Constructing the set of complete intersection numerical semigroups with a given Frobenius number. (Q360195)

A numerical semigroup \(\Gamma\) is a cofinite submonoid of \((\mathbb N,+)\). We call such a semigroup a complete intersection if the ring \(\Bbbk[[t^n:n\in\Gamma]]\) is a complete intersection. For a numerical semigroup \(\Gamma\) we define the Frobenius number \(f(\Gamma)=\max(\mathbb Z\setminus\Gamma)\), the multiplicity \(m(\Gamma)=\min(\Gamma\setminus\{0\})\), the genus \(g(\Gamma)=|\mathbb N\setminus\Gamma|\) and the embedding dimension \(e(\Gamma)\) as the number of minimal generators. The gluing of two semigroups \(\Gamma_1\) and \(\Gamma_2\), introduced first by \textit{C. Delorme} [Ann. Sci. Éc. Norm. Supér. (4) 9, 145-154 (1976; Zbl 0325.20065)] with a different terminology, is an operation that yields a semigroup \(\Gamma\) with \(e(\Gamma)=e(\Gamma_1)+e(\Gamma_2)\) and preserves several properties. In particular, Delorme showed that the collection of complete intersections can be constructed recursively via gluing. In the paper under review the authors use gluing to provide bounds for \(e(\Gamma)\) and \(m(\Gamma)\) in terms of \(f(\Gamma)\) under the assumption that \(\Gamma\) is a complete intersection. They also present an algorithm to construct the (finite) collection of complete intersections with fixed Frobenius number. Then the authors focus their attention on certain subclasses of the complete intersections, namely free and telescopic numerical semigroups and semigroups associated to singularities of irreducible plane curves. For each of these classes, some inequalities involving the generators of \(\Gamma\) are proved, allowing improved versions of the algorithm mentioned above. The authors conclude the paper by presenting some experimental results obtained by implementing the algorithms in \texttt{GAP}. More specifically, they provide data on the asymptotic behavior of the proportion of complete intersections (and the other classes) in the set of semigroups with fixed genus, as the genus increases. These computations suggest that most complete intersections are free.