Pseudo-Frobenius numbers versus defining ideals in numerical semigroup rings (Q1643532)

Let \(H\) be a numerical semigroup, i.e. a cofinite additive submonoid of \(\mathbb{N}\), and let \(k[H]\) denote the semigroup ring of \(H\) over a field \(k\). The pseudo-Frobenius numbers of \(H\) are defined as \[ \mathrm{PF}(H) = \big\{ \alpha \in \mathbb{Z} \setminus H \, | \, \alpha + n \in H \text{ for all } n \in H\setminus\{0\}\big\} \] and they correspond to the generators of the socle of the (minimal monomial) artinian reduction of \(k[H]\). The main result of the paper provides a structure theorem for \(n\)-generated numerical semigroups \(H\) whose pseudo-Frobenius numbers are of the form \[ \{ \alpha, 2\alpha, \ldots, (n-1)\alpha\}. \] In particular, the defining equations of such \(k[H]\) are completely described as the minors of a \(2\times n \) matrix whose entries are certain powers of the variables. The rings that satisfy the conditions of this theorem are examples of almost Gorenstein rings. This theorem also gives nice generalizations of well known results for \(n=3\). The proof of the main result relies on a careful analysis of the numerical invariants of such a numerical semigroup \(H\), as well as a comparison map of canonical modules induced from the Eagon-Northcott complex of a \(2\times n\) matrix.