Gorenstein and totally reflexive orders (Q517055)

The main objective of the article is to study orders over Cohen-Macaulay rings. \textit{M. van den Bergh} [in: The legacy of Niels Henrik Abel. Papers from the Abel bicentennial conference, University of Oslo, Oslo, Norway, June 3--8, 2002. Berlin: Springer. 749--770 (2004; Zbl 1082.14005)] introduced the notion of a noncommutative crepant resolution of a commutative ring \(R\). This is a not necessarily commutative ring \(\text{End}_R(M)\) for some \(R\)-module \(M\) such that \(\text{End}_R(M)\) itself is a maximal Cohen-Macaulay \(R\)-module. When \(R\) is a Gorenstein ring, this leads to strong results and connections to geometric notions. For instance, appropriate conditions on \(R\) gives that \(R\) having a non-commutative crepant resolution is equivalent to \(\text{Spec}R\) having a commutative crepant resolution. This article considers a construction in the non-Gorenstein case which gives back the main results from that case. Dao, Ingalls, Faber, Iyama, Takahashi, Vial and Wemyss apply the definition of a noncommutative crepant resolution in the Gorenstein case directly to the non-Gorenstein case. Then some of the corresponding results hold, others don't. In this article, the author gives an alternative definition of a noncommutative resolution. The condition of \(\text{End}_R(M)\) being maximal Cohen-Macaulay is replaced by the stronger condition of total reflexivity. An \(R\)-module \(M\) is totally reflexive if it is reflexive, and \(\text{Ext}^i_R(M,R)=\text{Ext}^i_R(M^\ast,R)=0\) for all \(i>0\). This condition certainly gives the main results from the Gorenstein case back, the problem is that such orders cannot exist over non-Gorenstein rings. Following this, the author discusses the Gorenstein property of an order, which holds for a noncommutative crepant resolution over Gorenstein \(R\), and give necessary and sufficient conditions for \(\text{End}_R(R\oplus\omega)\) to be a Gorenstein order when \(R\) is Cohen-Macaulay with canonical module \(\omega\). The last issue of the article is to use results of \textit{O. Iyama} and \textit{Y. Nakajima} [J. Noncommut. Geom. 12, No. 2, 457--471 (2018; Zbl 1419.16012)] to find the global dimension of \(\text{End}_R(R\oplus\omega)\) in the case of abelian quotient singularities, starting with a review of the case of \(R\) being Gorenstein. The article gives a nice review on noncommutative crepant resolutions, with a good illustration of the issues when trying to generalize to the non-Gorenstein case. The examples are explicit and are very clearly explained.