Intersection theorems under dimension constraints (Q2368655)

The authors study an analog of the intersection problem for finite sets by adding some dimension constrains. Two vectors of \(\{0,1\}^n\) are called \(t\)-intersecting if they have at least \(t\) common ones. For \(A\subseteq \{0,1\}^n\) denote by \(\dim(A)\) the dimension of the vector space spanned by \(A\). The first part of the paper is devoted to computing the value \(J_t(n,k)\), which is the maximum size of a \(t\)-intersecting family \(A\subseteq \{0,1\}^n\) with \(\dim(A)=k\). A conjecture from an earlier paper of the authors is established that provides exact values of \(J_t(n,k)\): \[ J_t(n,k) = \begin{cases} \sum^{k-1}_{i=k-1-{n-t\over 2}} {k-1\choose i} + \sum^{k-1}_{i={n+t\over 2}} {k-1\choose i}, &\text{if }n+t \text{ is even}\\ 2\sum^{k-2}_{i=k-1-{n-t+1\over 2}} {k-2\choose i} + 2\sum^{k-1}_{i={n+t-1\over 2}} {k-2\choose i}, &\text{if } n+t \text{ is odd}\end{cases} \] for \(t > n - k + 1\) and some range of the parameters. It turns out that this problem can be reduced to a weighed version of the intersection problem for systems of finite sets. The authors also study a diametric problem under the same constrains. The second part is devoted to the uniform case when \(A\subseteq \{0,1\}^n_\omega\), that is, each vector of \(A\) has exactly \(\omega\) ones and \(\dim(A)\leq k\). The corresponding maximum size of a \(t\)-intersecting family in this case is denoted by \(J_t(n,k,\omega)\). It is conjectured that for \(\omega\leq n/2\) one has \(J_t(n,k,\omega)= \max\{| U_k| \cap \{0,1\}^{n-1}_{\omega-1}\}\), where the maximum runs over all \(k\)-dimensional subspaces \(U_k\) of \(\{0,1\}^{n-1}\). This conjecture is established for \(t=1\) and \(k<2\omega\) and for any fixed \(t\), \(1\leq t\leq\omega\), and large \(k\).