{"id":"3df6d0b5-9f08-4f24-b893-65e7f6da9780","arxiv_id":"2506.18142","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":1,"one_line_summary":"A measure-zero, non-sticky Kakeya set exists in R^2, and non-trivial sticky and non-sticky measure-zero Kakeya sets exist in R^d.","lead":"This paper constructs a Kakeya set, a set that contains a unit line segment in every direction, that is also of Lebesgue measure zero and non-sticky. It also provides the first non-trivial high-dimensional examples of such sets, showing that a natural conjecture about sticky and non-sticky Kakeya sets is false.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 3.3(3) relies on condition (M) for the zero-measure projection estimate, but the paper never proves that (M) is compatible with (*)_k, (**)_k, and (***)_k; the measure-zero conclusion is therefore conditional on an unstated combinatorial existence result.","rationale":"The reader's weakest assumption identifies the same point: Proposition 3.3(3) depends on condition (M), whose compatibility with the other construction conditions is not demonstrated. This is indeed the most load-bearing concern because the zero-measure property of the Kakeya set is obtained only through the zero-projection estimate, and that estimate is exactly where the unproved (M) is used. The rest of the construction, including the Hausdorff and packing dimension computations and the passage from C to a Kakeya set via Proposition 2.3, is sound. I do not see an internal inconsistency or a contradiction with known results; the issue is a missing existence argument for the invariant pattern. The gap is likely repairable, since small cases admit such patterns by hand, but as written the proof of Proposition 3.3(3) is incomplete. The higher-dimensional sections inherit this issue when they reuse C, but the primary theorem is unaffected beyond the (M) gap. Therefore the reader's CONDITIONAL verdict should stand; no stronger objection is warranted.","tokens_in":11409,"tokens_out":42504,"duration_ms":378900,"concrete_test":"For a single repeat with n_{2k-1}=n_{2k}=n, explicitly construct a translation-invariant pattern P_n on the 4^n by 4^n grid such that: (i) the odd-stage selection has exactly 2^n squares in every column; (ii) the even-stage selection places at least one subsquare in every column of width 4^{-2n}; (iii) both the +45-degree and -45-degree projections contain an exactly overlapping pair of subsquares. Verify this construction for all n>=1, or at least for n=1,2,3 by exhaustive search. If such P_n exists for all n, the (M) compatibility is established and the proof of Proposition 3.3(3) can be repaired by inserting the explicit pattern. If no such pattern exists for some n, the measure-zero claim is unsupported and Theorem 1.2 is in jeopardy.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The zero-measure conclusion in Theorem 1.2 passes through Proposition 3.3(3), where m_1(pi_+(C))=0 is proved by an inductive covering argument. The argument multiplies the number of projected intervals by (4^{n1+n2}-1) at each repetition, and that multiplication is justified only if the pattern of subsquares inside every parent square at the even stage is identical, i.e. if condition (M) holds. The paper introduces (M) as an 'additional condition' and, in Proposition 3.3, simply states 'we will require C to satisfy (M)', but it never shows that the chosen sequence n_k admits a choice of subsquares satisfying (M) together with the column condition (*)_k, the column-covering condition (**)_k, and the exact-overlap condition (***)_k at every stage. These are simultaneous finite combinatorial constraints; for instance, a pattern that trivially satisfies (*)_k and (**)_k (bottom rows with cyclic column assignments) can fail (***) for one of the two diagonal projections because the required parity assignments prevent an exactly overlapping pair. If no pattern satisfying all three conditions and identical across parents exists, the bound m_1(pi_+(C)) <= sqrt(2) * prod_j (1 - 4^{-(n_{2j-1}+n_{2j})})^{m_j} collapses, and the Kakeya set may fail to have Lebesgue measure zero. The gap is not a disagreement with consensus; it is an omitted existence proof for a crucial combinatorial object, and it directly affects the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper constructs a compact Cantor-type set C in [0,1]^2 with pi_0(C)=[0,1], H^1(C) finite and positive, packing dimension strictly larger than 1, and two diagonal projections of zero Lebesgue measure. It then invokes a criterion (Proposition 2.3) built on the Besicovitch projection theorem and Fubini to convert C into a non-sticky Kakeya set of Lebesgue measure zero in R^2; a product with [0,1]^{d-2} gives the same in higher dimensions. The paper also gives two genuinely higher-dimensional constructions: a sticky Kakeya set from products of the four-corner Cantor set (Theorem 4.2) and a non-sticky one from products of C (Theorem 4.3), both of Hausdorff dimension d and not formed by a trivial Cartesian product with a line.","tokens_in":11600,"tokens_out":36292,"duration_ms":345932,"significance":"The main result, if fully justified, resolves a natural question: measure-zero Kakeya sets need not be sticky, so the sticky and non-sticky cases are not distinguished by Lebesgue measure. The construction is explicit and the proof strategy is attractive, using standard tools (Besicovitch projection, Marstrand, product inequalities). The paper is careful in many places: Proposition 2.4 gives a self-contained lower bound for packing dimension of fractal cubes, and the higher-dimensional examples in Section 4 address a gap in the literature. The central issue is that the proof of the zero-measure projection claim rests on an unproved combinatorial existence statement for condition (M). This is likely fixable, but until it is supplied the main theorem is conditional.","major_comments":[{"comment":"The proof that m_1(pi_+(C))=m_1(pi_-(C))=0 requires the Cantor set to satisfy condition (M), because the product bound (1 - 4^{-(n_{2j-1}+n_{2j})})^{m_j} in the covering argument is obtained by assuming that the even-stage pattern inside every parent square is identical, so that the exact-overlap condition (***)_k provides a reduction in every parent. Condition (M) is introduced only as an 'additional condition', and the paper never proves that the sequence n_k chosen in Proposition 3.3 admits a construction satisfying (M) simultaneously with (*)_k, (**)_k, and (***)_k at every stage and every repetition. This is a load-bearing gap: the conclusions m_1(pi_+(C))=m_1(pi_-(C))=0, and hence the Lebesgue measure zero of the Kakeya set via Proposition 2.3, depend on it. Please add a lemma that explicitly constructs the uniform pattern and verifies all required conditions.","section":"Section 3, condition (M) and Proposition 3.3(3)"},{"comment":"The step from (***)_1 to 'pi_+(C) is contained in at most 4^{n1+n2}-1 intervals of length sqrt(2)*4^{-(n1+n2)}' uses the fact that, because n1=n2, each repetition selects exactly one square in each fine column, so that without the duplicate pair the projection would occupy at most 4^{n1+n2} distinct diagonal classes. This one-per-column fact is never stated or proved; if several squares were allowed per column, a single duplicate would not in general reduce the covering number by one. Please make this counting explicit, since it is essential to the product estimate.","section":"Section 3, Proposition 3.3(3), counting step"}],"minor_comments":[{"comment":"The word 'Haudorff' should be 'Hausdorff'.","section":"Theorem 4.3 statement"},{"comment":"In the displayed computation, the term m_k(n_{2k-1}+n_{2k}) appears to be missing the factor 3; the following line uses m_k(3n_{2k-1}+n_{2k}). Please correct the typo.","section":"Section 3, Proposition 3.3(2), displayed formula"},{"comment":"The stickiness conclusion needs dim_P(A_B)=d-1, but the text states only the Hausdorff dimension of A_B. Since A_B is bi-Lipschitz equivalent to C_{d-1} x C_{d-1} and dim_P(C0)=dim_H(C0)=1/2, the product inequality (2.2) gives the required packing dimension; please state this explicitly.","section":"Theorem 4.2, stickiness conclusion"},{"comment":"The spelling 'Bescovitch' should be 'Besicovitch'.","section":"Abstract and throughout"},{"comment":"Condition (M) is formulated for C_{2k} with k>=1; when the first pair (n1,n2) is repeated, the same uniformity is needed for the patterns inside each element of C_2. Please make the indexing uniform or add a sentence covering the initial repetition.","section":"Section 3, condition (M)"}],"recommendation":"major_revision","confidential_remarks":"The paper is likely correct and the missing condition-(M) lemma is probably straightforward, but as written the main theorem is conditional. If the authors provide a rigorous construction of the uniform pattern, I would support publication. The higher-dimensional results are a nice addition, and I do not see concerns about novelty or citation practice."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper's central object is genuinely new—an explicit measure-zero non-sticky Kakeya set in R^2, plus higher-dimensional examples not obtained by Cartesian product. The construction in Section 3 is clean and the paper is well written. My main concern is the same as your stress-test: Proposition 3.3(3) needs condition (M) for the projection estimate, and compatibility of (M) with (*), (**), (***) is never proved. This is a real gap, not a nitpick, because the measure-zero conclusion hangs directly on it. The paper says (***) 'can easily be achieved' and Lemma 3.1 only proves (**) follows from (*) by counting; it never shows a single pattern satisfying all three together, let alone one that repeats identically in every parent square. So the proof of Theorem 1.2 is conditional on an unstated combinatorial existence claim. I don't think the claim is false—the examples for n1=n2=1 make it plausible—but it needs an actual argument.\n\nThe rest of the paper is solid. The dimension computations for the Cantor set (Hausdorff dimension 1, packing dimension >1) are correct, and the route from C to a Kakeya set via Proposition 2.3 and the Besicovitch projection theorem is standard. The high-dimensional examples in Section 4 are a nice addition; the sticky example's computation in Theorem 4.2 says 'Hausdorff dimension' where it should say 'packing dimension' for stickiness, but the packing dimension follows from the same bi-Lipschitz equivalence, so that is minor.\n\nThis paper is for people working on Kakeya constructions and fractal geometry. I would send it to peer review: the construction is important enough and the gap is likely fixable. The referee should ask for a rigorous existence proof for the patterns satisfying (M) and all inductive conditions, or for a projection estimate that avoids (M). If the gap is repaired, this is a citeable contribution. My recommendation: engage with it, and request a revision.","headline":"Genuinely new construction with a real but repairable gap in Proposition 3.3(3); deserves serious peer review.","tokens_in":12252,"tokens_out":14670,"would_cite":false,"duration_ms":125335,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["28A78","28A80"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper constructs a compact Kakeya set in R^2 of Lebesgue measure zero that is non-sticky, and extends this to all higher dimensions.","keywords":["Kakeya set","Besicovitch set","Lebesgue measure zero","sticky Kakeya sets","non-sticky Kakeya sets","packing dimension","Hausdorff dimension","Besicovitch projection theorem"],"falsifier":"Run the induction of Proposition 3.3 with $n_{2k-1}=n_{2k}$ and check at the first even stage whether condition (M) can be met without emptying a column in condition $(**)_1$; if the identical-pattern requirement forces an empty column, the estimate $m_1(\\pi_+ C)\\le \\sqrt2\\,\\prod_j(1-4^{-(n_{2j-1}+n_{2j})})^{m_j}$ fails and the proof of Proposition 3.3(3) collapses.","tokens_in":11083,"feed_emoji":"📐","tokens_out":9662,"duration_ms":90626,"temperature":0.7,"pith_summary":"A Kakeya set contains a unit line segment in every direction. The paper's main result is a compact Kakeya set in $\\mathbb{R}^2$ of Lebesgue measure zero that is non-sticky, meaning its line-parameter set has packing dimension strictly above $1$. This matters because measure-zero Kakeya sets are the hard case for the Kakeya set conjecture, and the sticky/non-sticky distinction is central to the recent resolution in $\\mathbb{R}^3$; if non-stickiness forced positive measure, the restricted sticky conjecture and the full conjecture would be equivalent. The paper shows that route is blocked: non-sticky examples can have zero area. Section 4 then builds sticky and non-sticky measure-zero Kakeya sets in every dimension that are not Cartesian products of a planar example with $\\mathbb{R}^{d-2}$ and have full Hausdorff dimension $d$.","feed_headline":"Zero-measure Kakeya sets can be non-sticky","feed_subtitle":"A Cantor-set construction yields a direction-covering set of area zero whose line parameters have packing dimension above 1.","key_machinery":"The central object is a Cantor set $C$ built from fractal squares with side lengths $4^{-(n_1+\\cdots+n_k)}$ and two alternating selection rules: odd stages keep $2^{3n}$ squares per parent with exactly $2^n$ per column, while even stages keep $2^n$ squares per parent with at least one per subcolumn and a pair overlapping exactly under the $45^\\circ$ projections. A uniform-pattern condition (M) makes $C$ a uniform fractal cube and supplies the product estimate for the diagonal projections. Proposition 2.3 is the bridge: any compact parameter set $C$ whose first-coordinate projection contains an interval, whose packing dimension exceeds $1$, which has finite one-dimensional Hausdorff measure, and which has two distinct zero-measure projections, yields a non-sticky measure-zero Kakeya set through the line family $K_0$ and the Besicovitch projection theorem.","core_discovery":"The central claim is Theorem 1.2: there exists a non-sticky Kakeya set of Lebesgue measure zero in $\\mathbb{R}^2$, and hence in $\\mathbb{R}^d$ for every $d>2$. The proof builds a Cantor set $C\\subset[0,1]^2$ that has positive finite one-dimensional Hausdorff measure, packing dimension greater than $1$, and zero Lebesgue measure for both diagonal projections. Using $C$ as the parameter set of affine lines, the union of the corresponding unit segments, after finitely many rotations, is a compact Kakeya set; because the parameter set has packing dimension greater than $1$, the set is non-sticky, and by the Besicovitch projection theorem together with Fubini's theorem, the set has measure zero. The paper also proves Theorem 4.2 and Theorem 4.3: in $\\mathbb{R}^d$ there are sticky and non-sticky Kakeya sets of measure zero and Hausdorff dimension $d$ that are not formed by taking a Cartesian product with $\\mathbb{R}^{d-2}$.","pith_inferences":["A natural next step is to test whether condition (M) can be weakened or dropped; if the diagonal-projection estimate can be obtained without identical patterns in every parent square, the construction would become substantially more flexible.","The Proposition 2.3 recipe suggests a general search: any compact family of parameter sets in $[0,1]^2$ with packing dimension above $1$ and two zero-measure projections automatically yields a non-sticky measure-zero Kakeya set, so the problem reduces to producing such Cantor sets.","One could investigate whether a single zero-measure projection plus a lower entropy bound suffices in place of the two diagonal projections, which would shorten the route from fractal-square data to a non-sticky measure-zero Kakeya set."],"forward_implications":["In $\\mathbb{R}^2$, there is a compact Kakeya set of Lebesgue measure zero whose line-parameter set has packing dimension strictly above $1$, so non-stickiness does not force positive measure.","Taking products with $\\mathbb{R}^{d-2}$ gives non-sticky measure-zero Kakeya sets in every dimension $d>2$.","Theorem 4.2 provides a sticky Kakeya set in $\\mathbb{R}^d$ of measure zero and Hausdorff dimension $d$ that is not a Cartesian product of a planar Kakeya set with $\\mathbb{R}^{d-2}$.","Theorem 4.3 provides the analogous non-sticky example in $\\mathbb{R}^d$, also of measure zero and Hausdorff dimension $d$, so both high-dimensional constructions satisfy the full-dimension conclusion of the Kakeya set conjecture without being trivial products."],"supporting_citations":[{"why":"It supplies the Besicovitch projection theorem, which converts two zero-measure diagonal projections into almost-everywhere zero-measure slices in Proposition 2.3.","marker":"[5]"},{"why":"It gives the comparison between packing dimension and upper entropy dimension of a measure, used in Proposition 2.4 to lower-bound the packing dimension of the fractal cubes.","marker":"[6]"},{"why":"It provides the product dimension inequalities and Marstrand's projection and slicing theorems used for the higher-dimensional examples in Section 4.","marker":"[1]"},{"why":"It introduces the sticky/non-sticky distinction and the sticky Kakeya conjecture that frame Theorem 1.2 as a counterexample to the expected equivalence.","marker":"[14]"},{"why":"It provides the classical four-corner Cantor construction of a measure-zero Kakeya set and the lines-between-two-Cantor-sets viewpoint used in Section 4.1.","marker":"[11]"}],"fun_headline_variants":["Measure-zero Kakeya sets need not be sticky","Explicit non-sticky Kakeya set of area zero","Non-sticky Kakeya set with Lebesgue measure zero","Cantor-set based non-sticky Kakeya set of measure zero"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof that the diagonal projections have zero measure rests on the requirement that every parent square at each even construction stage use the same subsquare pattern, and the paper does not verify that the chosen block sizes can meet that requirement together with the other construction conditions.","fun_headline_variants_meta":{"raw":{"variants":["Measure-zero Kakeya sets need not be sticky","Explicit non-sticky Kakeya set of area zero","Non-sticky Kakeya set with Lebesgue measure zero","Cantor-set based non-sticky Kakeya set of measure zero"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000713,"raw_usage":{"total_tokens":3198,"prompt_tokens":930,"completion_tokens":2268,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":546,"completion_tokens_details":{"reasoning_tokens":2196}},"tokens_in":546,"tokens_out":2268,"duration_ms":16889,"temperature":1.0,"reasoning_tokens":2196,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T18:56:09.268689+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the induction of Proposition 3.3 with $n_{2k-1}=n_{2k}$ and check at the first even stage whether condition (M) can be met without emptying a column in condition $(**)_1$; if the identical-pattern requirement forces an empty column, the estimate $m_1(\\pi_+ C)\\le \\sqrt2\\,\\prod_j(1-4^{-(n_{2j-1}+n_{2j})})^{m_j}$ fails and the proof of Proposition 3.3(3) collapses.","supporting_citations":[{"cited_title":"Number 85","cited_arxiv_id":null,"evidence_quote":"It supplies the Besicovitch projection theorem, which converts two zero-measure diagonal projections into almost-everywhere zero-measure slices in Proposition 2.3."},{"cited_title":"Relationships between different dimensions of a mea- sure.Monatsh","cited_arxiv_id":null,"evidence_quote":"It gives the comparison between packing dimension and upper entropy dimension of a measure, used in Proposition 2.4 to lower-bound the packing dimension of the fractal cubes."},{"cited_title":"American Mathematical Society, Providence, RI, [2023].doi:10.1090/surv/276","cited_arxiv_id":null,"evidence_quote":"It provides the product dimension inequalities and Marstrand's projection and slicing theorems used for the higher-dimensional examples in Section 4."},{"cited_title":"Stein and Rami Shakarchi.Real analysis, volume 3 ofPrinceton Lectures in Analy- sis","cited_arxiv_id":null,"evidence_quote":"It provides the classical four-corner Cantor construction of a measure-zero Kakeya set and the lines-between-two-Cantor-sets viewpoint used in Section 4.1."}],"review_version":2}