On the Apéry sets of monomial curves (Q1955765)

Let \(S\subseteq{\mathbb N}_0\) be a numerical semigroup, that is an additive semigroup having finite complement in \({\mathbb N}_0\), and let \(n_1<\cdots<n_b\) be its unique minimal generating set. Letting \(K\) be a field, the ring \({\mathcal S}:=K[T^{n_1},\dots,T^{n_b}]\subseteq K[T]\) is the coordinate ring of the curve in affine \(b\)-space parametrised by the monomials \(X_i=T^{n_i}\), for \(1\leq i\leq b\). In continuation of earlier work by the same authors [Contemp. Math. 502, 45--58 (2009; Zbl 1183.13003)], the topic of the present paper is a further study of the local properties of \({\mathcal S}\) at the origin, in terms of the arithmetic properties of \(S\). To this end, notation and results from [loc. cit.] are first recalled in detail: Let \({\mathcal R}\) be the completion of \({\mathcal S}\) with respect to the \(T\)-adic valuation \(\nu_T\), let \({\mathcal M}:=\langle T^{n_1},\dots,T^{n_b}\rangle\subset{\mathcal R}\) be its unique maximal ideal, and let \({\mathcal G}:=\bigoplus_{n\geq 0}{\mathcal M}^n/{\mathcal M}^{n+1}\) be the associated graded local ring. An ideal \(I\) of \(S\) is a non-empty subset such that \(I+S\subseteq S\), and letting \(e:=n_1\), the Apéry set of \(I\) is defined as \(A(I):=\{i\in I;i-e\not\in I\}\); hence we have \(A(I)=\{\omega^I_0,\dots,\omega^I_{e-1}\}\), where \(\omega^I_i\equiv i\pmod{e}\). Then, letting \(M:=S\setminus\{0\}\) be the unique maximal ideal of \(S\), we have \(\nu_T({\mathcal M}^n\setminus\{0\})=nM\), the \(n\)-fold sum of \(M\) with itself, where \(0\cdot M:=S\). Letting \(A(nM)=\{\omega^n_0,\ldots,\omega^n_{e-1}\}\), associated with \(S\) we get the Apéry matrix \([\omega^n_i]_{n,i}\), with rows and columns indexed by \(n\in{\mathbb N}_0\) and \(0\leq i\leq e-1\), respectively. To elucidate the stucture of \({\mathcal G}\), let \({\mathcal F}:=\bigoplus_{n\geq 0}T^{en}{\mathcal R}/T^{en}{\mathcal M}\), which can be seen as a univariate polynomial ring, in the variable \(T^e\). The natural embedding \({\mathcal F}\rightarrow{\mathcal G}\) being a finite ring extension, the indecomposable direct sum decomposition of \({\mathcal G}\) as an \({\mathcal F}\)-module can be read off from the Apéry matrix, as can be the Cohen-Macaulay and Gorenstein properties and the Hilbert function of \({\mathcal G}\). Having this in place, and the case \(b=2\) being well-understood, the paper now focusses on the case \(b=3\): the Apéry matrix method yields simple combinatorial criteria when \({\mathcal G}\) is Cohen-Macaulay; Buchsbaum, but not Cohen-Macaulay; \(2\)-Buchsbaum, but not Buchsbaum; and new proofs of earlier results by \textit{V. A. Sapko} [Commun. Algebra 29, No. 10, 4759--4773 (2001; Zbl 1051.13005)], \textit{Y.-H. Shen} [Commun. Algebra 39, No. 5, 1922--1940 (2011; Zbl 1264.13007)] and \textit{M. D'Anna} et al. [J. Commut. Algebra 3, No. 2, 147--168 (2011; Zbl 1237.13014)]. Moreover, this method shows that \({\mathcal G}\) has non-decreasing Hilbert function in the case \(b=3\), and in the case \(b=4\) whenever \({\mathcal G}\) is Buchsbaum.