Flag vectors of Eulerian partially ordered sets (Q1840824)

From the combinatorial viewpoint Eulerian posets, i.e., finite graded posets in which every interval has the same number of elements of even and odd rank, are both nice and plentiful. From the same viewpoint the flag-vector, i.e., the \(2^n\)-tuple of flag numbers \(f_S(P)\) as \(S\) ranges over \([1,\dots,n]\) when \(P\) (with \(\widehat 0\), \(\widehat 1\)) is of rank \((n+1)\) written in a standard ordered fashion, encodes structural information on \(P\) in a perhaps surprisingly strong and quite efficient way. It is then always of interest to find interesting descriptions of structures related to flag-vectors, these descriptions in turn providing information on the flag-vectors themselves and these in their turn relating to the underlying posets \(P\). Thus, the problem dealt with here, i.e., the description of the closed cone of flag \(f\)-vectors of Eulerian posets, falls into this category. This ``description'' comes in the form of a variety of results, e.g., general linear inequalities valid for all Eulerian flag-vectors, both of known types but only proven up until now for smaller classes of posets and others of newer form, or the identification of special collections of points. Sample results include Theorem 3.2 and Theorem 4.2. Proving Theorem 4.2 requires the appearance of the class of half-Eulerian posets (which when doubled yield Eulerian posets) and these thus turn out to be interesting combinatorial objects themselves. In fact, the posets which are produced are examples due to Billera and Hetyei which give rise to extreme rays of the cone. In an appendix the cone of Eulerian flag vectors is completely determined through rank seven. In this quite substantial paper the \(cd\)-index, itself also somewhat indirectly related to the flag-vector of a given Eulerian poset, is also employed (as in Proposition 2.9 for example) to yield information on the cone and extreme elements in it. Not yet a complete story in all cases, it is nevertheless a considerable addition to what is known in this area, well beyond the problem mainly under investigation here.