On counting numerical semigroups by maximum primitive and Wilf's conjecture
Manuel Delgado, Neeraj Kumar, Claude Marion
TL;DR
This paper introduces counting numerical semigroups by maximum primitive, defining $A_n$ and relating it to the Frobenius-based counts $N_n$ via Möbius transforms. It proves $A_n$ and $N_n$ are asymptotically equivalent and proves almost all semigroups with large maximum primitive satisfy Wilf's conjecture, building on a key criterion: if a semigroup has many left-over elements in $(m,2m)$ then Wilf holds. A central tool is the injective but not surjective map $\Phi$ between the max-primitive and Frobenius-counting families, plus a divisor-based partition linking $A_n$ and $N_n$ through $N_n = \sum_{d|n} A_d$ and Möbius inversion. The results yield explicit growth bounds, lower bounds, and asymptotics for $A_n$, and establish that almost all $S\in\mathcal{A}_n$ satisfy Wilf as $n\to\infty$, with new infinite Wilf-satisfying classes arising from the derived criterion. Overall, the work deepens the connection between different counting paradigms in numerical semigroups and extends Wilf-type results to a broader combinatorial setting.
Abstract
We introduce a new way of counting numerical semigroups, namely by their maximum primitive, and show its relation with the counting of numerical semigroups by their Frobenius number. For any positive integer $n$, let $A_{n}$ denote the number of numerical semigroups whose maximum primitive is $n$, and let $N_{n}$ denote the number of numerical semigroups whose Frobenius number is $n$. We show that the sequences $(A_{n})$ and $(N_{n})$ are Möbius transforms of one another. We also establish that almost all numerical semigroups with large enough maximum primitive satisfy Wilf's conjecture. A crucial step in the proof is a result of independent interest: a numerical semigroup $S$ with multiplicity $\mathrm{m}$ such that $|S\cap (\mathrm{m},2 \mathrm{m})|\geq \sqrt{3\mathrm{m}}$ satisfies Wilf's conjecture.
