Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated


Source. Unnumbered opening statement, p. 125, of P. Frankl, "On families of finite sets no two of which intersect in a singleton," Bull. Austral. Math. Soc. 17 (1977), no. 1, 125-134, doi:10.1017/S0004972700025521. Pages are the journal's own, as on the source card.

Statement

Let XX be a set of nn elements and F\mathcal F a family of kk-element subsets of XX. The paper proves (p. 125): if n>n0(k)n>n_0(k), k≥4k\ge4 and ∣F∣>(n−2k−2)\lvert\mathcal F\rvert>\binom{n-2}{k-2}, then F\mathcal F has two members F,GF,G with ∣F∩G∣=1\lvert F\cap G\rvert=1.

Equivalently, in the language of §1 (pp. 125-126): an (n,{0,2,3,…,k−1},k)(n,\{0,2,3,\ldots,k-1\},k)-system, a family of kk-subsets of an nn-set in which any two different members meet in a number of elements from {0,2,3,…,k−1}\{0,2,3,\ldots,k-1\}, has at most (n−2k−2)\binom{n-2}{k-2} members when k≥4k\ge4 and n≥n0(k)n\ge n_0(k), as the conjecture is restated there. The paper attributes the conjecture to Erdős and Sós, citing Erdős's problem paper in the Proceedings of the Fifth British Combinatorial Conference (1975), and records that Katona proved the case k=4k=4 (unpublished). The bound is attained by the (n−2k−2)\binom{n-2}{k-2} kk-sets containing two fixed elements, any two of which share at least two elements.

The paper gives no explicit value of n0(k)n_0(k). The conclusion is delivered by Theorem 2 (p. 132), whose range is n>n0(k)+2(n0(k)k)n>n_0(k)+2\binom{n_0(k)}{k} with n0(k)n_0(k) the bound from Theorem 1.

Read depth. Claims checked: the statement and the definitions it uses were read on the print.

Proof pointer

Lemmas 1 to 3 (pp. 126-128) and Theorems 1 and 2 (pp. 128-133). The condition says exactly that each link Fx={F−x:x∈F∈F}\mathcal F_x=\{F-x : x\in F\in\mathcal F\} is an intersecting family. The proof studies the minimal kernels of large sunflowers (Δ\Delta-systems) in each link, proves the structural Theorem 1, and iterates it in Theorem 2.

Bears on

  • Problem 702: this statement, with its range n>n0(k)n>n_0(k), is the problem's corrected Statement, and the paper proves it. The paper says nothing about smaller nn.