Characterizing face and flag vector pairs for polytopes (Q2189744)

For a \(d\)-polytope \(P\) the integer vector \((f_0(P),f_1(P),\ldots,f_{d-1}(P))\) is the \(f\)-\textit{vector} of \(P\) where \(f_i(P)\) is the number of \(i\)-faces of \(P\). More generally, the integer vector \((f_S(P))_{S\subseteq \{0,1,\ldots,d-1\}}\) is the \textit{flag vector} of \(P\), where \(f_S(P)\) denotes the number of chains \(F_1 \subset\cdots\subset F_r\) of faces of \(P\) where \(\{\dim(F_1),\ldots,\dim(F_r)\} = S\). A celebrated result by Steinitz from 1906 characterizes all integer vectors \((f_0,f_1,f_2)\) that are \(f\)-vectors of \(3\)-polytopes. Such a characterization for \(d\)-polytopes when \(d\geq 4\) remains an open problem. However, during the years from 1967 and 1974, Grünbaum, Barnette and Reay characterized all integer vectors \((f_i,f_j)\) for which there is a \(4\)-polytope \(P\) with \(f_i = f_i(P)\) and \(f_j = f_j(P)\). The first main theorem of this article completely characterizes all ordered pairs \((f_0,f_{03})\) of flag numbers of \(4\)-polytopes. The second and third main theorems provide descriptions of \((f_0,f_{d-1})\) that can be realized by \(d\)-polytopes. A \textit{polytopal pair} is an ordered pair \((n,m)\) of integers such that there is a \(d\)-polytope \(P\) with \(n\) vertices and \(m\) facets, that is \((n,m) = (f_0(P),f_{d-1}(P))\). By the Upper Bound Theorem (UBT) necessary conditions for \((n,m)\) to be a polytopal pair is (i) \(m \leq f_{d-1}(C(d,n))\), and dually (ii) \(n\leq f_{d-1}(C(d,m))\) where \(C(d,n))\) is the \(d\)-dimensional cyclic polytope on \(n\) vertices. The second main theorem states that for even \(d\geq 2\) a pair \((n,m)\) is polytopal if and only if it satisfies both the mentioned UBT conditions (i) and (ii) and \(n+m\geq\binom{3d+1}{\lfloor d/2\rfloor}\). The third main theorem states that for odd \(d\geq 5\), every \((n,m)\) satisfying the UBT conditions (i) and (ii) and \(n+m\geq\binom{3d+1}{\lfloor d/2\rfloor}\) is a polytopal pair with infinitely many exceptional cases both when (a) \(m\) is odd and \(f_{d-1}(C(d,n-1))\leq m\) and, dually, also when (b) \(n\) is odd and \(f_{d-1}(C(d,m-1))\leq n\). This article is well written and contains a wealth of beautiful and explanatory figures.