REVIEW 2 major objections 4 minor 19 references
A lemma on a finite union-closed family of finite sets and its applications
T0 review · 2 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash
Pith's one-line read This paper claims that a sharp transfer lemma for union-closed families, applied through an iterative reduction, makes Frankl's union-closed sets conjecture equivalent to Nagel's conjecture.
desk verdict The core lemma is new, sharp, and correct, and the Frankl–Nagel equivalence is real, but the proof is too terse where it matters. 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 the ratio identity (2.1): with x counting sets A in F that omit i and contain j and arise as B\{i} from a set B containing both, and y counting the analogous sets without j, |F_j|/|F| = (|G_j|+x)/(|G_j|+|G_{/j}|+x+y). Since 0 <= y <= |G_{/j}|, the assumption |G_j|/|G| >= c bounds the denominator and forces |F_j|/|F| >= 1/(1+2(1-c)/c). The paper's further tool is an iterative 'reduction method' in which elements are deleted one at a time, a frequency bound is obtained in the reduced family, and Lemma 1.1 lifts the bound back to the original family.
What would settle it
Enumerate all union-closed families on {1,...,m} for m = 12 (a range already checked for Frankl's conjecture in the literature cited by the paper), compute the ordered element frequencies, and test whether any family satisfies Frankl's conjecture yet violates Nagel's bound 1/($2^{{k-1}}$+1) for some k; finding one would disprove the claimed equivalence.
Extended reading notes
Core claim
The paper's central claim is Lemma 1.1: for any finite union-closed family F and any element i, let G be the family obtained by deleting i from every set in F; if some j != i belongs to at least a fraction c in (0,1] of the sets of G, then j belongs to at least 1/(1+2(1-c)/c) of the sets of F. The proof uses an exact ratio identity that expresses |F_j|/|F| as a quotient of counts in G plus two correction terms, and the inequality is sharp in the sense that equality can occur. As applications, the paper argues that Frankl's conjecture is equivalent to Nagel's conjecture, and that for any A in F with |A| >= 2, some y in A occurs in at least 1/($2^{{|A|-2}}$+1) of the sets of F.
Load-bearing premise
The equivalence claim in Proposition 3.1 rests on an unstated 'reduction method': for k >= 3, the authors delete the top k-1 elements one at a time, apply Frankl's conjecture to each reduced family, and assume without proof that Lemma 1.1 restores the required bound at each restoration step.
Editorial extensions
If this is right
- If Proposition 3.1 is right, proving Frankl's conjecture immediately proves Nagel's conjecture, so any counterexample to Nagel's conjecture would also be a counterexample to Frankl's.
- For any union-closed family and any set A of size at least 2, some element of A is present in at least 1/(2^{|A|-2}+1) of the sets, improving the previous best bound of 1/(2^{|A|-1}+1).
- Any future constant lower bound c for the most frequent element in a reduced family can be converted into a frequency bound 1/(1+2(1-c)/c) for the original family; with Liu's c = 0.38234 this gives a second-element bound of about 0.23635.
- The transfer inequality is optimal, since the six-set family in Example 1.2 realizes equality when c = 1/2 and the conclusion is 1/3.
- Under the assumption that the smallest nonempty set in F has size at least 2, the lemma yields two distinct elements with frequencies at least 1/2 and 1/3, respectively.
Reading between the lines
- A formal induction for the 'reduction method' would make the claimed equivalence completely explicit; as written, the k >= 3 case is asserted rather than shown, and that step is what carries the equivalence.
- Read as a lifting principle, Lemma 1.1 says that frequency lower bounds survive deletion-and-restoration with a predictable penalty, so any improvement in a reduced family automatically improves bounds in the original family.
- A natural test is whether the sharp constant can be improved when the deleted element i is itself known to be frequent; the proof only uses the crude bound y <= |G_{/j}| and leaves that structure unused.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper introduces Lemma 1.1, a counting inequality for a finite union-closed family F. For a fixed element i, let G be the family obtained by deleting i from every member of F. The lemma states that if an element j has frequency at least c in G, then its frequency in F is at least 1/(1+2(1-c)/c). The proof is an elementary counting argument. The authors then present three applications: they claim that Frankl's conjecture is equivalent to Nagel's conjecture, they complement a lemma of Nagel on elements of a fixed set A in F, and they make remarks about stronger versions of Frankl's conjecture. The central lemma and the k=2 case of Proposition 3.1 are proved in detail; however, the proofs for k≥3 in Proposition 3.1 and |A|≥4 in Proposition 3.3 are only sketched via an unstated 'reduction method'.
Significance. If the equivalence in Proposition 3.1 is established, the paper gives a surprising and potentially useful link: Frankl's conjecture would imply the full Nagel conjecture, including the currently open k=2 case. The proof of Lemma 1.1 is clean, self-contained, and sharp as demonstrated by Example 1.2. The iterative restoration argument that fills the reduction method is sound (applying Lemma 1.1 with c = 1/(2^r+1) after each restoration), but it is not written down in the manuscript. The paper is short and would be a useful contribution once the missing proofs are supplied.
major comments (2)
- [Section 3.1, proof of Proposition 3.1] The proof for k ≥ 3 is not given; it says only 'using the reduction method, we can obtain that for any k = 3,...,m, (3.2) holds.' This is the load-bearing step for the claimed equivalence, because (3.2) for k ≥ 3 is part of Nagel's conjecture. Please write out the induction: for fixed k, form H = {A \ {1,...,k-1} : A ∈ F}, apply Frankl's conjecture to H to obtain an element i ≥ k with frequency at least 1/2 in H, then restore the deleted elements one at a time. If after r restorations the frequency of i is at least 1/(2^r+1), Lemma 1.1 with c = 1/(2^r+1) yields frequency at least 1/(2^{r+1}+1) after the next restoration. After k-1 restorations, f_i(F) ≥ 1/(2^{k-1}+1), and since i ≥ k, inequality (3.1) gives f_k(F) ≥ 1/(2^{k-1}+1). This argument should be included in the paper.
- [Section 3.2, proof of Proposition 3.3] The case |A| ≥ 4 is dismissed with 'by the reduction method, we can easily obtain the result.' This is again a load-bearing omission because Proposition 3.3 is stated for all |A| ≥ 2. Please provide the analogous induction: order the elements of A as x1,...,x_r with r = |A|, delete x1,...,x_{r-2}, apply the known result for 2-element sets to the reduced family to get an element y among the two remaining elements with frequency at least 1/2, and restore x_{r-2},...,x1 one at a time using Lemma 1.1. After r-2 restorations, the frequency of y in F is at least 1/(2^{r-2}+1), which is exactly (3.3).
minor comments (4)
- [Section 2, proof of Lemma 1.1] The application of Lemma 2.1 is terse; for readability, state explicitly that one takes a = |G_j| + |G_{/j}| + y, b = |G_j|, c = x, d = x, so that b/a is bounded below via (2.2) and d/c = 1.
- [Section 3.3, Remark 3.4] The term 'S-Frankl's conjecture' in item (iii) is not defined; it should be 'S2-version' as introduced above, or a definition should be given.
- [Section 1, introduction] The historical summary states 'the author in [13] proved' but the reference is to Poonen; the wording would be clearer as 'Poonen [13] proved' to avoid ambiguity.
- [Throughout] There are several typographical issues in the extracted text (fractions rendered inline, e.g., '1 1+2(1−c)/c'); the final typeset version should ensure equations are legible.
Circularity Check
No significant circularity: Lemma 1.1 is proved from elementary set cardinality identities, and the applications are conditional arguments that do not smuggle their conclusions into their hypotheses.
full rationale
The paper's central result, Lemma 1.1, is proved directly from the identity |F_j|/|F| = (|G_j| + x)/(|G_j| + |G_/j| + x + y), the assumption |G_j|/|G| >= c, and the elementary weighted-mean inequality of Lemma 2.1. No parameter is fitted to the target conclusion, and no external citation is used to establish the lemma. The applications are conditional: Proposition 3.1 assumes Frankl's conjecture as a hypothesis, applies it to the reduced union-closed family G, and uses Lemma 1.1 to transfer a frequency bound for an element i >= 2 back to F; the same reconstruction is then iterated for k >= 3 via the stated ‘reduction method’. This is a standard conditional derivation, not circular, because the conclusion (Nagel's conjecture) is not assumed at any point. The brief ‘reduction method’ sentence is terse, but an independent reconstruction using Lemma 1.1 with c = 1/(2^r + 1) at each restoration step supplies the missing details and confirms the argument is valid. Proposition 3.3 is analogous and equally non-circular. The only self-citation, [3] by Cui and Hu for the S2-version, is background context and is not load-bearing for any derived claim. The paper therefore exhibits no self-definitional step, no fitted input renamed as a prediction, and no self-citation chain that forces the main results.
Assumptions & free parameters
assumptions (3)
- domain assumption Frankl's conjecture is assumed true for all finite union-closed families
- standard math The projection of a union-closed family under A -> A\{i} is union-closed
- standard math The mediant inequality: if b/a >= k and d/c >= k then (b+d)/(a+c) >= k
Cite this review
Pith. "Pith review of A lemma on a finite union-closed family of finite sets and its applications." pith.science (2026). https://pith.science/paper/OBDEFRXU
@misc{pith2026250711008,
author = {Pith},
title = {Pith review of: A lemma on a finite union-closed family of finite sets and its applications},
year = {2026},
howpublished = {\url{https://pith.science/paper/OBDEFRXU}},
note = {Machine review of arXiv:2507.11008}
}
abstract
Suppose that $\mathscr{F}$ is a finite union-closed family of sets with $\cup_{A\in \mathscr{F}}A=\{1,2,\ldots,m\}$ and $m\geq 2$. Fix $i\in \{1,2,\ldots,m\}$ and denote $\mathscr{G}:=\{A\backslash \{i\}: A\in \mathscr{F}\}$. For $j\in \{1,2,\ldots,m\}\backslash\{i\}$, let $\mathscr{G}_j:=\{A\in\mathscr{G}: j\in A\}$ and $\mathscr{F}_j:=\{A\in\mathscr{F}: j\in A\}$. In this note, we will prove a lemma which says that if $\frac{|\mathscr{G}_j|}{|\mathscr{G}|}\geq c\,(c\in (0,1])$, then $\frac{|\mathscr{F}_j|}{|\mathscr{F}|}\geq \frac{1}{1+2(1-c)/c}$. Several applications of this lemma will be given.
Reference graph
Works this paper leans on
-
[1]
Boˇ snjak I., Markovi´ c P.: The 11-element case of Frankl’s conjecture, Electron. J. Combin. 15(1), #88, 17 pp. (2008) 6
work page 2008
-
[2]
Bruhn H., Schaudt O.: The journey of the union-closed sets conjecture, Graphs Combin. 31(6), 2043-2074 (2015)
work page 2015
-
[3]
Cui Z., Hu Z.-C.: Two stronger versions of the union-closed sets conjecture, Adv. Math. (China) 50(6), 829-851 (2021)
work page 2021
-
[4]
Das S., Wu S.: Frequent elements in union-closed set families, arXiv:2412.03862v3 (2025)
work page Pith review arXiv 2025
-
[5]
Ellis D., Ivan M.-R., Leader I.: Small sets in union-closed families, Electron. J. Combin. 30(1), #1.8, 6 pp. (2023)
work page 2023
-
[6]
Gilmer J.: A constant lower bound for the union-closed conjecture, arXiv:2211.09055 (2022)
arXiv 2022
-
[7]
Hu Z.-C., Li S.-L.: The 6-element case of S1-Frankl conjecture (I), J. Sichuan Univ. (Natural Sci. Edi.), 57(1), 11-26 (2020)
work page 2020
-
[8]
Kabela A., Pol´ ak M., Teska J.: The number of abundant elements in union-closed families without small sets, arXiv:2212.09279v2 (2023)
work page Pith review arXiv 2023
Show all 19 references
-
[9]
Karpas I.: Two results on union-closed families, arXiv:1708.01434 (2017)
2017 arXiv
-
[10]
Liu J.B.: Improving the lower bound for the union-closed sets conjecture via conditional IID coupling, arXiv:2306.08824v1 (2023)
2023 arXiv
-
[11]
Lo Faro G.: Union-closed sets conjecture: improved bounds, J. Comb. Math. Comb. Comput. 16, 97-102 (1994)
1994
-
[12]
Nagel N.: Notes on the union closed sets conjecture, arXiv:2208.03803v2 (2023)
2023 arXiv
-
[13]
Poonen B.: Union-closed families, J. Comb. Theory, Ser. A 59(2), 253-268 (1992)
1992
-
[14]
(Ed.): Graphs and Order, Reidel, Dordrecht/Boston (1985)
Rival I. (Ed.): Graphs and Order, Reidel, Dordrecht/Boston (1985)
1985
-
[15]
Roberts I., Simpson J.: A note on the union-closed sets conjecture, Austral. J. Comb. 47, 265-267 (2010)
2010
-
[16]
G., Renaud J.-C.: On the union-closed sets conjecture, Ars Combin
Sarvate D. G., Renaud J.-C.: On the union-closed sets conjecture, Ars Combin. 27, 149-154 (1989)
1989
-
[17]
P., Enumerative Combinatorics, Vol
Stanley R. P., Enumerative Combinatorics, Vol. I, Wadsworth & Brooks/Cole Advanced Books & Software, Monterey, CA (1986)
1986
-
[18]
Studer L.: An asymptotic version of Frankl’s conjecture. Amer. Math. Monthly 128(7), 652-654 (2021)
2021
-
[19]
Vuˇ ckovi´ c B., Zivkovi´ c M.: The 12-element case of Frankl’s conjecture, IPSI BgD Transactions on Internet Research, 13(1), 65-71 (2017) 7
2017
Reviewed August 6, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.