Geometric algebras on projective surfaces (Q615852)

The aim of the article is to study a broader class of geometric algebras than have been studied so far. These algebras are twisted homogeneous algebras and their subalgebras. The ground field \(\mathbf{k}\) is algebraically closed, and if \(X\) is a projective variety, \(\sigma\in\text{Aut}(X)\), and \(\mathcal L\) a quasicoherent sheaf on \(X\), \(\mathcal L^\sigma:=\sigma^\ast\mathcal L\). The geometric algebras studied in this article are given by geometric data: The tuple \(\mathbb D=(X,\mathcal L,\sigma,\mathcal A,\mathcal D,\mathcal C,s)\) is ADC data if \(X\) is a projective surface, \(\sigma\) an automorphism of \(X\), \(\mathcal L\) an invertible sheaf on \(X\), \(s\) a positive integer, \(\mathcal D\) is the ideal sheaf of a \(0\)-dimensional subscheme of \(X\) such that all points in the cosupport of \(\mathcal D\) have distinct infinite \(\sigma\)-orbits, and \(\mathcal A\) and \(\mathcal C\) are ideal sheaves on \(X\) such that the cosupport of \(\mathcal C\) is \(0\)-dimensional and \(\mathcal A\mathcal C\subseteq\mathcal D\mathcal D^\sigma\cdots\mathcal D^{\sigma^{s-1}}\). Given ADC data \(\mathbb D=(X,\mathcal L,\sigma,\mathcal A,\mathcal D,\mathcal C,s)\), the sheaves \(\mathcal T_n\) are defined by setting \(\mathcal T_0=\mathcal O_X\) and \(\mathcal T_n=\mathcal A\mathcal D^{\sigma^{s}}\cdots\mathcal D^{\sigma^{n-1}}\mathcal C^{\sigma^n}\cdot\mathcal L\otimes\mathcal L^\sigma\otimes\cdots\otimes\mathcal L^{\sigma^{n-1}}\) for \(n\geq 1\). \(\mathcal T_n\) satisfy \(\mathcal T_n\mathcal T_m^{\sigma^{n}}\subseteq\mathcal T_{n+m}\), and a \(k\)-algebra \(T(\mathbb D):=\bigoplus_{n\geq 0}H^0(X,\mathcal T_n)\) is constructed by letting the multiplication be given by \(H^0(X,\mathcal T_n)\otimes H^0(X,\mathcal T_m)\overset{1\otimes\sigma^n}\rightarrow H^0(X,\mathcal T_n\otimes\mathcal T_m^{\sigma^n})\rightarrow H^0(X,\mathcal T_{n+m})\). These rings called ADC rings are the articles main object of study. These algebras occur naturally in the classification of noncommutative projective surfaces; connected \(\mathbb N\)-graded noetherian domains of GK-dimension \(3\). The algebras occurring is described by a geometric condition on the ADC data: When \(\sigma\in\text{Aut}(X)\), \(Z\) is a closed subscheme of \(X\), and \(A\subset\mathbb Z\) is finite, then \(\{\sigma^n Z\}_{n\in A}\) is \textit{critically transverse} if for all closed subschemes \(Y\) of \(X\), we have for all but finitely many \(n\in A\) that \(\mathcal Tor^{X}_i(\mathcal O_{\sigma^n Z},\mathcal O_Y)=0\) for all \(i\geq 0\). Let \(\mathbb D=(X,\mathcal L,\sigma,\mathcal A,\mathcal D,\mathcal C,s)\) be ADC data. Let \(Z\) be the subscheme of \(X\) defined by \(\mathcal D\) and let \(\Gamma\) be the subscheme defined by \(\mathcal C\). Using a primary decomposition of \(\mathcal A\), write \(\mathcal A=\mathcal I_\Omega\cap\mathcal I_\Lambda\) where \(\Omega\) is a curve and \(\mathcal I_\Lambda\) is maximal with respect to this condition. \(\mathbb D\) is called \textit{transverse} if the sets \(\{\sigma^n Z\}_{n\in\mathbb Z}\), \(\{\sigma^n \Omega\}_{n\in\mathbb Z}\), \(\{\sigma^n \Lambda\}_{n\geq 0}\), and \(\{\sigma^n \Gamma\}_{n\leq 0}\) are critically transverse. In an earlier article the author proves that if the ground field is uncountable, if \(R\) is a connected \(\mathbb N\)-graded noetherian domain of GK-dimension \(3\) such that \(Q_{\text{gr}}(R)\cong K[z,z^{-1};\sigma]\) for a field K and automorphism \(\sigma\) of \(K\), then there are transverse data \(\mathbb D=(X,\mathcal L,\sigma,\mathcal A,\mathcal D,\mathcal C,1)\) where \(\mathcal L\) \(\sigma\)-ample, and an integer \(k\) so that \(R^{(k)}\cong T(\mathbb D).\) The algebras \(T(\mathbb D)\) are therefore of interest. The main result in this article is easily stated: If \(\mathcal L\) is \(\sigma\)-ample, then transversality of the ADC data \(\mathbb D\) is necessary and sufficient for \(T(\mathbb D)\) to be noetherian. In addition, the author proves that in this situation, with \(T=T(\mathbb D)\), the functor \(\text{Hom}_{\text{qgr-}T}(T,-)\) has finite cohomological dimension, and further, qgr-\(T\) depends only on \(X\),\(\sigma\) and \(\mathcal D\). By defining a maximality condition on \(\mathbb D\), A nice application of this theory is that if \(\mathbb D=(X,\mathcal L,\sigma,\mathcal A,\mathcal D,\mathcal C,1)\) are maximal transverse ADC data, where \(X\) is a normal surface and \(\mathcal L\) is \(\sigma\)-ample, then \(T(\mathbb D)\) is a maximal order. The author recalls the definition and basic properties of bimodule algebras, and study the Artin-Zhang \(\chi\)-conditions for \(T(\mathbb D)\). Going through the article, this is a good consistent theory. The definitions are explicit, the propositions clear, and the proofs are detailed and thoroughly treated. This is a nice article for those who would understand the subject. Also, a surprising observation by the author is that developing techniques in graded ring theory that do not require the algebras to be generated in degree one may be important for future research.