REVIEW 5 minor
Robust Repulsion for Growing Crowns in Linear Hypergraphs
T0 review · 0 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash
Pith's one-line read In a crown-free linear hypergraph, an edge attaining the local reciprocal-degree bound is exactly a line shared by $k-1$ projective planes, and neighboring edges are repelled to lower degrees.
desk verdict A genuine structural result: equality forces glued projective planes, and repulsion yields a strict but modest global coefficient gap. 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 engine is the local crown system around a fixed edge $e=\{v_1,\dots,v_r\}$: for each $i$, the petal family $F_i(e)$ consists of the $(r-1)$-sets $f\setminus\{v_i\}$ over all edges $f\ne e$ through $v_i$. Linearity makes each $F_i(e)$ a matching, and a $C^r_{1,k}$-crown with base $e$ is exactly a rainbow matching of size $k$ in these $r$ color classes. The matching-blocker inequality bounds the total unused-color petal mass by incidences on a maximum rainbow matching. At equality every color class has size $(k-1)(r-1)$, so a perfect blocker forces the system to split into $t$ affine planes of order $r-1$; adjoining the base vertices completes them to $t$ projective planes glued along $e$. Sharp-state repulsion then follows from the fact that a line in one constituent plane can meet any other edge in at most one point.
What would settle it
Construct a linear $r$-uniform $k$-crown-free hypergraph with a sharp edge $e$ whose vertices all have degree $D=(k-1)(r-1)+1$ and whose neighborhood is not $k-1$ projective planes of order $r-1$ glued along $e$; for instance, $r=4$, $k=3$, $D=7$, $q=3$ already tests the classification. No such construction can exist if the classification theorem is true.
Extended reading notes
Core claim
The central discovery is a finite-geometric classification of equality in the local reciprocal-degree bound. For a linear $C^r_{1,k}$-free $r$-graph $H$ and an edge $e$, the quantity $\Phi_H(e)=\sum_{v\in e}1/d_H(v)$ is always at least $r/D$ with $D=(k-1)(r-1)+1$; the paper proves that equality holds exactly when $e$ and every edge meeting it form $t=k-1$ projective planes of order $q=r-1$ whose common intersection is precisely the line $e$. In other words, the petals through the vertices of $e$ split into $t$ affine planes of order $q$, and adjoining the base line completes each to a projective plane. No Desarguesian hypothesis is needed for this rigidity. The same structure then repels: any adjacent edge must contain many vertices of degree at most $D^-=(k-2)(r-1)+1$, and this local gap accumulates into a strict improvement over the previous global edge-count coefficient $D/r$.
Load-bearing premise
The load-bearing premise is that the hypergraph is linear, meaning any two edges share at most one vertex; if two edges could share two vertices, the petal families would not be matchings, the trace skeleton would fail, and the projective-plane classification and repulsion proof would not go through.
Editorial extensions
If this is right
- Sharp edges are pairwise disjoint, so equality neighborhoods cannot overlap through a vertex.
- Every edge meets at most $k-2$ sharp edges, and the number of low-degree vertices is bounded below by $|V^-(H)|\ge \frac{r(D-1)(q-t+2)}{(t-1)D^-}|S(H)|$.
- For fixed $k$ and all sufficiently large $r$, every such hypergraph satisfies $|E(H)|\le \frac{D(D-1)}{r(D-1)+1}|V(H)|$, improving the coefficient $D/r$ by $r^{-2}+O_k(r^{-3})$.
- There is no nonempty $D$-regular linear $C^r_{1,k}$-free $r$-graph; more generally $r|E(H)|\le D|V(H)|-c(H)$, where $c(H)$ is the number of edge-containing components.
- For $k=3$, every connected component containing a sharp edge is exactly two projective planes of order $r-1$ glued along that edge, with $r+2(r-1)^2$ vertices and $1+2r(r-1)$ edges.
Reading between the lines
- The colored-matching argument is a template that should transfer to other acyclic configurations, such as uniform hypertrees or linear paths, where equality in a local degree bound might also force a rigid finite geometry.
- Because the equality classification uses no Desarguesian assumption, the sharp-state construction can be realized from any projective plane, including non-Desarguesian ones; probing a non-Desarguesian order such as $q=9$ would test whether the repulsion constant depends on the plane's inner structure.
- Theorem 7.3 reduces the asymptotic problem to an average defect estimate on neighbors of near-sharp edges; if the paper's Conjecture 7.5 holds, the edge-density limsup drops below $k-1$ for every fixed $k$, and the conjecture is likely easiest to test first for $t=2$, where the sharp component is just two glued planes.
- The nonexistence of sharp states when no projective plane of order $r-1$ exists suggests that extremal density in crown-free hypergraphs may be sensitive to the arithmetic of $r-1$, not only to the shape of the forbidden configuration.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. This paper studies linear r-uniform hypergraphs that contain no copy of the k-crown C^r_{1,k}. The main local result is the reciprocal-degree inequality Φ_H(e) ≥ r/D for every edge e, where D = (k-1)(r-1)+1. The paper classifies all equality cases: e is sharp exactly when e and every edge meeting it form t = k-1 projective planes of order q = r-1 whose common intersection is precisely the line e. Around this classification it proves a repulsion theorem (every edge meeting a sharp edge contains at least q-t+2 vertices of degree at most D^-), a strengthened quadratic packing bound, the nonexistence of nonempty D-regular crown-free linear hypergraphs, and unconditional global coefficient gaps strictly smaller than D/r. It also develops a quantitative near-sharp packet localization theorem and a conditional constant-gap closure based on an explicitly stated conjecture. The proofs combine a colored matching reformulation, a matching-blocker inequality, a greedy degree-witness lemma, and finite-geometric classification lemmas.
Significance. If correct, this is a substantial advance in linear Turán theory. The sharp structural classification of equality in the local reciprocal-degree bound is new and is derived rather than assumed, with no reliance on Desarguesian planes. The repulsion and packing estimates are explicit, and the unconditional improvement over Adak's coefficient D/r is significant. The paper carefully separates the unconditional results from the conditional conjecture-based part, and the near-sharp localization is presented with quantitative bounds. The authors also re-prove the local reciprocal-degree inequality rather than importing it as a black box, which strengthens the reliability of the whole chain.
minor comments (5)
- [§5 (Theorem 1.2)] When Lemma 5.1 is applied to Π_2 to conclude that each chosen edge g_i contains at most one point of Π_2, the proof should explicitly verify that g_i is not a line of Π_2. This is true because g_i contains the affine point w_i of Π_1, whereas Π_2 shares only the line e with Π_1 and w_i is not on e. The same one-line justification is also needed for the earlier application of Lemma 5.1 to Π_1. The claim is correct, but the hypothesis of Lemma 5.1 is currently skipped.
- [§6 (Corollary 1.5)] The sentence 'Summing (28) over E(K) and using (31) within the component' is imprecise: (31) is an equality for the whole hypergraph, whereas for a single component K one only gets ∑_{e∈E(K)} Φ_H(e) = ∑_{v∈V(K)} deg_K(v)/d_H(v) ≤ |V(K)|. This inequality suffices for the argument, but the equality case, which is used to prove strictness, should be spelled out: equality would force every edge of K to be sharp and d_H(v)=deg_K(v) for every vertex v, making K D-regular and contradicting the first part.
- [§2 (Lemma 2.1)] The phrase 'any common point of their outside sets is therefore an additional intersection' is confusing, since two edges from different colors do not share a vertex of e. It should be rephrased to say that a common point of A∈F_i(e) and B∈F_j(e) would be an intersection of the corresponding full edges, and linearity therefore permits at most one such point.
- [§6 (after Eq. (17))] The displayed difference between the coefficients should be D/(r(r(D-1)+1)); the present typesetting appears to show D/r times (r(D-1)+1), which is not the intended quantity. Please correct the formatting.
- [§4 (Lemma 4.2)] In Lemma 4.2, the sentence 'At any stage, at most s+1≤t colors are either already used or forbidden' is correct only if s is understood as the number of selected sets before the current extension step. As written it could be misread, so a short clarification would help.
Circularity Check
No significant circularity: the sharp-state classification and all global consequences are derived from the paper's own lemmas, not from fitted inputs or self-citation.
full rationale
The paper re-proves the local reciprocal-degree inequality in Proposition 3.3 via Lemma 3.1, and Lemma 3.1 is proved directly by greedy petal selection, so the bound r/D is not imported as an unexamined input. The central classification (Theorem 1.1) follows from the equality case of Proposition 3.3, which forces every petal family to have size tq, and then Theorem 4.7 derives the affine-plane structure from the matching-blocker equations; the projective planes are a consequence of the sharpness assumption rather than being assumed. Theorem 1.2 repulsion is proved from the classified neighborhood using Lemma 5.1, and the packing/defect estimates in Theorem 1.3 are double-counting consequences. The external citations to Bruck-Ryser and Lam et al. are used only to establish non-existence of projective planes of certain orders, translating them into coefficient improvements via Theorem 1.7; this is not load-bearing for the main derivation. Conjecture 7.5 is explicitly labeled conjectural, and Theorem 7.6 is conditional on it, so no unproven input is passed off as a derivation. No fitted parameter is renamed as a prediction, and no cited uniqueness theorem is used to forbid alternatives. The manuscript's own limitation statements are respected: the conditional constant-gap result is openly conditional, and the k=3 component theorem is proved rather than assumed. The derivation chain is self-contained given the stated linearity and crown-freeness hypotheses.
Assumptions & free parameters
assumptions (3)
- domain assumption Two distinct edges of H meet in at most one vertex (linearity).
- standard math An affine plane of order q can be completed to a projective plane of order q by adding one point for each parallel class.
- standard math Bruck-Ryser theorem: if q is 1 or 2 modulo 4 and q is not a sum of two squares, no projective plane of order q exists; and no projective plane of order 10 exists.
Cite this review
Pith. "Pith review of Robust Repulsion for Growing Crowns in Linear Hypergraphs." pith.science (2026). https://pith.science/paper/NGB3IDD5
@misc{pith2026260801568,
author = {Pith},
title = {Pith review of: Robust Repulsion for Growing Crowns in Linear Hypergraphs},
year = {2026},
howpublished = {\url{https://pith.science/paper/NGB3IDD5}},
note = {Machine review of arXiv:2608.01568}
}
abstract
Put $q=r-1$, $t=k-1$, and $D=tq+1$. For an edge $e$ of a linear $C^r_{1,k}$-free $r$-uniform hypergraph, define \[ \delta_H(e)=\sum_{v\in e}\frac{1}{d_H(v)}-\frac{r}{D}. \] The defect satisfies $\delta_H(e)\ge 0$. At equality, every vertex of $e$ has degree $D$, and the petal trace at $e$ is a disjoint union of $t$ affine planes of order $q$. Equivalently, restoring the base line gives $t$ projective planes of order $q$ with common line $e$. For growing crowns, the equality structure is stable in the following sense. If $q_j\to\infty$, $2\le t_j\le q_j$, and $e_j$ is an edge of a finite linear $C_{1,t_j+1}^{q_j+1}$-free hypergraph satisfying \[ \frac{t_j^2}{q_j}\to0, \qquad t_j^3\delta_{H_j}(e_j)\to0, \] then, for every fixed $0<\theta<1$, \[ \frac{ |\{f\ne e_j:f\cap e_j\ne\varnothing,\ \delta_{H_j}(f)\ge\theta/t_j^2\}| }{(q_j+1)(t_jq_j)} \to1. \] Thus an edge close to equality is adjacent almost entirely to edges with defect of order at least $t_j^{-2}$. A uniform form gives absolute constants $Q_0,\varepsilon_0,c_0>0$ such that, whenever $q\ge Q_0t^2$ and $\delta_H(e)<\varepsilon_0/t^3$, at least $\tfrac12(q+1)tq$ neighbors of $e$ have defect at least $1/(100t^2)$. Consequently, \[ c^{\mathrm{lin}}_{q+1,t+1}\le t-\frac{c_0}{t}, \] where $c^{\mathrm{lin}}_{r,k}$ denotes the asymptotic linear Tur\'an coefficient for $C^r_{1,k}$.
Reviewed August 15, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.