Grothendieck groups, convex cones and maximal Cohen-Macaulay points (Q2231125)

Let \(R\) be a commutative Noetherian ring and let \(\mathrm{G}_0(R)\) be the Grothendieck group of finitely generated \(R\)-modules. Let \(\mathrm{H}(R)\) be the quotient of \(\mathrm{G}_0(R)\) by the subgroup generated by pseudo-zero modules i.e. modules \(M\) such that \(\text{ht}(\text{ann}(M))\geq 2\). The ambient Euclidean space in which majority of the work in this paper is carried out is \(\mathrm{H}(R)_{\mathbb{R}}:=\mathrm{H}(R)\otimes_{\mathbb{Z}}\mathbb{R}\), under the assumption that it is finite dimensional. The primary objects of study are \(\mathrm{C}(R)\) and \(\mathrm{C}_r(R)\), the convex cones in \(\mathrm{H}(R)_{\mathbb{R}}\) spanned by maximal Cohen-Macaulay (MCM) modules and MCM modules of rank \(r\) respectively. Topological properties of \(\mathrm{C}_r(R)\) are used to characterize the condition that \(R\) has only finitely many rank \(r\) MCM points (defined below) in \(\mathrm{C}_r(R)\). By viewing MCM modules of rank one as elements of the divisor class group \(\mathrm{Cl}(R)\), a direct approach to a conjecture of \textit{H. Dao} and \textit{K. Kurano} [Math. Ann. 364, No. 3--4, 713--736 (2016; Zbl 1346.13020)] is also considered. \par The motivation for this work is as follows. \textit{K. Kurano} [Invent. Math. 157, No. 3, 575--619 (2004; Zbl 1070.14007)] introduced and developed the notion of Grothendieck groups modulo numerical equivalence, denoted \(\overline{\mathrm{G}_0(R)}\). \textit{C-Y. J. Chan} and \textit{K. Kurano} [Trans. Am. Math. Soc. 368, No. 2, 939--964 (2016; Zbl 1342.13016) ] introduce and explore the Cohen-Macaulay cone \(\mathrm{C}_{\mathrm{CM}}(R)\), which is defined as the convex cone spanned by MCM modules in \(\overline{\mathrm{G}_0(R)}\otimes_{\mathbb{Z}}\mathbb{R}\). When \(R\) is a Cohen-Macaulay local ring of dimension at most three, the canonical map from \(\mathrm{G}_0(R)\) to \(\overline{\mathrm{G}_0(R)}\) factors through \(\mathrm{H}(R)\). This connection motivates the study of the objects in consideration in this paper. \par Section 2 of the paper deals with establishing properties of the group \(\mathrm{H}(R)\) and the space \(\mathrm{H}(R)_{\mathbb{R}}\), while section 3 proves various results concerning the interior, closure and boundary of \(\mathrm{C}(R)\). Section 4 is motivated by the open question as to whether \(\mathrm{C}_{\mathrm{CM}}(R)\) is polyhedral. In Theorem 4.1, the author considers the polyhedrality of \(\mathrm{C}(R)\) and under certain technical assumptions constructs a ``large'' convex polyhedral subcone of \(\mathrm{C}(R)\) when a module finite extension of \(R\) has a polyhedral cone. The author shows that the result applies to Gorenstein normal local rings possessing a simple (ADE) simgularity as a finite extension. \par Sections 5 and 6 approach the aforementioned conjecture of Dao-Kurano, which says that under mild hypothesis a Cohen-Macaulay normal local ring has only finitely many rank one MCM modules up to isomorphism. In Theorem 5.4, the author gives characterizations of the existence of only finitely many rank \(r\) MCM points in \(\mathrm{C}_r(R)\) in terms of the topological properties of \(\mathrm{C}_r(R)\). Here a MCM point of rank \(r\) refers to a point represented by a rank \(r\) MCM module. Section 6 approaches the conjecture more directly by giving sufficient conditions for the finiteness of the number of MCM points on the line defined by a reflexive ideal in \(\mathrm{Cl}(R)\), when \(R\) is Gorenstein. More precisely, Theorem 6.8 says that if \(R\) is Gorenstein local of dimension at least two with an isolated singularity and \(I\) is a non-zero reflexive ideal of \(R\) satisfying certain conditions on cohomological dimension or asymptotic depth, there exist only a finite number of MCM points on \(\mathbb{Z}I\) in \(\mathrm{Cl}(R)\). The author gives several examples where the hypothesis of Theorem 6.8 is satisfied. On the other hand, Theorem 6.14 assumes \(R\) is a Gorenstein normal local ring and determines the MCM points on the line defined in \(\mathrm{Cl}(R)\) by a height one, rigid ideal such that \(R/I\) is Gorenstein.