An Erdős-Ko-Rado theorem for the subcubes of a cube (Q762158)

Let be the vectors \(X=(x_ 1,...,x_ n)\) with \(x_ i\in \{0,...,k\}\) the elements of the poset P \((i=1,...,n)\). The order in P is given by \(x\leq y\) iff \(x_ i=y_ i\) or \(x_ i=0\) for all i. (Another aspect of the examination of this poset is given in \textit{N. Metropolis} and \textit{G.-C. Rota}, SIAM J. Appl. Math. 35, 689-694 (1978; Zbl 0402.05010). A subset F of P is called Erdős-Ko-Rado family, if for all x,y\(\in F\) it holds \(x\nless y\), \(x\ngtr y\) and there exists an \(z\in P\) with exactly one non-zero component such that \(z\leq x\), \(z\leq y\). Let \({\mathcal F}\) be the set of all vectors \(f=(f_ 0,...,f_ n)\) for which there is an EKR family F in P such that \(f_ i\) equal to the number of elements of F with exactly i non-zero components. The author describes the vertices of the convex closure \(<{\mathcal F}>\) of \({\mathcal F}\) in the Euclidean \((n+1)\)- space (K\(\geq 2)\). They are as follows: (0,...,0) and \(\left(0,...,0,\binom{n-1}{i-1} k^{i-1},0,...,0\right)\) (i-component). The above problem for \(k=1\) was raised and solved by the reviewer, \textit{P. Frankl} and \textit{G. O. H. Katona} in terms of the extremal set theory [Combinatorica 4, 21-34 (1984; Zbl 0544.05001)]. The rest of the paper generalizes the above result for a more complicated poset showing that the received polyhedron is identical with \(<{\mathcal F}>\).