Gorenstein liaison of 0-dimensional schemes (Q1401494)

Liaison theory concerns an equivalence relation among equidimensional projective subschemes of the same dimension. This equivalence relation is generated by {direct links}. Roughly, schemes \(V\) and \(V'\) are directly linked if their union is arithmetically Gorenstein. Liaison theory has seen a surge of activity since it was placed in a modern context by \textit{C. Peskine} and \textit{L. Szpiro} [Invent. Math. 26, 271--302 (1974; Zbl 0298.14022)] in 1974. The majority of results for the first two decades dealt with the special case where the links were in fact via complete intersections (CI-liaison), but in the last 7 years or so much of the focus has changed to the more general case (G-liaison), beginning with the memoir by \textit{J. O. Kleppe}, \textit{J. C. Migliore}, \textit{R. Miró-Roig}, \textit{U. Nagel} and \textit{C. Peterson} [``Gorenstein liaison, complete intersection liaison invariants and unobstructedness'', Mem. Am. Math. Soc. 732 (2001; Zbl 1006.14018)], although the groundwork had been laid by \textit{P. Schenzel} [J. Math. Kyoto Univ. 22, 485--498 (1982; Zbl 0506.13012)] in 1982/83. The most complete picture is for codimension two, where complete intersections and Gorenstein ideals coincide. However, recent work gives some hope that changing the focus to Gorenstein liaison will allow us to prove results for higher codimension that are analogous those in codimension two, even though the same statements are false for CI-liaison in higher codimension. One particularly nice codimension two result is a classical result due independently to Apery and to Gaeta (which now is a special case of a much bigger theory): in modern language it says that for zero-dimensional subschemes of \({\mathbb P}^2\), there is only one liaison class. This is known to be false in higher codimension, if we restrict to CI-liaison. One of the most intriguing questions at the moment is whether there is just one G-liaison class for zero-dimensional subschemes of higher projective space. (This is in turn a special case of a more general open question which we omit here.) The author, together with \textit{R. Miró-Roig}, has proved many striking results for G-liaison of subschemes of rational normal scrolls [J. Pure Appl. Algebra 164, 325--343 (2001; Zbl 1070.14522)]. In this paper, the author takes a step toward answering the above question. She proves that a {general} set of points on a smooth rational normal scroll surface is in the G-liaison class of a complete intersection (i.e. {glicci}). This generalizes a result of \textit{R. Hartshorne} [Collect. Math. 53, No. 1, 21--48 (2002; Zbl 1076.14065)] for points on a quadric surface. As a consequence, the author proves that a general set of \(\leq N+4\) points in \({\mathbb P}^N\) is glicci. The Gorenstein links are produced using the theory developed in the memoir of Kleppe et al. mentioned above, which develops G-liaison as a theory of divisors.