On a conjecture regarding the product version of the Hilton-Milner theorem

Lihua Feng; Lu Lu; Qifan Wang; Xucheng Bu; Zejun Huang

A two-center construction disproves the product Hilton-Milner conjecture for n linear in k, while the conjecture holds for n larger than a quadratic threshold.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-14 16:00 UTC pith:3NZFI2W2

load-bearing objection Clean refutation plus large-n confirmation of the Frankl-Wang product Hilton-Milner conjecture; the two-center construction is the real novelty.

arxiv 2607.06443 v2 pith:3NZFI2W2 submitted 2026-07-07 math.CO

On a conjecture regarding the product version of the Hilton-Milner theorem

Xucheng Bu , Lihua Feng , Zejun Huang , Lu Lu , Qifan Wang This is my paper

classification math.CO MSC 05D0505C65

keywords Hilton-Milner theoremcross-intersecting familiesproduct versionnon-trivial intersecting families2-cover graphsminimal coversextremal set theory

verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The classical Hilton-Milner theorem bounds the size of a non-trivial intersecting family of k-subsets. Frankl and Wang asked for the analogous product maximum when two families of different uniformities are required only to be non-trivial and cross-intersecting, and conjectured that the maximum is always realized by a natural pair of Hilton-Milner-type families. This paper shows that the conjecture is false once n is only a constant multiple of k: a simple two-center construction produces a strictly larger product for every fixed ℓ ≥ 3 and all sufficiently large k, throughout a linear interval 2k+1 ≤ n ≤ (c_ℓ-ε)k. At the same time the paper proves that the same conjecture becomes true as soon as n exceeds a quadratic threshold 100ℓk^{2}; moreover equality is attained only by the Hilton-Milner-type pairs. The argument rests on a careful comparison of the sizes of the 2-cover graphs of the two families and on crude but uniform bounds for their higher-order minimal covers.

Core claim

For every fixed ℓ ≥ 3 and all large enough k the two-center families of Theorem 2.1 satisfy |F||G| > M_k(n,k,ℓ)M_ℓ(n,k,ℓ) throughout the linear range 2k+1 ≤ n ≤ (c_ℓ-ε)k, so Conjecture 1.5 fails; conversely, when n > 100ℓk^{2} every non-trivial cross-intersecting pair satisfies the conjectured product bound, with equality only for the Hilton-Milner-type configurations.

What carries the argument

The 2-cover graphs P and Q that encode the minimal covers of size 2 of the two families; their cross-intersecting property, degree bounds Δ(P)≤ k, Δ(Q)≤ℓ and the resulting product bound |P||Q|≤ kℓ-ℓ/2 (or the common-center star case) control all subsequent size estimates.

Load-bearing premise

The quadratic threshold n>100ℓk^{2} is chosen large enough to absorb every error term that arises from the crude cover-size bounds; a sharper analysis of the same cover graphs could lower the threshold.

What would settle it

For a concrete triple such as ℓ=3, k=20 and n=50 (which lies inside the claimed counter-example interval), compute the exact product |F||G| of the two-center construction and compare it with M_k M_ℓ; if the inequality is reversed, the linear-range claim fails.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Desk Editor's Note

Clean refutation plus large-n confirmation of the Frankl-Wang product Hilton-Milner conjecture; the two-center construction is the real novelty.

read the letter

The paper does two concrete things: it kills Conjecture 1.5 of Frankl-Wang for every fixed ℓ≥3 in a linear range of n (roughly 2k to c_ℓ k), and it proves the same conjecture once n exceeds a quadratic threshold 100ℓk^{2}, with a full equality characterization. Both statements are new.

What works well is the two-center construction in Section 2. They take two disjoint (k-1)-sets A,B, form FA={2}∪A and FB={1}∪B, then build F as all k-sets containing {1,2} plus those two, and G as the natural three-part family that forces intersection with both. The size formulas are elementary, the non-triviality check is clean, and the ratio comparison reduces to the simple function R_ℓ(c)=2/c(1-(1-1/c)^ℓ) whose root defines the explicit constant c_ℓ. That is a genuine counter-example family, not a one-off numerical curiosity.

The large-n half is a standard but carefully executed cover-graph argument. They define the 2-cover graphs P and Q, bound their degrees and products (Lemma 3.3 is exhaustive), control higher-order minimal covers by the usual encoding, and feed the three cases into the product comparison against Mk Mℓ. All the binomial ratios and geometric series are tracked with explicit constants; nothing is hand-waved. The only softness is the deliberately loose 100 in the threshold n>100ℓk^{2}, which absorbs the crude |P|≤k^{2}, |Q|≤ℓ^{2} bounds. A tighter analysis could almost certainly lower it, but that does not affect correctness.

Citations are appropriate (EKR, Hilton-Milner, Pyber, Frankl-Wang, Huang). No circularity, no free parameters beyond the explicit constant they chose for convenience. This is solid finite combinatorics that a specialist in intersecting families will want to see. I would send it to referees without hesitation; the math checks out line-by-line.

Referee Report

0 major / 5 minor

Summary. The paper studies the product version of the Hilton–Milner theorem for non-trivial cross-intersecting families F ⊂ binom([n],k) and G ⊂ binom([n],ℓ). Frankl–Wang conjectured that |F||G| ≤ Mk(n,k,ℓ)Mℓ(n,k,ℓ) for n ≥ 2k > 2ℓ ≥ 4, with the bound attained by the natural Hilton–Milner-type pairs. The authors first disprove the conjecture in a linear range: for every fixed ℓ ≥ 3 and all sufficiently large k they exhibit an explicit two-center construction (Theorem 2.1) whose product strictly exceeds Mk Mℓ whenever 2k+1 ≤ n ≤ (cℓ − ε)k (Theorem 2.2), where cℓ = 1/(1−qℓ) and qℓ is the unique root in (1/2,1) of ∑_{i=0}^{ℓ−1} q^i = 2. Second, they prove that the conjecture holds for n > 100ℓk^{2} and 3 ≤ ℓ < k, and completely characterize the extremal pairs as the Hilton–Milner-type configurations (Theorem 3.1). The positive direction proceeds by reducing to maximal pairs, encoding their 2-covers as graphs P and Q, bounding higher-order minimal covers via an injective encoding argument, and comparing the resulting size estimates against the closed-form Hilton–Milner numbers.

Significance. The work settles the status of a recent conjecture of Frankl–Wang by supplying both a robust family of counter-examples (valid for every fixed ℓ ≥ 3 in a positive-density linear interval of n) and a complete equality characterization in the large-n regime. The two-center construction is elementary yet sharp enough to produce an explicit threshold cℓ that can be computed for each ℓ; the large-n proof introduces a clean structural analysis of 2-cover graphs that may be reusable for other product-type EKR problems. All size formulas and asymptotic comparisons are fully explicit, so the results are immediately verifiable and falsifiable. The only free parameter is the deliberately generous constant 100 appearing in the quadratic threshold; this does not affect the qualitative claims.

minor comments (5)

The constant 100 in the hypothesis n > 100ℓk^{2} of Theorem 3.1 (and of Lemmas 3.6–3.7) is chosen only to absorb crude geometric-series bounds. A short remark indicating that the same argument works for any sufficiently large absolute constant, or a brief calculation of a smaller admissible constant, would improve readability without changing the proof.
In the definition of the two-center families (page 3) the sets A,B ⊂ [3,n] are required to be disjoint (k−1)-sets; it would be helpful to note explicitly that such sets exist precisely when n ≥ 2k+1, which is already the standing hypothesis of Section 2.
Lemma 3.3 Case 3 (P contains two disjoint edges) lists several sub-cases for |Q| = m ∈ {1,2,3,4}. The inequalities |P||Q| < kℓ − ℓ/2 are elementary but slightly tedious; a one-line table or a uniform bound |P| ≤ 4 would make the case easier to check.
Typographical: the arXiv identifier of the second Frankl–Wang paper is written “arXiv:2605.09246” in the references; the year 2026 appearing in several citations is presumably a placeholder and should be updated once the final versions appear.
In the display of Rℓ(c) after (2.3) the factor 2/c is written as a fraction; a parenthetical remark that Rℓ(c) > 1 precisely when c < cℓ would make the comparison with the definition of cℓ immediate.

Circularity Check

0 steps flagged

No circularity: explicit two-center construction and cover-graph bounds are self-contained combinatorial arguments.

full rationale

The paper's two main claims are established by direct, elementary calculations that do not reduce to their own inputs. Theorem 2.1 constructs concrete families F and G, verifies cross-intersection and non-triviality by hand, and obtains exact size formulas by partitioning and binomial counting. Theorem 2.2 then forms the ratio of those sizes to the closed-form Hilton-Milner numbers Mk and Mℓ, expands the binomials asymptotically in ck = n/k, and shows that the resulting function Rℓ(c) exceeds 1 precisely when 2 ≤ c < cℓ; the comparison is an ordinary inequality of explicit rational functions of c and ℓ, with no fitted parameters. In the large-n regime, maximality reduces the problem to the 2-cover graphs P and Q; Lemmas 3.2–3.5 bound their degrees, sizes and higher-order minimal covers by purely combinatorial encoding arguments (at most r^s covers of size s, etc.). Lemmas 3.6–3.7 convert those bounds into concrete upper estimates for |F| and |G| that are then compared, case by case, against the lower bounds for Mk Mℓ; every constant (n > 100ℓk^{2}, geometric ratio 2kℓ/n ≤ 1/2, etc.) is written out and verified by elementary arithmetic. Equality characterization follows by observing that the only surviving case forces the families to coincide with the Hilton-Milner-type pairs. No step is definitional of its conclusion, no parameter is fitted and then re-predicted, and the few self-citations are ordinary references to prior statements of the conjecture or classical EKR/Hilton-Milner results; none is load-bearing for the new inequalities. The derivation is therefore free of circularity.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 2 invented entities

The work is pure finite combinatorics. It rests only on the classical binomial-coefficient identities and the definition of cross-intersecting families; no free parameters are fitted and no new physical or combinatorial entities are postulated beyond the explicit two-center construction used for the counter-example.

free parameters (1)

threshold constant 100 in n>100ℓk^{2}
Chosen large enough so that all error terms 8kℓ^{2}/n, 4k^{2}ℓ^{2}/n etc. fall below the main terms; not fitted to data but selected by hand for convenience of the inequalities in Lemmas 3.6-3.7.

axioms (2)

standard math Standard binomial-coefficient identities and the elementary inequalities (1-x)^{2}≥1-2x, Weierstrass product inequality
Used throughout the asymptotic comparisons of Sections 2 and 3.
domain assumption A pair of families is maximal with respect to cross-intersection if no further set can be added while preserving the cross-intersecting property
Standard reduction in extremal set theory; invoked at the beginning of Section 3 to restrict attention to maximal pairs.

invented entities (2)

two-center construction (families F and G of Theorem 2.1) no independent evidence
purpose: Explicit counter-example to Conjecture 1.5 in the linear range of n
Defined by two special k-sets FA, FB and three blocks of ℓ-sets; purely combinatorial, no independent physical existence claimed.
2-cover graphs P and Q no independent evidence
purpose: Encode the minimal 2-element covers of the two families so that degree and intersection properties can be analysed
Standard encoding device in the Hilton-Milner literature; introduced in Section 3.1.

pith-pipeline@v1.1.0-grok45 · 17223 in / 2450 out tokens · 20745 ms · 2026-07-14T16:00:46.655520+00:00 · methodology

0 comments

Cite this review

Pith. "Pith review of On a conjecture regarding the product version of the Hilton-Milner theorem." pith.science (2026). https://pith.science/paper/3NZFI2W2

@misc{pith2026260706443,
  author       = {Pith},
  title        = {Pith review of: On a conjecture regarding the product version of the Hilton-Milner theorem},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3NZFI2W2}},
  note         = {Machine review of arXiv:2607.06443}
}

Share X Bluesky LinkedIn Reddit HN

read the original abstract

Recently, Frankl and Wang considered a product version of the classical Hilton-Milner theorem. They conjectured that, if $\mathcal{F} \subset \binom{[n]}{k}$ and $\mathcal{G} \subset \binom{[n]}{\ell}$ are non-trivial cross-intersecting families with $n \geq 2k > 2\ell \geq 4$, the maximum of $|\mathcal{F}||\mathcal{G}|$ is attained by the natural Hilton-Milner-type configurations. In this paper, we present two main results concerning this conjecture. Firstly, we show that the conjecture does not hold in general. By introducing a two-center construction, we prove that for every fixed integer $\ell \geq 3$ and all sufficiently large $k$, the conjecture is false in a linear range $2k+1 \leq n \leq (c_\ell - \epsilon)k$ for any $0 < \epsilon < c_\ell - 2$, where $c_\ell > 2$ is an explicit constant. Secondly, we prove that the conjecture holds when $n > 100\ell k^2$ and $3 \leq \ell < k$, and we completely characterize the extremal families. Our proofs rely on the size of minimal covers and analyzing the structural properties of $2$-cover graphs.

Review history (2 revisions) →

discussion (0)

Reference graph

Works this paper leans on

13 extracted references · 2 linked inside Pith

[1]

P. Borg, C. Feghali, The maximum sum of sizes of cross-intersecting families of subsets of a set, Discrete Math.345(11) (2022) 112981

2022
[2]

Erd˝ os, C

P. Erd˝ os, C. Ko, R. Rado, Intersection theorems for systems of finite sets,Quart. J. Math. Oxford Ser. (2)12(2) (1961) 313–320

1961
[3]

Frankl, On the maximum of the sum of the sizes of non-trivial cross-intersecting families,Com- binatorica44(1) (2024) 15–35

P. Frankl, On the maximum of the sum of the sizes of non-trivial cross-intersecting families,Com- binatorica44(1) (2024) 15–35

2024
[4]

Frankl, The maximum of the product of non-trivial cross-intersecting 3-graphs,Combinatorics and Number Theory15(1) (2025) 1–8

P. Frankl, The maximum of the product of non-trivial cross-intersecting 3-graphs,Combinatorics and Number Theory15(1) (2025) 1–8

2025
[5]

Frankl, A

P. Frankl, A. Kupavskii, A size-sensitive inequality for cross-intersecting families,European J. Com- bin.62 (2017) 263–271

2017
[6]

Frankl, N

P. Frankl, N. Tokushige, Some best possible inequalities concerning cross-intersecting families,J. Combin. Theory Ser. A61(1) (1992) 87–97

1992
[7]

Frankl, J

P. Frankl, J. Wang, A product version of the Hilton-Milner Theorem,J. Combin. Theory Ser. A 200 (2023) 105791

2023
[8]

Frankl, J

P. Frankl, J. Wang, A product version of the Hilton-Milner Theorem II, (2026) arXiv:2605.09246

Pith/arXiv arXiv 2026
[9]

A. J. W. Hilton, An intersection theorem for a collection of families of subsets of a finite set,J. London Math. Soc.2(3) (1977) 369–376

1977
[10]

A. J. W. Hilton, E. C. Milner, Some intersection theorems for systems of finite sets,Quart. J. Math. 18(1) (1967) 369–384

1967
[11]

Huang, A sharp product bound for non-trivial cross-intersecting families, (2026) arXiv:2606.23322

Y. Huang, A sharp product bound for non-trivial cross-intersecting families, (2026) arXiv:2606.23322

Pith/arXiv arXiv 2026
[12]

Matsumoto, N

M. Matsumoto, N. Tokushige, The exact bound in the Erd˝ os-Ko-Rado theorem for cross-intersecting families,J. Combin. Theory Ser. A52 (1989) 90–97

1989
[13]

Pyber, A new generalization of the Erd˝ os-Ko-Rado theorem,J

L. Pyber, A new generalization of the Erd˝ os-Ko-Rado theorem,J. Combin. Theory Ser. A43(1) (1986) 85–90. 12

1986

[1] [1]

P. Borg, C. Feghali, The maximum sum of sizes of cross-intersecting families of subsets of a set, Discrete Math.345(11) (2022) 112981

2022

[2] [2]

Erd˝ os, C

P. Erd˝ os, C. Ko, R. Rado, Intersection theorems for systems of finite sets,Quart. J. Math. Oxford Ser. (2)12(2) (1961) 313–320

1961

[3] [3]

Frankl, On the maximum of the sum of the sizes of non-trivial cross-intersecting families,Com- binatorica44(1) (2024) 15–35

P. Frankl, On the maximum of the sum of the sizes of non-trivial cross-intersecting families,Com- binatorica44(1) (2024) 15–35

2024

[4] [4]

Frankl, The maximum of the product of non-trivial cross-intersecting 3-graphs,Combinatorics and Number Theory15(1) (2025) 1–8

P. Frankl, The maximum of the product of non-trivial cross-intersecting 3-graphs,Combinatorics and Number Theory15(1) (2025) 1–8

2025

[5] [5]

Frankl, A

P. Frankl, A. Kupavskii, A size-sensitive inequality for cross-intersecting families,European J. Com- bin.62 (2017) 263–271

2017

[6] [6]

Frankl, N

P. Frankl, N. Tokushige, Some best possible inequalities concerning cross-intersecting families,J. Combin. Theory Ser. A61(1) (1992) 87–97

1992

[7] [7]

Frankl, J

P. Frankl, J. Wang, A product version of the Hilton-Milner Theorem,J. Combin. Theory Ser. A 200 (2023) 105791

2023

[8] [8]

Frankl, J

P. Frankl, J. Wang, A product version of the Hilton-Milner Theorem II, (2026) arXiv:2605.09246

Pith/arXiv arXiv 2026

[9] [9]

A. J. W. Hilton, An intersection theorem for a collection of families of subsets of a finite set,J. London Math. Soc.2(3) (1977) 369–376

1977

[10] [10]

A. J. W. Hilton, E. C. Milner, Some intersection theorems for systems of finite sets,Quart. J. Math. 18(1) (1967) 369–384

1967

[11] [11]

Huang, A sharp product bound for non-trivial cross-intersecting families, (2026) arXiv:2606.23322

Y. Huang, A sharp product bound for non-trivial cross-intersecting families, (2026) arXiv:2606.23322

Pith/arXiv arXiv 2026

[12] [12]

Matsumoto, N

M. Matsumoto, N. Tokushige, The exact bound in the Erd˝ os-Ko-Rado theorem for cross-intersecting families,J. Combin. Theory Ser. A52 (1989) 90–97

1989

[13] [13]

Pyber, A new generalization of the Erd˝ os-Ko-Rado theorem,J

L. Pyber, A new generalization of the Erd˝ os-Ko-Rado theorem,J. Combin. Theory Ser. A43(1) (1986) 85–90. 12

1986

On a conjecture regarding the product version of the Hilton-Milner theorem

Core claim §

What carries the argument

Load-bearing premise §

What would settle it §

discussion (0)

Core claim

Load-bearing premise

What would settle it