Symmetries on almost symmetric numerical semigroups. (Q1941743)

Denote by \(\mathbb N\) the set of nonnegative integers. A numerical semigroup \(H\) is a set of nonnegative integers closed under addition and such that \(0\in H\) and \(\mathbb N\setminus H\) has finitely many elements. The cardinality of \(\mathbb N\setminus H\) is the genus of \(H\), \(\mathrm g(S)\). Let \(M=H\setminus\{0\}\), which is the unique maximal ideal of \(H\). Then \(H\) is generated as a monoid by \(M\setminus(M+M)\); indeed this is the only minimal generating system of \(M\), and its cardinality is known as the embedding dimension of \(H\), \(\mathrm e(H)\). The multiplicity of \(H\) is the least positive integer belonging to \(H\), and it is denoted by \(\mathrm m(H)\). It can be easily shown that \(\mathrm e(H)\leq\mathrm m(H)\). If the equality holds, then \(H\) is said to be of maximal embedding dimension. The integer \(\max(\mathbb Z\setminus H)\) is the Frobenius number of \(H\), \(\mathrm F(H)\). Clearly it is the smallest integer such that \(\mathrm F(H)+(\mathbb N\setminus\{0\})\subseteq H\). This is why if an integer \(f\) not in \(H\) has the property that \(f+M\subseteq H\), then \(f\) is said to be a pseudo-Frobenius number. The set of pseudo-Frobenius numbers is denoted by \(\mathrm{PF}(H)\), and corresponds with the set of maximal elements in \(\mathbb Z\setminus H\) with respect to the ordering \(h\leq_Hh'\) if \(h'-h\in H\). The cardinality of \(\mathrm{PF}(H)\), \(\mathrm t(H)\), is the type of \(H\) (it is the Cohen-Macaulay type of the semigroup ring associated to \(H\)). Observe that \(\mathrm F(H)\in\mathrm{PH}(H)\). For every \(h\in H\cup (\mathrm{PH}(H)\setminus\{\mathrm F(H)\})\), \(\mathrm F(H)-h\not\in H\). From this it follows that \(2\mathrm g(H)\geq\mathrm F(H)+\mathrm t(H)\). The numerical semigroup \(H\) is said to be almost symmetric if the equality holds. This concept was first introduced by \textit{V. Barucci} and \textit{R. Fröberg} [J. Algebra 188, No. 2, 418-442 (1997; Zbl 0874.13018)] and some characterizations where also presented. Let \(n\in M\). The Apéry set of \(n\) in \(H\) is \(\mathrm{Ap}(H,n)=\{ h\in H\mid h-n\not\in H\}\). It is well known that \(\mathrm{PF}(H)=\mathrm{Maximals}_{\leq_H}(\mathrm{Ap}(H,n)-n)\). The author proves that almost symmetry of \(H\) can be characterized in terms of certain symmetries in \(\mathrm{Ap}(H,n)\) or equivalently in \(\mathrm{PF}(H)\). More explicitly: if the elements of \(\mathrm{Ap}(H,n)\) are arranged as \(\{0<\alpha_1<\cdots<\alpha_m\}\cup\{\beta_1<\cdots<\beta_{\mathrm t(H)-1}\}\), with \((\mathrm{PF}(H)\setminus\{\mathrm F(H)\})+n=\{\beta_1<\cdots<\beta_{\mathrm t(H)-1}\}\) and \(\alpha_m=\mathrm F(H)+n\), then \(H\) is almost symmetric if and only if for all suitable \(i,j\), \(\alpha_i+\alpha_{m-i}=\alpha_m\) and \(\beta_j+\beta_{\mathrm t(H)-j}=\mathrm F(H)\), or equivalently, \(f_i+f_{\mathrm t(H)-i}=\mathrm F(H)\), \(i\in\{1,\dots,\mathrm t(H)-1\}\) (writing \(\mathrm{PF}(H)=\{f_1<\cdots<f_{\mathrm t(H)}\}\), and thus \(f_{\mathrm t(H)}=\mathrm F(H)\)). In this way, the author generalizes a well known characterization for irreducible numerical semigroups (irreducible means that it cannot be expressed as the intersection of numerical semigroups containing it). Let \(H^*=H-M=\{z\in\mathbb Z\mid z+M\subseteq H\}\) be the dual of \(H\). The author describes the notable elements of \(H^*\) in terms of those of \(H\), and as a consequence he proves that for \(H\) almost symmetric, \(H^*\) is almost symmetric if and only if \(\mathrm m(H)=\mathrm t(H)+\mathrm t(H^*)\). Barucci and Fröberg proved that if \(H\) has maximal embedding dimension, then \(H\) is almost symmetric if and only if \(H^*\) is symmetric. With the tools introduced in Section 3, a new proof of this fact is presented. Also, it is shown that for \(H\) almost symmetric, \(\mathrm e(H)=\mathrm m(H)-1\) if and only if \(H^*\) is almost symmetric and not symmetric. Let \(I\) be an interval of real numbers with nonempty interior. Define \(\mathrm S(I)=(\bigcup_{k\in\mathbb N}kI)\cup\mathbb N\). Then \(\mathrm S(I)\) is a numerical semigroup and there exists \(a,b,c\in\mathbb N\), \(c<a<b\), such that \(\mathrm S(I)\) is the set of nonnegative integer solutions to \(ax\bmod b\leq cx\). A numerical semigroup is proportionally modular if it is of this form. Indeed \(I\) can be taken to be \([b/a,b/(a-c)]\). If \(c=1\), then the semigroup is said to be modular. If either \(H\) is a half-line or \(H=\mathrm S(]b/a,b/(a-1)[)\), then \(H\) is called an opened modular numerical semigroup. The author proves that \(\mathrm S(]b/a,b/(a-1)[)^*=\mathrm S([b/a,b/(a-1)])\) and that every opened modular numerical semigroup is almost symmetric. For embedding dimension three, almost symmetry coincides with irreducibility. Every irreducible numerical semigroup is either symmetric (\(\mathrm{PH}=\{\mathrm F(H)\}\)) or pseudo-symmetric (\(\mathrm{PH}=\{\mathrm F(H)/2,\mathrm F(H)\}\)). Symmetric embedding dimension three proportionally modular numerical semigroups were characterized by \textit{J. C. Rosales}, \textit{P. A. García-Sánchez} and \textit{J. M. Urbano-Blanco} [J. Number Theory 128, No. 3, 453-476 (2008; Zbl 1185.11023)]. The author gives a description for the pseudo-symmetry, having in this way a characterization of all almost symmetric proportionally modular numerical semigroups with embedding dimension three. Let \(A\) be the minimal generating set of \(H\), and let \(A_1\cup A_2=A\) be a non-trivial partition. Set \(d_i=\gcd(A_i)\), and \(H_i\) be the numerical semigroup generated by \(A_i/d_i\), \(i\in \{1,2\}\). Then \(A=d_1H_1+d_2H_2\) and \(\gcd(d_1,d_2)=1\). We say that \(H\) is a gluing of \(H_1\) and \(H_2\) if \(d_1\in H_2\), \(d_2\in H_1\) and they are not minimal generators. This concept is inspired in \textit{C. Delorme}'s paper [Ann. Sci. Éc. Norm. Supér. (4) 9, 145-154 (1976; Zbl 0325.20065)] though the term gluing was first introduced by \textit{J. C. Rosales} [Semigroup Forum 55, No. 2, 152-159 (1997; Zbl 0951.20042)] (Rosales generalized the idea of ``suite distingué'' to affine semigroups; indeed Rosales also showed in his paper that Lemma 6.5 can be seen as a characterization of gluing). The author proves that the type of \(H\) is the product of the types of \(H_1\) and \(H_2\), and that the gluing of almost symmetric numerical semigroups is never almost symmetric if either \(H_1\) or \(H_2\) is not symmetric. This contrasts with the symmetric case: Delorme showed that gluings of symmetric numerical semigroups are always symmetric. The manuscript contains a good bunch of examples illustrating the results presented.