Partitions of a finite three-complete poset (Q1910551): Difference between revisions

If \(P\) is a finite poset, then there is a minimal number of disjoint chains \(\{T_1, \dots, T_m\}\) which cover it. Uniqueness is not required here. Interesting properties are easy to prove if the number \(m\) is one and much less so even if the number is as low as two. When the number is three the situation is vastly more complicated. Nevertheless, Daykin and Daykin were able to prove interesting results in this situation and in 1985 they conjectured that if \(m = 3\) then \(P = R_1 \cup \cdots \cup R_n\), with \(R_1 < R_2 < \cdots < R_n\) (i.e., if \(x \in R_i\) and \(y \in R_j\), \(i < j\), then \(x < y\)) and such that for each \(i \in \{1, \dots, n\}\), \(j \in \{1,2,3\}\), either \(R_i\) and \(T_j\) are disjoint or if \(p\) and \(q \in R_i\) and different chains, then \(p\) and \(q\) are incomparable. The authors prove this conjecture via a detailed construction based on two equivalence relations: \(\delta\) (on chains \(T\)): \(t_1 \delta t_2\) iff \(\text{above} \{t_1\}\setminus T = \text{above} \{t_2\} \setminus T\) and \(\text{below} \{t_1\} \setminus T = \text{below} \{t_2\}\setminus T\) for \(t_1, t_2 \in T \subseteq P\); \(\rho\) (on equivalence classes defined by \(\delta\) on the \(T_i\)): \(T^z_i \rho T^k_j\) iff \((\text{above }T^z_i)\setminus T^z_i = (\text{above }T^k_j)\setminus T^k_j\) or \((\text{below }T^z_i)\setminus T^z_i = (\text{below }T^k_j)\setminus T^k_j\). From the unions \(M_\ell\) of the elements in the equivalence classes \(K_1\) of the (not so easy to prove) equivalence relation a careful construction allows the blocks \(M_1\) to be used to build the looked for sets \(R_i\) which settles the conjecture, leaving hope for further steps along the lines developed in this paper.