Gorenstein biliaison and ACM sheaves (Q1879670)

The equivalence relation of liaison has been studied classically for complete intersection links (CI-liaison), and more recently for arithmetically Gorenstein links (G-liaison). For codimension two the two notions coincide, and the theory is well understood. In higher codimension, there are striking results that suggest that G-liaison may have more similarities to the codimension two case than CI-liaison. See the reviewer's book ``Introduction to liaison theory and deficiency modules'' [Progress in Mathematics 165, Birkhäuser, Boston (1998; Zbl 0921.14033)] for details. For technical reasons, it is more reasonable to study the equivalence relation generated by an \textit{even} number of links, and linearly equivalent divisors on arithmetically Cohen-Macaulay (ACM) varieties are evenly linked. These ideas lead to the notions of CI-biliaison and G-biliason respectively. An important open question is whether the codimension two theorems due to \textit{P. A. Rao} [Invent. Math. 50, 205--217 (1979; Zbl 0406.14033) and Math. Ann. 258, 169--173 (1981; Zbl 0493.14009)] that classify the even liaison classes carry over to G-biliaison in higher codimension, even though it is known that they do not carry over to CI-biliaison. The special cases of most interest are (1) whether ACM subschemes (e.g.\ all zero-dimensional schemes) of the same codimension in \(\mathbb P^n\) are always in the same biliaison class, and (2) for curves, whether the deficiency module (a.k.a. the Rao module) completely determines the G-biliaison class. The current paper makes a nice contribution to the theory. The authors study G-biliaison for codimension two subschemes of a normal ACM scheme, \(X\), in \(\mathbb P^n\). They give a new approach by relating it to the category of ACM sheaves on \(X\). (These are sheaves that correspond to the graded maximal Cohen-Macaulay modules on the homogeneous coordinate ring of \(X\).) Their main results give a criterion for all codimension two ACM subschemes of \(X\) to be in the same G-biliaison class, in terms of the ACM sheaves on \(X\). For certain choices of \(X\), they use their theory to give affirmative answers to the special cases mentioned above for codimension two subschemes of \(X\). In particular, their new approach also allows them to extend work of \textit{J. Lesperance} [Collect. Math. 52, 219--230 (2001; Zbl 1074.14527)] concerning biliaison classes of certain curves in \(\mathbb P^4\). On the other hand, they show that if \(X\) is the Veronese surface in \(\mathbb P^5\), there are infinitely many G-biliaison classes of zero-dimensional schemes on \(X\) (but this is not a counterexample to the question since classes may be joined by leaving \(X\)).