Two results on set families: sturdiness and intersection

TL;DR

This paper proves a bound on the sturdiness of IU-families and establishes a tight upper bound on the sum of sizes of cross t-intersecting separated families, resolving open problems in extremal set theory.

Contribution

It confirms a conjecture on the sturdiness of IU-families and provides a counterexample to an open problem on cross t-intersecting separated families.

Findings

Proved that the sturdiness of IU-families is at most 2^{n-4}.

Established a tight upper bound on the sum of sizes of cross t-intersecting separated families.

Provided explicit counterexamples to an open problem on separated families.

Abstract

This paper resolves two open problems in extremal set theory. For a family $\mathcal{F} \subseteq 2^{[n]}$ and $i, j\in [n]$, we denote $\mathcal{F} (i,\bar{j})=\{F\backslash\{i\}: F\in \mathcal{F}, F\cap\{i,j\}=\{i\}\}$. The sturdiness $\beta (\mathcal{F})$ is defined as the minimum $|\mathcal{F} (i,\bar{j})|$ over all $i\neq j$. A family $\mathcal{F}$ is called an IU-family if it satisfies the intersection constraint: $F\cap F'\neq \emptyset $ for all $F,F'\in \mathcal{F}$, as well as the union constraint: $F\cup F' \neq [n]$ for all $F,F'\in \mathcal{F}$. The well-known IU-Theorem states that every IU-family $\mathcal{F}\subseteq 2^{[n]}$ has size at most $ 2^{n-2}$. In this paper, we prove that if $\mathcal{F}\subseteq 2^{[n]}$ is an IU-family, then $\beta (\mathcal{F})\le 2^{n-4}$. This confirms a recent conjecture proposed by Frankl and Wang. As the second result, we establish a tight upper bound on the sum of sizes of cross $t$-intersecting separated families. Our result not only extends a previous theorem of Frankl, Liu, Wang and Yang on separated families, but also provides explicit counterexamples to an open problem proposed by them, thereby settling their problem in the negative.