Intersection theorems for \(\{0,\pm1\}\)-vectors and \(s\)-cross-intersecting families (Q1681899)

Let \([n] := \{1, \dots, n\}\) and \({[n] \choose k} := \{A \subseteq [n] : |A| = k\}\). The classical Erdős-Ko-Rado theorem [Quart. J. Math. Oxford (2) 12, 313--320 (1961; Zbl 0100.01902)] states that, if \(1 \leq k \leq n/2\), then the size of any family of pairwise intersecting sets in \({[n] \choose k}\) is at most \({n-1 \choose k-1}\). The authors consider two variants of the setting of this theorem. Let \(\mathcal{L}_k\) denote the set of all vectors \( v \in \{0,1,-1\}^n\) such that \(\langle v, v \rangle = k\). Let \(F(n,k,l)\) denote the maximum size of a subset \(\mathcal{V}\) of \(\mathcal{L}_k\) such that \(\langle v, w \rangle \geq l\) for every \(v, w \in \mathcal{V}\). Let \(f(k,l) := \sum_{i=0}^{l/2} {k \choose i}\) if \(l\) is even, and let \(f(k,l) := 2\sum_{i=0}^{(l-1)/2} {k-1 \choose i}\) if \(l\) is odd. The authors prove that, for \(0 \leq l \leq k\) and \(n \geq n_0(k)\), \(F(n,k,l) = {n-l \choose k-l}\) and \(F(n,k,-l) = f(k,l){n \choose k}\). They also determine \(F(n,3,0)\) for \(n \geq 3\). Moreover, they prove that \(\max\{|\mathcal{V}| : \mathcal{V} \subset \mathcal{L}_k, \; \forall v, {\mathbf w} \in \mathcal{V} \; \; \langle v, w \rangle \neq -l-1\} = f(k,l){n \choose k} + O(n^{k-1})\) for \(0 \leq l < k\). Two families \(\mathcal{A}\) and \(\mathcal{B}\) are \(s\)-cross-intersecting if \(|A \cap B| \geq s\) for every \(A \in \mathcal{A}\) and \(B \in \mathcal{B}\). A family \(\mathcal{A}\) is \(t\)-intersecting if \(|A \cap B| \geq t\) for every \(A, B \in \mathcal{A}\). Let \(k > s > t \geq 1\), \(k \geq 2s-t\). For \(0 \leq i < s-t\), let \(\mathcal{M}_i(k,s,t) := \{A \subset [k] : |A| \geq s, |A \cap [t+2i]| \geq t+i\} \cup \{A \subset [k] : |A| \geq k-s+t\}\), \(\mathcal{A}_i := \{A \in {[n] \choose k} : A \cap [k] \in \mathcal{M}_i(k,s,t)\}\) and \(\mathcal{B}_i = \{[k]\}\). Let \(\mathcal{A}_{s-t} := \{A \in {[n] \choose k} : |A \cap [2s-t]| \geq s\}\) and \(\mathcal{B}_{s-t} := \{B \in {[n] \choose k} : [2s-t] \subset B\}\). For \(0 \leq i \leq s-t\), \(\mathcal{A}_i\) and \(\mathcal{B}_i\) are non-empty \(s\)-cross-intersecting \(t\)-intersecting families. The authors prove that, for \(n \geq n_0(s,t,k)\), the maximum sum of sizes of two non-empty \(s\)-cross-intersecting \(t\)-intersecting subfamilies \(\mathcal{A}\) and \(\mathcal{B}\) of \({[n] \choose k}\) is \(\max\{|\mathcal{A}_i| + |\mathcal{B}_i| : 0 \leq i \leq s-t\}\).