Moduli spaces for point modules on naïve blowups (Q365890)

An essential contribution to noncommutative projective geometry is the classification of noncommutative projective planes by Artin, Tate and Van den Bergh. Noncommutative projective planes are what is known as Artin-Schelter regular algebras of dimension 3, noncommutative graded rings taking the place of graded polynomial algebras in three variables. To classify the noncommutative planes \(R\), the main method is to study the point modules: cyclic graded modules with the Hilbert series of a point in projective space. Constructing a moduli scheme for the points modules of \(R\), a morphism of \(R\) into a known ring is obtained. This gives information of the structure of the noncommutative ring, or plane, itself. This technique works in a more general way. For an algebraically closed field \(k\), a \(k\)-algebra \(R\) is called strongly noetherian if, for any commutative noetherian \(k\)-algebra \(C\), \(R\otimes_k C\) is again noetherian. A general result states that if \(R\) is a strongly noetherian \(\mathbb N\)-graded \(k\)-algebra, then its point modules are parametrized by a projective scheme. Then it is possible to construct a morphism from a strongly connected \(k\)-algebra \(R\) (\(R_0\simeq k\)) to a twisted homogeneous coordinate ring on the scheme \(X\) parametrizing point modules. A class of noncommutative graded algebras, known as naïve blowups, are constructed that are noetherian, but no strongly noetherian. The point modules over naïve blowups, viewed as objects of noncommutative projective geometry, cannot behave well in families. It is proved that there is no fine moduli scheme of finite type for such modules. This article develops the moduli theory of point modules for naïve blowups S. The authors prove that there is an analog of a Hilbert scheme of one point on \(\text{Proj}(S)\) that is an infinite blowup of a projective variety. This is a quasicompact and noetherian fpqc-algebraic stack. Also, they prove that there is a coarse moduli space for one point on \(\text{Proj}(S)\) which is the projective variety from which the naïve blowup is constructed. More precisely, let \(X\) be a projective variety over \(k\) of dimension \(\geq 2\). Let \(\sigma:X\overset\sim\rightarrow X\) be an automorphism, and let \(\mathcal L\) be a \(\sigma\)-ample invertible sheaf. Let \(P\in X\) be such that its \(\sigma\)-orbit is infinite with dense infinite subsets (critically dense). For \(n\geq 0\), let \(\mathcal I_n=\mathcal I_P\mathcal I_P^\sigma\cdots\mathcal I_P^{\sigma^{n-1}}\) and \(\mathcal L_n=\mathcal L\otimes\mathcal L^\sigma\otimes\cdots\mathcal L^{\sigma^{n-1}}\), where \(\mathcal L^\sigma=\sigma^\ast\mathcal L\). Define \(\mathcal S=\mathcal I\bigoplus\mathcal L_n\), and let \(S\) be its ``global sections'': \(S=S(C,\mathcal L,\sigma,P)=\underset{n\geq 0}\oplus H^0(X,\mathcal S_n)\). Then \(S\) is the naïve blowup associated to the data \((X,\mathcal L,\sigma,P)\) (\(S\) is supposed to be generated in degree one). A \textit{point module} is a graded cyclic \(S\)-module \(M\) with Hilbert series \(1+t+t^2+\cdots.\) \(M\) is an \textit{embedded point module} if a surjection of graded modules \(S\rightarrow M\) of graded modules are given, inducing the concept of equivalence of embedded point modules. The article starts by constructing the moduli stack \(X_{\infty}\) for embedded point modules: There is an inverse system of schemes \(\cdots\rightarrow X_n\rightarrow X_{n-1}\rightarrow\cdots\rightarrow X\) where \(X_n\) is the blowup of \(X\) at \(\mathcal I_n\), and \(X_\infty=\underset\longleftarrow\lim X_n\). Then \(X_\infty\) is a noetherian fpqc-algebraic stack, \(X_\infty\rightarrow X\) is quasicompact. \(X_\infty\) is a fine moduli space for embedded point modules. In addition, \(X_\infty\) is discrete, its points have no stabilizers. so \(X_\infty\) is a \(k\)-space. However, \(X_\infty\) does not seem to have an étale cover by a scheme and so is not an algebraic space. By definition, the noncommutative projective scheme associated to \(S\) is the quotient category Qgr-\(S\)=Gr-\(S\)/Tors-\(S\) of graded right \(S\)-mdoules by the full subcategory of locally bounded modules. A \textit{point object} in Qgr-\(S\) is the image of a shift of a point module. If \(S\) is a commutative graded algebra generated in degree 1, Qgr-\(S\) is equivalent to the category of quasicoherent sheaves on \(\text{Proj}(S)\), this is why Qgr-\(S\) is the noncommutative analogue of a projective scheme. If \(R\) is strongly noetherian and generated in degree 1, Artin and Stafford shows that point objects of Qgr-\(R\) are parametrized by the same projective scheme \(X\) that parametrize embedded point modules. For naïve blowups \(S=S(X,\mathcal L,\sigma,P)\) the main results are the following: \hskip0,2cm - The algebra \(S\) is notherian but not strongly noetherian. There is no fine moduli scheme of finite type over \(k\) parametrizing point objects of Qgr-\(S\). - The variety \(X\) is a coarse moduli scheme for point objects in Qgr-\(S\). - There is a fine moduli space \(X_\infty\) for embedded \(S\)-point modules of \(S\). \hskip0,2cm The authors recall the standard notation for module categories over rings and bimodule algebras, to measure ampleness and to study ample sequences, they use, and define, relative Castelnuovo-Mumford regularity. Included, and needed for proving the main results, general properties of infinite blowups are proved. It should be mentioned that the authors focuses on infinite blowups as the limits rather than as pro-objects. The theme of the article is one of the more sophisticated, also philosophically, but the authors treat this in a very nice way, with references to standard literature. This article gives a good oversight over noncommutative geometry, and in particular makes the generalization of commutative, projective planes more visible.