Some special cases of the Eisenbud-Green-Harris conjecture (Q2441056)

Let \(S = k[x_1, x_2, \ldots, x_n]\) be the polynomial ring in \(n\) variables over the field \(k\) with the ordering \(x_1 > x_2 > \cdots > x_n\). This paper studies the Eisenbud--Green--Harris (EGH) Conjecture which suggests a characterization of the Hilbert functions of homogeneous ideals \(I\) in \(S\) which contain a regular sequence. More precisely, the EGH Conjecture states: EGH Conjecture [\textit{D. Eisenbud} et al., in: Days on algebraic geometry, Orsay, France, July 20--26, 1992. Journées de géométrie algébrique d'Orsay, France, juillet 20--26, 1992. Paris: Société Mathématique de France, Astérisque. 218, 187--202 (1993; Zbl 0819.14001)]: If \(I \subset S\) is a homogeneous ideal containing a regular sequence of forms \(f_1, \ldots, f_r\) of degrees \(a_1, \ldots, a_r\), where \(2 \leq a_1 \leq a_2 \leq \cdots \leq a_r\) and \(1 \leq r \leq n\), then there exists a homogeneous ideal in \(S\) containing \(x_1^{a_1}, \ldots, x_r^{a_r}\) with the same Hilbert function. \noindent This generalizes the well-known Macaulay's Theorem which says that given any homogeneous ideal \(I \subset S\), there exists a lex ideal with the same Hilbert function (from \textit{F. S. Macaulay} [Proc. Lond. Math. Soc. (2) 26, 531--555 (1927; JFM 53.0104.01)]. Much work has been done in trying to prove the EGH Conjecture. For example, \textit{G. F. Clements} and \textit{B. Lindström} [J. Comb. Theory 7, 230--238 (1969; Zbl 0186.01704)] and \textit{S. M. Cooper} and \textit{L. G. Roberts} [J. Commut. Algebra 1, No. 3, 361--380 (2009; Zbl 1209.13018)] showed that if \(I\) contains the regular sequence \(x_1^{a_1}, \ldots, x_r^{a_r}\), then there is a lex ideal \(L\) such that \(L + (x_1^{a_1}, \ldots, x_r^{a_r})\) (called a lex-plus-powers ideal) has the same Hilbert function as \(I\). \textit{S. M. Cooper} also proved the conjecture in special cases when \(n=3\) [J. Pure Appl. Algebra 212, No. 1, 122--131 (2008; Zbl 1130.13009)] and in a more geometric setting [in: Progress in commutative algebra 1. Combinatorics and homology. De Gruyter Proceedings in Mathematics, 167--198 (2012; Zbl 1250.13021)]. \textit{G. Caviglia} and \textit{D. Maclagan} [Math. Res. Lett. 15, No. 2--3, 427--433 (2008; Zbl 1154.13001)] proved that the EGH Conjecture holds when \(a_i > \sum_{j=1}^{i-1}(a_j-1)\) for fixed integers \(2 \leq a_1 \leq a_2 \leq \cdots \leq a_n\). This paper includes a list of references for additional proved cases of the EGH Conjecture. The original EGH Conjecture was stated in the case of \(a_1 = \cdots = a_n = 2\). In this special case, Richert verified (in unpublished work) the conjecture for \(n \leq 5\), Cooper proved the case in a geometric setting [Zbl 1250.13021], and \textit{J. Herzog} and \textit{D. Popescu} [Compos. Math. 113, No. 1, 1--22 (1998; Zbl 0908.13008)] proved the conjecture when \(k\) is of characteristic zero and \(I\) is minimally generated by generic quadratic forms. The first main result of this paper is to prove the EGH Conjecture in the special case that \(a_1 = \cdots = a_n = 2\) and \(2 \leq n \leq 4\). The proof uses the fact that the EGH Conjecture holds true if and only if \(\dim I_k(I_{d+1}) \geq \dim_k(S_1J_d + (x_1^{a_1}, \ldots, x_n^{a_n})_{d+1})\) where \(J\) is a lex-plus-powers ideal such that \(\dim_k(I_d) = \dim_k(J_d)\). The author also shows the conjecture is true for regular sequences of 2-forms of very special type (see Proposition 2.3). The remainder of the paper concentrates on almost complete intersections. \textit{C. A. Francisco} [J. Algebra 276, No. 2, 737--760 (2004; Zbl 1102.13009)] proved the following for almost complete intersections: Fix integers \(2 \leq a_1 \leq \cdots \leq a_n\) and let \(d\) be an integer such that \(d \geq a_1\). Let \(I \subset S\) be a homogeneous ideal minimally generated by forms \(f_1, \ldots, f_n, g\), where \(f_1, \ldots, f_n\) is a regular sequence, \(\deg(f_i) = a_i\) and \(\deg(g) = d\). Let \(J = (x_1^{a_1}, \ldots, x_n^{a_n}, m)\), where \(m\) is the greatest monomial in lex order in degree \(d\) not in \((x_1^{a_1}, \ldots, x_n^{a_n})\). Then \(\dim_k(I_{d+1}) \geq \dim_k(J_{d+1})\). The author presents two different proofs of this result in the special case for \(a_1 = \cdots = a_n = 2\). The first proof studies \(S/((f_1, \ldots, f_n):I)\) and the second proof uses the minimal free resolution of \(S/(x_1, x_2, \ldots, x_i)\).