Rigid Hilbert functions (Q810095)

Let A,m,k be a local Cohen-Macaulay ring of dimension \( d,\) embedding codimension h, multiplicity e, and initial degree a. Let \(P_ A(z)=\sum^{\infty}_{n=0}H_ A(n)z^ n \) be the Hilbert-Poincaré series (generating function for the Hilbert function \(H_ A(n)=\dim_ km^ n/m^{n+1})\), and \(h_ A(x)=\sum^{d-1}_{j=0}(-1)^ je_ j\left( \begin{matrix} x+d-1-j\\ d-1-j\end{matrix} \right) \) be the Hilbert polynomial. The goal of this paper is to determine \(P_ A\) from a minimal amount of information about the coefficients of \(h_ A\); the terminology of the title refers to situations in which the Hilbert polynomial determines the Hilbert function (and similarly for the sum- transforms \(H^ i_ A\) and the polynomials \(h^ i_ A\) that correspond to these). The first main result is that if \(e_ 1\leq et-\left( \begin{matrix} h+t\\ h+1\end{matrix} \right)\) for some \(t>0\), then \(P_ A(z)=(\sum^{t- 1}_{i=0}\left( \begin{matrix} h+i-1\\ i\end{matrix} \right)z^ i+cz^ t)/(1- z)^ d \) for some \(c\geq 0\). Also in this case \(gr_ mA\) is Cohen- Macaulay. If \(e_ 1=et-\left( \begin{matrix} h+t\\ h+1\end{matrix} \right)+1\) an additional hypothesis is needed: for \(a=t-1\) and for \(a=t+1\) (minimum and maximum possible) formulas for \(P_ A\) are obtained, and again \(gr_ mA\) is Cohen-Macaulay. For \(a=t\) and \(d\leq 1\) the authors find that \(P_ A(z)=(\sum^{t-1}_{i=0}\left( \begin{matrix} h+i-1\\ i\end{matrix} \right)z^ i+bz^ t+z^{t+1})/(1-z)^ d\) for some \(b\geq 0\), but it is open whether this remains true for \(d\geq 2\). Furthermore, if the Poincaré series does have this form, then \(gr_ mA\) has depth at least d-1, but it can fail to be Cohen-Macaulay. Finally the authors discuss the graded case, showing first that \(e_ 1=e-1+\left( \begin{matrix} e-h\\ 2\end{matrix} \right)\) implies \(P_ A(z)=(1+hz+\sum^{e-h}_{i=2}z^ i)/(1-z)^ d\), and second that if A is Gorenstein with \(e_ 1=e(t-1)\) and \(a\geq t\), then A is extremal, \(a=t\), and \(P_ A\) is completely determined.