The Frobenius problem for generalized repunit numerical semigroups (Q2682903)

Let \(b\) and \(n\) be integers greater than one, and let \(a_1=(b^n-1)/(b-1)\) (a repunit number; all its digits in base \(b\) are 1). Choose \(a\) coprime with \(a_1\), and set \(a_i=a_{i-1}+a b^{i-2}\), for all \(i\ge 2\). Denote by \(S_a(b,n)\) the submonoid of \((\mathbb{N},+)\) generated by the sequence \(\{a_i\}_{i\ge 1}\) (where \(\mathbb{N}\) denotes the set of nonnegative integers). The condition \(\gcd(a,a_1)=1\) forces \(\mathbb{N}\setminus S_a(b,n)\) for have finitely many elements, and thus \(S_a(b,n)\) is a numerical semigroup. The authors prove that the minimal generating set of \(S_a(b,n)\) is \(\{a_1,\dots,a_n\}\) and, thus, the embedding dimension of this numerical semigroup is \(n\). These semigroups are named generalized repunit numerical semigroups, and are a generalization of Meresenne numerical semigroups and repunit numerical semigroups. They are also closed under the action of the affine map \(x\mapsto bx+a-(b^n-1)\). Given a numerical semigroup \(S\) and a nonzero element \(m\) of \(S\), the Apéry set of \(m\) in \(S\) is the set \(\operatorname{Ap}(S,m)=\{ s\in S : s - m \not\in S\}\). This set has exactly \(m\) elements, \(\max(\operatorname{Ap}(S,m))-m\) is precisely the largest integer not in \(S\), known as the Frobenius number of \(S\), and the cardinality of \(\mathbb{N}\setminus S\) is \((\sum_{w\in \operatorname{Ap}(S,m)} w)/m-(m-1)/2\), known as the genus of \(S\) (these two expressions of the Frobenius number and genus of \(S\) are usually known as Selmer's formulas). Also, the set of pseudo-Frobenius numbers of \(S\), \(\operatorname{PF}(S)=\{z\in \mathbb{Z}\setminus{S}: z+(S\setminus\{0\})\subseteq S\}=-m+\operatorname{Maximals}_{\le_S}(\operatorname{Ap}(S,m)\), where \(\le_S\) is the order on \(\mathbb{Z}\) induced by \(S\), that is, \(a\le_S b\) if \(b-a\in S\). The cardinality of \(\operatorname{PF}(S)\) is the type of \(S\). For \(m\) equal to the multiplicity of \(S\) (the least positive integer in \(S\)), one usually writes \(\operatorname{Ap}(S)\) instead of \(\operatorname{Ap}(S,m)\). The authors provide a recursive procedure to compute \(\operatorname{Ap}(S_a(b,n))\) in terms \(\operatorname{Ap}(S_a(b,n-1))\). With a complete description of the elements of the Apéry set of the multiplicity of \(S_a(b,n)\), formulas for the Frobenius number, genus, and pseudo-Frobenius numbers of \(S_a(b,n)\) are derived. In particular it is shown that the type of \(S_a(b,n)\) is \(n-1\), and consequently \(S_a(b,n)\) fulfills Wilf's conjecture. The authors also prove that for every element in \(\operatorname{Ap}(S_a(b,n))\), all its factorizations have the same length, and thus \(S_a(b,n)\) is an homogeneous numerical semigroup.