Multiple structures with arbitrarily large projective dimension supported on linear subspaces (Q897761)

The authors investigate the possibility of a finite characterization of multiple structures in projective space \(\mathbb P^n_k\), where \(k\) is an algebraically closed field, defined on linear subspaces of small codimension under the assumption of unmixedness. \textit{N. Manolache} gave such a characterization for scheme-theoretically Cohen-Macaulay multiple structures of degree at most 4 [Math. Z. 210, No. 4, 573--579 (1992; Zbl 0784.14020)] as well as for locally complete intersection multiple structures of degree at most 6 [\textit{N. Manolache}, Math. Z. 219, No. 3, 403--411 (1995; Zbl 0837.14039)]. In addition, \textit{B. Engheta} gave a characterization of unmixed ideals of height 2 and multiplicity 2 [J. Algebra 316, No. 2, 715--734 (2007; Zbl 1132.13006)]. In contrast to these characterizations, the main result of this paper shows that no such finite characterization is possible under the unmixedness assumption alone. In particular, the authors prove the following: Theorem. For any integers \(h, e \geq 2\) with \((h,e) \not = (2,2)\) and integer \(p \geq 5\), there exists a homogeneous ideal \(I_{h,e,p}\) in a polynomial ring \(R\) over the algebraically closed field \(k\) such that: \(I_{h,e,p}\) has height \(h\); the Hilbert-Samuel multiplicity of \(R/I_{h,e,p}\) is \(e\); the projective dimension of \(R/I_{h,e,p}\) is at least \(p\); \(I_{h,e,p}\) is primary to a linear prime \((x_1, \ldots, x_h)\). After reviewing necessary background, at the beginning of Section 3 the authors provide the following useful summary of their approach to the proof of the above theorem. For the ease of notation, as in the main theorem, denote by \(L_{h,e,p}\) an ideal in a polynomial ring \(R\) over an algebraically closed field \(k\) such that \(L_{h,e,p}\) has height \(h\), \(R/L_{h,e,p}\) has Hilbert-Samuel multiplicity \(e\), and the canonical module for \(R/L_{h,e,p}\) has projective dimension greater than or equal to \(p\). Using this notation, the authors first define four key families of primary ideals with large projective dimension and nice resolutions and canonical modules. The four key families are \(L_{2,5,p}, L_{2,6,p}, L_{2,20,p}\) and \(L_{3,6,p}\). Then by linking via a complete intersection from one of these four key families, the authors present primary and radical linear ideals of height 2 and any multiplicity greater than or equal to 3 with arbitrarily large projective dimension. To obtain examples with arbitrary height, one adds extra linear generators. It is also noted that an additional construction is needed for the case of \(L_{3,2,p}\) case since the case for \(L_{2,2,p}\) is finite. Section 4 of the paper is dedicated to the details of the construction for the four key families. The key ideals were originally found with the use of Macaulay2 and have many generators (see Section 5 for an example demonstrating the minimal number of generators for \(I_{2,4,p}\)). The paper concludes with a discussion of classifying ideals with Serre's (\(S_2\)) property. The authors note that the ideals constructed in the paper satisfy Serre's (\(S_1\)) property but none are (\(S_2\)). The authors close with the following question: Question: Is there a finite classification of homogeneous unmixed or primary ideals of a given height and multiplicity that satisfy Serre's (\(S_2\)) condition? Is there a classification for such ideals that are (\(S_2\)) on the punctured spectrum?