A new proof of a theorem of Harper on the Sperner-Erdős problem
Let P be a finite poset. A weight on P is a function \(v: P\to \{x\in R\); \(x>0\}\), a rank on P is a function \(r: P\to \{0,1,2,...\}\) such that \(r(x)=0\), if x is minimal, and \(r(x)=r(y)+1\), if \(x>y\) (i.e., x covers y). For any subset \(S\subseteq P\) define \(v(S)=\sum_{x\in S}v(x)\). Put \(N_ i=\{x\in P;\quad r(x)=i\},\) \(r(P)=\max \{r(x);\quad x\in P\}.\) The weighted and ranked poset P is called normal, if \(v(A)/v(N_ i)\leq v(R(A))/v(N_{i+1})\) for all \(A\subseteq N_ i\) and \(i\in \{0,...,r(P)- 1\}\) where \(R(A)=\{x\in N_{i+1}\); \(x>a\) for some \(a\in A\}\). Let (P,v) and (Q,w) be weighted posets. A function \(\phi\) : \(P\to Q\) is called a flow morphism, if (i) \(\phi\) is surjective, (ii) \(x<y\) implies \(\phi (x)<\phi (y)\), (iii) \(w(x)=v(\phi^{-1}(x))\) for all \(x\in Q\), (iv) the poset induced on \(\phi^{-1}(x)\cup \phi^{-1}(y)\) is normal and has rank 1 for all x,y\(\in Q\) with \(x<y\). Using the framework of category theory, \textit{L. H. Harper} [Adv. Appl. Math. 1, 158-181 (1980; Zbl 0469.90028)] proved the following theorem: If there is a flow morphism \(\phi\) from (P,v) onto (Q,w), then max \(\{\) v(A); \(A\subseteq P\) is an \(antichain\}=\max \{w(B)\); \(B\subseteq Q\) is an antichain\(\}\). In the paper under review, the author gives another proof of this theorem. This proof is based on the method of linear programming and on the fact that relation graphs of posets are perfect.
- A short proof of the existence of k-saturated partitions of partially ordered sets
- Blocking and anti-blocking pairs of polyhedra
- scientific article; zbMATH DE number 3828714 (Why is no real title available?)
- scientific article; zbMATH DE number 3636041 (Why is no real title available?)
- scientific article; zbMATH DE number 3225808 (Why is no real title available?)
- Maximal sized antichains in partial orders
- Optimal representations of partially ordered sets and a limit Sperner theorem
- The global theory of flows in networks
- The structure of Sperner k-families
This page was built for publication: A new proof of a theorem of Harper on the Sperner-Erdős problem
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1065006)