Fibonacci-like growth of numerical semigroups of a given genus. (Q1955596)

Let \(\mathbb N\) denote the set of nonnegative integers. A numerical semigroup \(S\) is a submonoid of \(\mathbb N\) with finite complement in \(\mathbb N\). Thus there exists a largest integer not belonging to \(S\), which is denoted by \(f(S)\), and it is known as the Frobenius number of \(S\). Also \(S\cup\{f(S)\}\) is also a numerical semigroup provided that \(S\neq\mathbb N\). Indeed, if we repeat this operation \(g\) times we finally reach \(\mathbb N\), where \(g\) is the cardinality of \(\mathbb N\setminus S\), the genus of \(S\). In this construction, \(f(S)\) becomes a minimal generator of \(S\cup\{f(S)\}\), that is, it is not the sum of nonzero elements in this semigroup. Clearly, the genus of \(S\cup\{f(S)\}\) is precisely \(g-1\). The converse operation would be removing minimal generators greater than \(f(S)\), obtaining in this way numerical semigroups with genus \(g+1\). Thus we can arrange the set of all numerical semigroups in a tree rooted in \(\mathbb N\), and all semigroups with genus \(g\) are at depth \(g\) (one gets rid of generators smaller than the Frobenius number to avoid repetitions in the process of generating all numerical semigroups with genus \(g+1\) from those of genus \(g\)). The minimal generators of \(S\) greater than \(f(S)\) are called effective generators of \(S\), and the efficacy of \(S\) is the cardinality of its set of effective generators. If \(s\) is an effective generator, then we say that \(S\setminus\{s\}\) descends from \(S\). The descent is weak if every effective generator of \(S\setminus\{s\}\) is already an effective generator of \(S\); and strong otherwise, saying in this case that \(S\setminus\{s\}\) is strongly descended (\(\mathbb N\) is considered strongly descended). Any numerical semigroup is obtained by a chain of weak descents from a strongly descended semigroup. So if we denote by \(\mathcal S\) the set of all strongly descended semigroups, \(n_g\) the number of numerical semigroups of genus \(g\), and for \(S\in\mathcal S\), \(N_g(S)\) the set of weak descendants of \(S\) having genus \(g\), we get \[ n_g=\sum_{S\in \mathcal S}N_g(S). \] The author then divides the set of strongly descended numerical semigroups into three sets to get an estimate of \(n_g\), and then uses it to show that \[ \lim_{g\to\infty}\frac{n_g}{\varphi^g}=\mathfrak S, \] where \(\varphi=\frac{1}{2}(1+\sqrt 5)\) (the golden ratio) and \(\mathfrak S\) is a constant. While proving this formula the author also finds an affirmative answer to several open conjectures on the semigroup tree of numerical semigroups proposed by \textit{M. Bras-Amorós} [Semigroup Forum 76, No. 2, 379-384 (2008; Zbl 1142.20039)] and \textit{Y. Zhao} [Semigroup Forum 80, No. 2, 242-254 (2012; Zbl 1204.20080)], namely, that \[ \lim_{g\to\infty}\frac{n_{g-1}+n_{g-2}}{n_g}=1 \] and \[ \lim_{g\to\infty}\frac{t_g}{n_g}=1, \] where \(t_g\) is the set of numerical semigroups of genus \(g\) such that three times the multiplicity exceeds the Frobenius number.