Codimension 3 arithmetically Gorenstein subschemes of projective \(N\)-space (Q954815): Difference between revisions

The authors study Gorenstein linkage of zero-dimensional subschemes of \(\mathbb{P}^3\). Recall that two subschemes of \(\mathbb{P}^n\) without common irreducible components whose union is a complete intersection (resp. an arithmetically Gorenstein, AG, scheme) are said to be \textit{CI-linked} (resp. \textit{G-linked}). The corresponding algebraic definition allows linkage between two arithmetically Cohen-Macaulay (ACM) schemes having common components. CI-liaison is rather well understood in codimension \(2\), but many interesting problems are open and seem more natural for G-liaison than for CI-linkage in higher codimensions. E.g., is it true that all ACM schemes, say of codimension \(3\) in \(\mathbb{P}^n\), is in the G-liaison class of a complete intersection (i.e. is glicci)? This question was asked by the reviewer, \textit{J.C. Migliore, R.M. Miro-Roig, U. Nagel} and \textit{C. Peterson} in [Mem. Am. Math. Soc. 732, (2001; Zbl 1006.14018)] and has initiated an interesting research activity. The present paper is concerned with glicciness of zero-schemes in which they often link with AG schemes of the form \(mH_X-K_X\) on some ACM curve \(X\) where \(H_X\) and \(K_X\) are the hyperplane and the canonical divisor class respectively. They show that a scheme consisting of \(n \geq 56\) general points in \(\mathbb{P}^3\) admits no strictly descending G-liaison or biliaison. This result does not imply non-glicciness, but it shows that one must first G-link to a larger set before eventually linking in possibly several steps to a CI (and further down to a point). Moreover they study in detail the problem of how many general points one can impose on a general AG zero-scheme, and they give a new proof for a result of the first author that a set of \(n \leq 19\) general points in \(\mathbb{P}^3\) is glicci. Indeed it is shown that such a set admits strictly descending G-liaisons down to a point. Furthermore they show that a general codimension 3 AG zero-scheme admits strictly descending CI-liaisons to a point, thus sharpening a well known licciness result of \textit{J. Watanabe} in this case [Nagoya Math.\ J.\ 50, 227--232 (1973; Zbl 0242.13019)]. To achieve their results they need to study the irreducible smooth family, PGor\((h)\), of AG codimension 3 subschemes of \(\mathbb{P}^n\). In particular they show that, under a numerical assumption, the general element of PGor\((h)\) is of the form \(mH_X-K_X\) above, and they give an inductive formula for dim PGor\((h)\) fitting their applications. As the authors remark, these results are mainly known, cf. \textit{M. Boij \ } [Bull. Lond. Math. Soc. 31, No. 1, 11--16 (1999; Zbl 0940.13008)] and the paper [Trans. Am. Math. Soc. 358, No. 7, 3133--3167, (2006; Zbl 1103.14005)] by the reviewer. Note that the reference [21] appearing several times in the paper should be to [J. Algebra 200, No. 2, 606--628 (1998; Zbl 0928.14005)] of the reviewer.