The proportion of Weierstrass semigroups (Q2377441)

Let \(S\) be a numerical semigroup. The finite complement set \(\mathbb{N} \setminus S\) is the gap set of \(S\). Its cardinality, called the genus of \(S\), is denoted by \(g\), the largest element of \(\mathbb{N} \setminus S\) by \(F\), and the smallest positive element in \(S\) by \(m\). Take now a projective algebraic complex curve \(C\) of genus \(g\). Weierstrass' theorem states that given \(p \in C\), there exist exactly \(g\) integers \(\alpha_i(p)\), \(1=\alpha_1(p) < \dots < \alpha_g(p) \leq 2g-1\), such that there does not exist a meromorphic function on \(C\) which has a pole of order \(\alpha_i(p)\) at \(p\), and no other singularities. Then \(\mathbb{N} \setminus \{\alpha_1(p), \dots, \alpha_g(p)\}\) is a numerical semigroup of genus \(g\). A semigroup \(S\) is called Weierstrass if there exist a curve \(C\) and \(p \in C\) such that \(S\) is this semigroup. A problem coming back from Hurwitz is to determine which semigroups are Weierstrass. The first proof that not all semigroups are Weierstrass was obtained in [\textit{R. O. Buchweitz}, Über Deformation monomialer Kurvensingularitäten und Weierstrasspunkte auf Riemannschen Flächen, PhD thesis, Hannover, (1976)]. He proved that if there exists \(n>1\), such that the cardinality of the set of all sums of \(n\) gaps of \(S\) is greater than \((2n-1)(g-1)\), then \(S\) is not Weierstrass. The first main result of the paper under review considers \(NB(g)\), the number of semigroups \(S\) of genus \(g\) for which there is some \(n\) such that the above inequality holds; and \(N(g)\) the number of semigroups of genus \(g\). Then Theorem 5 states that the limit \(NB(g)/N(g)\) is \(0\) when \(g \rightarrow \infty\). On the other hand, in [\textit{D. Eisenbud} and \textit{J. Harris}, Invent. Math. 87, 495--515 (1987; Zbl 0606.14014)] it was proved that if \(S\) is a semigroup of genus \(g\) with \(F<2m\) and weight \(W(S)<g-1\), then \(S\) is Weierstrass. The second main result of the paper under review is Proposition 20. Calling \(Q(g)\) the number of semigroups of genus \(g\) satisfying Eisenbud-Harris criterion, the limit of \(Q(g)/N(g)\) is 0 when \(g \rightarrow \infty\). These results leave open the question of the density of the subset of Weierstrass semigroups in the set of all numerical semigroups, since both Buchweitz' and Eisenbud-Harris' criteria are sufficient, but not necessary.