geometric inequality: direction-separated $1\times\delta$ tubes cannot cluster near a nested sequence of low-degree algebraic varieties, with a count bounded by $\delta^{d+1-n-\varepsilon}/(r_1\cdots r_d)$. The inequality is proved using the algebraic lemma that parametrizes semi-algebraic sets by smooth maps, then combined with a multilevel 'grains' decomposition and the standard linear/multilinear conversion to reach the maximal function statement. If the paper is correct, it also gives new lower bounds on the Hausdorff dimension of Besicovitch sets for certain intermediate $n$.

What carries the argument

The load-bearing object is the nested-variety tube-counting inequality (Theorem 1.9), which bounds how many direction-separated tubes can survive successive low-degree algebraic constraints. It is proved in Section 2 by combining Theorem 1.7 (the polynomial Wolff axiom for a single semi-algebraic set), Lemma 2.6 (tubes contained in a semi-algebraic set cannot expand much when extended, proved through the algebraic lemma parametrizing semi-algebraic sets by $C^r$ maps), and the tubular-neighborhood volume bound for real algebraic varieties. Section 3 organizes tubes into a tree of 'grains' (Proposition 3.5), and inequality (4.14) is the decisive step: the grains are read as nested varieties and Theorem 1.9 bounds the number of surviving sub-tubes per level. Lemma 1.4 then converts the resulting direction-separated multilinear estimate into a maximal function estimate.

What would settle it

Construct a direction-separated family of $1\times\delta$ tubes in $\mathbb{R}^5$ near a nested pair $Z_1\supset Z_2$ (for instance a plane and a line inside it) and count how many tubes have prescribed intersection fractions $r_1,r_2$ with the two $2\delta$-neighborhoods; if the number exceeds $C\delta^{1-5-\varepsilon}/(r_1r_2)$ by more than a $\delta^{-\varepsilon}$ factor for every $\varepsilon$, Theorem 1.9 and the claimed $d(5)=3.6$ fail. Alternatively, a Besicovitch set in $\mathbb{R}^5$ of Hausdorff dimension strictly below $3.6$ would contradict the maximal-function estimate at dimension $d=3.6$.

Watch

Extended reading notes

Core claim

The paper establishes that direction-separated multilinear Kakeya estimates hold at exponent $d=(n^2+k^2+n-k)/(2n)$ for $2\le k\le n$ (Theorem 1.3), and that these estimates convert into a Kakeya maximal function estimate at dimension $d(n)=\max_{2\le k\le n}\min(n-k+2,(n^2+k^2+n-k)/(2n))$ (Theorem 1.5). The engine is Theorem 1.9: for nested real algebraic varieties $Z_1\supset\cdots\supset Z_d$ of codimension at least $i$ and degree at most $E$, and radii $1\ge r_1\ge\cdots\ge r_d\ge\delta$, the number of direction-separated $1\times\delta$ tubes whose intersection with each $2\delta$-neighborhood inside $B(x,r_i)$ is at least $r_i|T|$ is at most $C(n,E,\varepsilon)\delta^{d+1-n-\varepsilon}/(r_1\cdots r_d)$. This nested inequality is proved from the polynomial Wolff axioms for a single semi-algebraic set together with a tube-extension lemma, and Section 4 applies it to the 'grains' produced by a multilevel polynomial partitioning tree. The conversion to the maximal function is via the linear/multilinear mechanism of Lemma 1.4, and the final exponents are new for all $n\ge5$ except $n=6$.

Load-bearing premise

The proof leans on the polynomial Wolff axiom (Theorem 1.7), cited without proof, which limits how many separated tubes can mostly lie inside a semi-algebraic set of bounded complexity; if that estimate fails for some semi-algebraic set, the nested-variety inequality and the new exponents do not follow.

Editorial extensions

If this is right

Kakeya maximal function estimates hold in $\mathbb{R}^n$ at dimension $d(n)\ge(2-\sqrt{2})n$ for every $n$, and for $n\ge5$, $n\ne6$, this strictly improves the previous best bound $(4n+3)/7$.
Because a maximal-function estimate at dimension $d$ forces every Besicovitch set to have Hausdorff dimension at least $d$, the theorem gives new lower bounds such as $4.857$ in $\mathbb{R}^7$ and $5.25$ in $\mathbb{R}^8$.
The direction-separated multilinear estimate at $d=(n^2+k^2+n-k)/(2n)$ is sharp for $k=n-1$, showing that the direction-separation condition—not just multilinearity—carries the improvement.
The final exponent is tied to Theorem 1.9 through inequality (4.14), so any later improvement to the nested-variety counting bound feeds directly into stronger maximal function estimates via the same proof.
As $n$ grows, the optimizing $k$ is approximately $(\sqrt2-1)n$, so $d(n)/n\to2-\sqrt2$; the gap to the conjectured dimension $n$ is a linear factor that this method does not close.

Reading between the lines

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

The tube-extension lemma uses only semi-algebraic structure, so the nested counting inequality should extend from algebraic varieties to general semi-algebraic sets of bounded complexity; Lemma 2.11 already gestures in this direction, and such a version would apply to Fourier restriction problems where wave packets concentrate near several algebraic levels at once.
The iterative cutting procedure in Lemma 3.10—alternating between shortening tubes and shrinking balls until two conflicting inequalities hold—is a reusable mechanism that could feed $k$-broad restriction estimates, potentially covering the missing case $n=6$ if assembled differently.
A natural stress test is whether the sharp examples for $k=n-1$ also saturate Theorem 1.9; if they do, the constant $2-\sqrt2$ is the true limit of this method rather than of the Kakeya problem itself.

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

Here's what you should know: this is a genuinely important paper—it improves the Kakeya maximal function exponent for all n≥5 except 6, via a new nested-variety tube-counting inequality and a multilevel grains decomposition. The main structure is sound and the writing is clear. But there's a hole in the proof of Lemma 2.6 that isn't mentioned in the paper, and it needs to be fixed before the theorem is fully justified.

The gap is in the proof of (2.23). The author defines J and J' as the boundaries of the images of two maps, and then claims that because the maps are continuous, they send ∂U onto J and J'. That's false for non-injective maps: think h(x)=x² on [-1,1]. The boundary of the image contains 0, which is not the image of the boundary. So the subsequent estimate |N_{δ^{2n-2}}(J)|≲δ^n does not control the symmetric difference between the two images. Yomdin–Gromov gives Lipschitz parametrizations but not injectivity; without an argument handling critical values, the lower bound on the slice is not justified. A standard fix would be to include the critical set of the polynomial map in the boundary estimate, using semialgebraic stratification or Sard's theorem with degree bounds; the pieces are all available, but they are not in the manuscript.

This matters because Lemma 2.6 feeds into Lemma 2.11 and Theorem 1.9, and Theorem 1.9 is the key geometric input used at inequality (4.14) to get the lower bounds on the D_i. So as written, the central exponent d(n) rests on that unproved claim. I want to be careful: this doesn't look like a wrong idea, it looks like a missing standard argument. The rest of the proof—the grains decomposition, the algebra in Section 4—checks out structurally. The paper is also honest about concurrent work and the history of the polynomial Wolff axioms.

My recommendation: send it to a serious referee. The result is important and the gap is likely repairable, but it should not be accepted as is. If the authors supply the missing Sard-type argument, this is a strong paper. For a reading group, it's actually a great example of why boundary-of-image arguments need care.

Referee Report

1 major / 3 minor

Summary. This paper proves new Kakeya maximal function estimates in R^n for n >= 5, n != 6. The main new ingredients are a geometric inequality (Theorem 1.9) bounding the number of direction-separated tubes that pass near a nested sequence of low-degree algebraic varieties, and a hierarchical 'grains' decomposition (Proposition 3.5) for families of tubes. These are combined with a direction-separated multilinear Kakeya estimate (Theorem 1.3) and the Bourgain-Guth multilinear-to-linear argument (Lemma 1.4) to obtain maximal function estimates at dimension d(n) = max_k min(n-k+2, (n^2+k^2+n-k)/(2n)), in particular d(n) >= (2 - sqrt(2))n for all n, improving earlier bounds in all dimensions except 2,3,4,6.

Significance. If the proofs are correct, the results are a substantial advance in the Kakeya problem: they improve the high-dimensional bound from (4n+3)/7 (Katz-Tao) and the intermediate-dimensional Hickman-Rogers bounds, and give new Hausdorff dimension estimates for Besicovitch sets in certain dimensions. The paper is well organized, with the main new lemmas proved in detail and external dependencies clearly cited. The grains decomposition (Section 3) and the nested-variety tube count (Theorem 1.9) are likely to be useful tools beyond this paper. The main concern is a gap in the proof of Lemma 2.6 that undermines Theorem 1.9 as written.

major comments (1)

[Section 2.2, proof of Lemma 2.6, derivation of (2.23)] The derivation of (2.23) is not justified as written. In the proof, J = boundary({(F(x),0)+t(G(x),1): x in U}) and J' = boundary({(F_{i0}(x),0)+t(G_{i0}(x),1): x in U}), and it is asserted that since the maps are continuous, they map boundary(U) onto J and J'. This is false for non-injective maps; for instance, h(x)=x^2 on [-1,1] has 0 in the boundary of h([-1,1]) but 0 is not in h({-1,1}). Consequently, the estimate |N_{delta^{2n-2}}(J)| << delta^n does not control the measure of the symmetric difference between the two images, because interior folds and critical values can create image-boundary components that do not come from boundary(U). Since (2.23) is precisely the step that converts the Yomdin-Gromov parametrization into a lower bound on |(F+tG)(U)|, Lemma 2.6 and therefore Corollary 2.10, Lemma 2.11, and Theorem 1.9 are not proved by the given argument. Theorem 1.9 is applied at (4.14) to obtain the D_i lower bounds used in (4.20)-(4.28), so the gap propagates to the main dimension estimate. A repair will likely require adding the critical locus of (F_{i0},G_{i0}) (or of the slice map) to boundary(U) and controlling its delta-neighborhood with semialgebraic degree bounds; this argument is absent from the manuscript.

minor comments (3)

[Section 4, equation (4.22)] Equation (4.22) and the following parenthetical remark use the index j where i is meant: the exponent should be k(k-1)/((n-i+1)(n-i)(n-i-1)), and the remark about the denominator being nonzero should refer to i, not j.
[Section 2.2, proof of Lemma 2.7] There is a typesetting artifact in the proof of Lemma 2.7 where '||P||/suppress L1(J)' appears; this should read ||P||_{L^1(J)}.
[Section 2.2, proof of Lemma 2.6] Several inequalities in the proof of Lemma 2.6 contain the garbled symbol '/greaterorsimilar' (for example, the line '|S'| /greaterorsimilar lambda delta^{n-1} >= delta^n'); these should be typeset as \gtrsim or \gtrsim_{\epsilon}.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the Kakeya bound is derived from cited external theorems and new intermediate lemmas, with no target estimate assumed or fitted.

full rationale

The paper's central claims are not assumed as inputs. Theorem 1.5 follows from Theorem 1.3 and the standard Bourgain-Guth/Hickman-Rogers multilinear-to-linear reduction (Lemma 1.4); Theorem 1.3 is proved via the grains decomposition (Proposition 3.5) and the nested-variety tube-counting inequality (Theorem 1.9). Theorem 1.9 is proved from Theorem 1.7 (Katz-Rogers polynomial Wolff axioms), Wongkew's volume bound, the Yomdin-Gromov algebraic lemma, and a semi-algebraic fiber-selection lemma. The exponent d(n) is obtained by optimizing the derived inequalities, not by fitting a parameter to the claimed bound. Several cited ingredients are by the author or collaborators, but they are prior published results with independent proofs: Theorem 1.7 is by Katz and Rogers, Lemma 2.5 is cited to the author's earlier paper with a cross-reference to Katz-Rogers, and the Yomdin-Gromov lemma is cited to Burguet. These citations are not load-bearing in a circular sense because none of them states or assumes the Kakeya maximal function estimate being proved. The possible objection flagged in the proof of Lemma 2.6 about the boundary-inclusion assertion is a correctness concern, not a circularity concern, and per the review rules it does not raise the circularity score.

Assumptions & free parameters 0 free parameters · 9 assumptions · 0 invented entities

The paper introduces no free parameters or invented entities. It relies on standard real algebraic geometry theorems and on the recent polynomial Wolff axioms of Katz-Rogers, cited as an external input. The grains decomposition and the nested-variety inequality are new and proven in the paper.

assumptions (9)

standard math Yomdin-Gromov algebraic lemma (Theorem 2.4) parameterizes compact semi-algebraic sets by maps with bounded derivatives.
Used in proof of Lemma 2.6 to control how much tubes can expand inside a semi-algebraic set; cited to Burguet [6].
standard math Wongkew's tubular neighborhood bound (Theorem 2.1): |N_rho(Z∩B)| ≤ C(n,E) rho^{n-d} r^d for 0<rho≤r.
Used to establish the growth condition (2.29) for algebraic varieties; cited to Wongkew [26].
standard math Milnor-Thom bound (Theorem 2.2) gives at most E(2E-1)^{n-1} connected components for a real variety of degree ≤ E.
Used in Lemma 2.6 and in digitization arguments; cited to Milnor [23].
standard math Basu-Pollack-Roy bound (Theorem 2.3) bounds the number of connected components of semi-algebraic sets of bounded complexity.
Used to control complexity in grain decompositions; cited to [1].
domain assumption Katz-Rogers polynomial Wolff axioms (Theorem 1.7): #{T ∈ T: |T∩S| ≥ r|T|} ≤ C(n,E,epsilon)|S| delta^{1-n-epsilon} r^{-n}.
Stated without proof and cited to [17]; load-bearing for Lemma 2.11 and hence Theorem 1.9. This is a deep external result.
standard math Polynomial partitioning theorem of Guth-Katz and Guth (Propositions 3.6, 3.7, Lemma 3.9) partitions point sets by low-degree polynomials.
Used to construct the grains tree in Section 3; cited to [11, 12].
standard math Bennett-Carbery-Tao multilinear Kakeya theorem (Theorem 1.2), with endpoint case by Guth, gives an L^{k/(k-1)} bound for products of k arbitrary tubes.
Used in Section 4 to bound grain interactions; cited to [2, 7, 9].
standard math Bourgain-Guth multilinear-to-linear Kakeya reduction (Lemma 1.4) converts a k-linear direction-separated estimate into a linear Kakeya maximal bound, provided d ≤ n-k+2.
The bridge between Theorem 1.3 and Theorem 1.5; cited to [5, 15].
standard math Selection lemma (Lemma 2.5) chooses one representative from each fiber of a semi-algebraic map with controlled complexity.
Used in proof of Lemma 2.6; cited to author's prior work [27, Lemma 2.6] and [17, Lemma 2.2].

how reviews work

0 comments

Cite this review

Pith. "Pith review of New Kakeya estimates using Gromov's algebraic lemma." pith.science (2026). https://pith.science/paper/Z7ADHWX5

@misc{pith2026190805314,
  author       = {Pith},
  title        = {Pith review of: New Kakeya estimates using Gromov's algebraic lemma},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Z7ADHWX5}},
  note         = {Machine review of arXiv:1908.05314}
}

abstract

This paper presents several new results related to the Kakeya problem. First, we establish a geometric inequality which says that collections of direction-separated tubes (thin neighborhoods of line segments that point in different directions) cannot cluster inside thin neighborhoods of low degree algebraic varieties. We use this geometric inequality to obtain a new family of multilinear Kakeya estimates for direction-separated tubes. Using the linear / multilinear theory of Bourgain and Guth, these multilinear Kakeya estimates are converted into Kakeya maximal function estimates. Specifically, we obtain a Kakeya maximal function estimate in $\mathbb{R}^n$ at dimension $d(n) = (2-\sqrt{2})n + c(n)$ for some $c(n)>0$. Our bounds are new in all dimensions except $n=2,3,4,$ and $6$.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 27 canonical work pages

[1]

S. Basu, R. Pollack, and M.-F. Roy. On the number of cells d eﬁned by a family of polynomials on a variety. Mathematika, 43:120–126, 1996

work page 1996
[2]

Bennett, A

J. Bennett, A. Carbery, and T. Tao. On the multilinear res triction and Kakeya conjectures. Acta Math., 196:261–302, 2006

work page 2006
[3]

Besicovitch

A. Besicovitch. Sur deux questions d’integrabilite des fonctions. J. Soc. Phys. Math. , 2:105– 123, 1919

work page 1919
[4]

Bochnak, M

J. Bochnak, M. Coste, and M.-F. Roy. Real algebraic geometry. Springer-Verlag, 1998

work page 1998
[5]

Bourgain and L

J. Bourgain and L. Guth. Bounds on oscillatory integral o perators based on multilinear esti- mates. Geom. Funct. Anal., 21:1239–1295, 2011. 33

work page 2011
[6]

D. Burguet. A proof of Yomdin-Gromov’s algebraic lemma. Israel J. Math. , 168:291–316, 2008

work page 2008
[7]

Carbery and S

A. Carbery and S. I. Valdimarsson. The endpoint multilin ear Kakeya theorem via the Borsuk- Ulam theorem. J. Funt. Anal., 264:1643–1663, 2012

work page 2012
[8]

A. Cordoba. The Kakeya maximal function and the spherica l summation multipliers. Am. J. Math., 99:1–22, 1977

work page 1977

Show all 27 references

[9]

L. Guth. The endpoint case of the Bennett-Carbery-Tao mu ltilinear Kakeya conjecture. Acta Math., 205:263–286, 2010

2010
[10]

L. Guth. A restriction estimate using polynomial parti tioning. J. Amer. Math. Soc. , 29:371– 413, 2016

2016
[11]

L. Guth. Restriction estimates using polynomial parti tioning II. Acta Math., 221:81–142, 2018

2018
[12]

Guth and N

L. Guth and N. Katz. On the Erd˝ os distinct distances pro blem in the plane. Ann. of Math. , 181:155–190, 2015

2015
[13]

Guth and J

L. Guth and J. Zahl. Polynomial Wolﬀ axioms and Kakeya-type estimates in R4. Proc. London Math. Soc., 117:192–220, 2018

2018
[14]

Hickman and K

J. Hickman and K. M. Rogers. Improved Fourier restricti on estimates in higher dimensions. Camb. J. Math. , 7:219–282, 2019

2019
[15]

Hickman and K

J. Hickman and K. M. Rogers. New Kakeya estimates using t he polynomial Wolﬀ axioms. arXiv e-prints, (arXiv:1901.01802), 2019

1901 arXiv
[16]

Hickman and J

J. Hickman and J. Zahl. A note of Fourier restriction and nested polynomial Wolﬀ axioms. arXiv e-prints, (arXiv:2010.02251), 2020

2010 arXiv
[17]

Katz and K

N. Katz and K. Rogers. On the polynomial Wolﬀ axioms. Geom. Funct. Anal., 28:1706–1716, 2018

2018
[18]

Katz and T

N. Katz and T. Tao. Recent progress on the Kakeya conject ure. Publ. Mat., 46:161–179, 2002

2002
[19]

Katz and J

N. Katz and J. Zahl. An improved bound on the Hausdorﬀ dim ension of Besicovitch sets in R3. J. Amer. Math. Soc. , 32:195–259, 2019

2019
[20]

Katz and J

N. Katz and J. Zahl. A Kakeya maximal function estimate i n four dimensions using plane- brushes. to appear, Rev. Mat. Iberoam., (also in arXiv:1902.00989), 2019

1902
[21]

K. Kuttler. Modern Analysis. CRC Press, 1998

1998
[22]

K. Kuttler. Real and Abstract Analysis. https://math.byu.edu/ klkuttle/541book.pdf, 2019

2019
[23]

J. Milnor. On the Betti numbers of real varieties. Proc. Amer. Math. Soc., 15:275–280, 1964

1964
[24]

T. Wolﬀ. An improved bound for Kakeya type maximal functions. Rev. Mat. Iberoam., 11:651– 674, 1995

1995
[25]

T. Wolﬀ. Recent work connected with the Kakeya problem. I n Prospects in mathematics (Princeton, NJ), pages 129–162, 1996. 34

1996
[26]

Volumes of tubular neighbourhoods of real alg ebraic varieties

Wongkew. Volumes of tubular neighbourhoods of real alg ebraic varieties. Paciﬁc J. Math. , 159:177–184, 1993

1993
[27]

J. Zahl. A discretized Severi-type theorem with applic ations to harmonic analysis. Geom. Funct. Anal., 28:1131–1181, 2018. 35

2018

Pith tools

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

[1] [1]

S. Basu, R. Pollack, and M.-F. Roy. On the number of cells d eﬁned by a family of polynomials on a variety. Mathematika, 43:120–126, 1996

work page 1996

[2] [2]

Bennett, A

J. Bennett, A. Carbery, and T. Tao. On the multilinear res triction and Kakeya conjectures. Acta Math., 196:261–302, 2006

work page 2006

[3] [3]

Besicovitch

A. Besicovitch. Sur deux questions d’integrabilite des fonctions. J. Soc. Phys. Math. , 2:105– 123, 1919

work page 1919

[4] [4]

Bochnak, M

J. Bochnak, M. Coste, and M.-F. Roy. Real algebraic geometry. Springer-Verlag, 1998

work page 1998

[5] [5]

Bourgain and L

J. Bourgain and L. Guth. Bounds on oscillatory integral o perators based on multilinear esti- mates. Geom. Funct. Anal., 21:1239–1295, 2011. 33

work page 2011

[6] [6]

D. Burguet. A proof of Yomdin-Gromov’s algebraic lemma. Israel J. Math. , 168:291–316, 2008

work page 2008

[7] [7]

Carbery and S

A. Carbery and S. I. Valdimarsson. The endpoint multilin ear Kakeya theorem via the Borsuk- Ulam theorem. J. Funt. Anal., 264:1643–1663, 2012

work page 2012

[8] [8]

A. Cordoba. The Kakeya maximal function and the spherica l summation multipliers. Am. J. Math., 99:1–22, 1977

work page 1977