Overview
This study examines $t$-intersecting families within direct product structures, specifically focusing on single layers and finite unions of layers. The framework considers a ground set $X$ partitioned into $p$ disjoint subsets, $X=X_1\mathbin{\dot\cup}\cdots\mathbin{\dot\cup}X_p$. Families of subsets $\mathcal{H}(\boldsymbol k)$ are defined such that for any subset $F \subseteq X$, the size of its intersection with each $X_i$ is $k_i$ (i.e., $|F\cap X_i|=k_i$ for every $i$). The core investigation revolves around the ordinary condition of $t$-intersection, where $|A\cap B|\ge t$.
Approach
A central methodological contribution is a generating-set method. This approach is applied to reduce every shifted extremal non-star system to at most $pt$ supporting points. This reduction is instrumental in deriving the coordinatewise linear Erdős–Ko–Rado thresholds.
The study also tackles a problem related to general $t$ under an explicit polynomial large-part hypothesis. This specific problem was previously posed by Kwan, Sudakov, and Vieira. The resolution of this problem involved demonstrating that every maximum family under these conditions is a product Hilton–Milner family.
Findings
- The generating-set method facilitates the establishment of coordinatewise linear Erdős–Ko–Rado thresholds. This finding improves upon previous quadratic and higher-degree hypotheses.
- For the multilayer result, it is proven that every maximum family is a full $t$-star. This conclusion holds without the necessity of a positive lower bound on each layer coordinate.
- For general $t$, under an explicit polynomial large-part hypothesis, every maximum family is identified as a product Hilton–Milner family, denoted $\mathcal{H}^\times(T,Y)$. In this formulation, $T\subseteq X$ represents a $t$-set, and $Y\subseteq X\setminus T$ is a set with at least two elements. This result addresses a problem initially posed by Kwan, Sudakov, and Vieira.
- The research provides an explicit product formula. Combined with an endpoint reduction technique, this allows for finite optimization and classification of all equality cases.
- The classification of equality cases specifically includes the family $\mathcal{A}_1^\times(T\cup Y)$ when $|Y|=2$.