Harbourne, Schenck and Seceleanu's conjecture (Q2629885)

Let \(R = k[x_1,\dots,x_r]\), where \(k\) is an infinite field. Let \(A = R/I\) be a graded artinian quotient of \(R\). We say that \(A\) has the \textit{weak Lefschetz property (WLP)} if multiplication by a general linear form induces a homomorphism of maximal rank from any degree component of \(A\) to the next. We consider ideals of the form \(I = (L_1^{a_1}, \dots, L_n^{a_n})\), where \(L_1,\dots,L_n\) are general linear forms. In 2011, Harbourne, Schenck and Seceleanu [\textit{B. Harbourne} et al., J. Lond. Math. Soc., II. Ser. 84, No. 3, 712--730 (2011; Zbl 1242.13016)] conjectured that if \(n = r+1\) and \(a_1 = \dots = a_{r+1} = 2\) (i.e. \(I\) is an almost complete intersection ideal generated by squares of general linear forms) then \(A\) fails the WLP for \(r=6\) and all \(r \geq 8\). Part of this conjecture, namely the case when \(r\) is even, was proved already by the author, \textit{U. Nagel}, and the reviewer in [Algebra Number Theory 6, No. 3, 487--526 (2012; Zbl 1257.13003)]. In the current paper, the author proves the case when \(r\) is odd. The key point is to determine the degree in which WLP fails. She also proves that in the remaining cases, the WLP does hold. This completely classifies those ideals of the stated form that have the WLP.