English

A cross-intersection theorem for subsets of a set

Combinatorics 2015-06-12 v1

Abstract

Two families A\mathcal{A} and B\mathcal{B} of sets are said to be cross-intersecting if each member of A\mathcal{A} intersects each member of B\mathcal{B}. For any two integers nn and kk with 0kn0 \leq k \leq n, let ([n]k){[n] \choose \leq k} denote the family of all subsets of {1,,n}\{1, \dots, n\} of size at most kk. We show that if A([m]r)\mathcal{A} \subseteq {[m] \choose \leq r}, B([n]s)\mathcal{B} \subseteq {[n] \choose \leq s}, and A\mathcal{A} and B\mathcal{B} are cross-intersecting, then ABi=0r(m1i1)j=0s(n1j1),|\mathcal{A}||\mathcal{B}| \leq \sum_{i=0}^r {m-1 \choose i-1} \sum_{j=0}^s {n-1 \choose j-1}, and equality holds if A={A([m]r) ⁣:1A}\mathcal{A} = \{A \in {[m] \choose \leq r} \colon 1 \in A\} and B={B([n]s) ⁣:1B}\mathcal{B} = \{B \in {[n] \choose \leq s} \colon 1 \in B\}. Also, we generalise this to any number of such cross-intersecting families.

Keywords

Cite

@article{arxiv.1402.3969,
  title  = {A cross-intersection theorem for subsets of a set},
  author = {Peter Borg},
  journal= {arXiv preprint arXiv:1402.3969},
  year   = {2015}
}

Comments

12 pages, submitted. arXiv admin note: text overlap with arXiv:1212.6955

R2 v1 2026-06-22T03:09:36.637Z