Constructing numerical semigroups of a given genus. (Q972584)

This is one of several recent papers concerning the problem of counting numerical semigroups by genus. Recall that a numerical semigroup is a subset of \(\mathbb{N}_0\) containing \(0\) that is closed under addition and has finite complement in \(\mathbb{N}\). For a given semigroup, the genus \(g\) is the size of this set of gaps, the Frobenius number \(f\) is the largest gap, and the multiplicity \(m\) is the size of the smallest nonzero element. Let \(n_g\) be the number of semigroups of genus \(g\). This paper improves the best known asymptotic lower bound for \(n_g\). It also gives two conjectures outlining a very promising approach to proving the conjecture of Bras-Amorós that the ratio \(n_g/n_{g-1}\) approaches \(\varphi\), the golden ratio. The author's approach is different than that of previous papers considering this problem. He focuses on counting all of the semigroups of genus \(g\) which satisfy \(f<3m\) and conjectures that asymptotically, almost all semigroups are of this form. The main idea is to count them by type, considering which elements between \(m\) and \(f-m\) the semigroup contains. Upper and lower bounds are given in terms of sums of Fibonacci numbers, which are estimated by standard techniques. This paper is well-written and the arguments are easy to follow. Data is presented that supports the author's conjectures. This paper would be useful reading for anyone interested in questions about counting numerical semigroups or the semigroup tree.