Generalized divisors on Gorenstein schemes (Q1337457)

As is well-known, the concept of divisors lies in the center of algebraic geometry, and its theory is well established for normal integral schemes. In recent times, the needs are felt to have a more general theory of ``generalized divisors'' over nonnormal schemes, e.g., in the classification theory of space curves. The author developed previously such a theory for Gorenstein curves. In this paper, he describes a satisfying general theory of generalized divisors. The schemes \(X\) under considerations are not necessarily integral but are only assumed to satisfy the following condition: the local rings \({\mathfrak O}_{X,x}\) are Gorenstein whenever \(\text{depth} ({\mathfrak O}_{X,x}) \leq 1\), i.e., \({\mathfrak O}_{X,x}\) is ``quasi-normal'' in the sense of Vasconcelos. After necessary preliminaries concerning reflexive modules, Gorenstein rings and duality, a generalized divisor is defined to be a reflexive coherent \({\mathfrak O}_X\)-submodule \(I\) of \({\mathcal K}_X\), the sheaf of total quotient rings of \({\mathfrak O}_X\). A generalized divisor \(I\) is said to be Cartier (resp. almost Cartier) if it is invertible (resp. invertible out of a closed subset of codimension \(\geq 2)\). The sum and the inverse of generalized divisors are defined, but in general the set of generalized divisors does not form a group. The set \(\text{GPic} (X)\) and the groups \(\text{APic} (X)\), \(\text{Pic} (X)\) of linear equivalence classes of generalized divisors, almost Cartier divisors and Cartier divisors are defined, and the general properties of these set and groups are examined in detail. As an application (or a dessert?) of the general theory, the author brushes up the foundations of the theory of liaison (or linkage) of subschemes of projective space, which have recently attracted some attention both in algebraic geometry and commutative algebra. The general theory is illustrated in the case of curves, some singular surfaces in \(\mathbb{P}^3\), and ruled cubic surfaces in detail, and the divergences from the classical theory of divisors on normal schemes are carefully examined. This clear exposition of the fundamental notions will be expected to provide useful tools in the study of algebraic geometry in future.