A short proof of Fel's theorems on 3-D Frobenius problem (Q1035295)

Let \(d^3 = (d_1, d_2, d_3)\) be three natural numbers such that \(\gcd(d_1,d_2, d_3) = 1\). The numerical semigroup \(S(d^3) = S\) is defined as the submonoid of \(\mathbb{N}\) given by \(\{d_1 x + d_2 y + d_3 z :\;x,y,z \in \mathbb{N}_0\}\), and has finite complement \(\Delta(S)\). The largest gap \(F(S)\) is called the Frobenius number \(F(S)\). In this paper the author studies the Hilbert series \(H(S) = \sum_{t \in S} t^s\), giving a shorter proof of formulas for \(H(S)\) and \(F(S)\) that were given by \textit{L. G. Fel} [Funct. Anal. Other Math. 1, No. 2, 119--157 (2006; Zbl 1194.20058)]. The result for \(F(S)\) builds on work of \textit{S. M. Johnson} [Can. J. Math. 12, 390--398 (1960; Zbl 0096.02803)]. The Apéry set of \(S\) with respect to the generator \(d_1\) is the set of elements of \(t\in S\) such that \(t-d_1\) is a gap. It is clear that this Apéry set determines \(H(S)\) and \(F(S)\). The set of pseudo-Frobenius numbers of \(S\) are those \(t \in \Delta(S)\) such that \(t+s \in S\) for all \(s\in \{S \setminus 0\}\). The cardinality of this set is called the type of \(S\). It is clear that the Frobenius number is the maximum of the pseudo-Frobenius numbers. A semigroup is called symmetric if for all \(t \in \Delta(S),\;F(S) - t \in S\). It is well known that symmetric semigroups are exactly those of type one. Moreover, a nonsymmetric semigroup given by three generators must have type two. See for example, the book of [\textit{J. C. Rosales, P. A. Garcia-Sánchez}, Numerical semigroups. Developments in Mathematics 20. Dordrecht: Springer (2009; Zbl 1220.20047)]. The proofs in this paper use a technique introduced by \textit{V. I. Arnold} [Funct. Anal. Other Math. 2, No. 2-4, 129--138 (2009; Zbl 1201.11009)], and extended by the author [Funct. Anal. Other Math. 2, No. 2--4, 111--127 (2009; Zbl 1208.11039)], which relates \(S\) to certain finite subsets of \(\mathbb{N} \times \mathbb{N}\), called the \(D_{d_i}\), and uses the geometry of these sets to determine the Apéry set and the pseudo-Frobenius numbers of the semigroup \(S\). Unfortunately, to understand the argument it is important to read it together with either of the two papers mentioned above as not all of the key concepts, such as the domains \(D_{d_i}\) or the term symmetric, are defined here.