A tight bound on the projective dimension of four quadrics (Q1748109)

Let \(S = k[x_1,\dots,x_n]\) be a polynomial ring over a field \(k\). Let \(I = (f_1,\dots,f_N)\) be a homogeneous ideal in \(S\), and set \(d_i = \deg (f_i)\) for all \(1 \leq i \leq N\). An important question due to M. Stillman is whether there is a bound on the projective dimension \(\text{pd}(S/I)\) that depends only on \(d_1,\dots,d_N\) and \(N\). This was settled by \textit{T. Ananyan} and \textit{M. Hochster} [Math. Res. Lett. 19, No. 1, 233--244 (2012; Zbl 1276.13014)], leaving open the question of how optimal the bounds can be shown to be. Different authors, including Ananyan and Hochster and also including the present authors, have previously considered the case of \(N\) quadrics (i.e. homogenous polynomials of degree 2) and the case of a small number of cubics, obtaining sharper bounds, but these are still far from optimal. In the present paper, the authors show that if \(I\) is generated by four quadrics then the projective dimension of \(S/I\) is at most 6. Combined with examples of \textit{J. McCullough} [Proc. Am. Math. Soc. 139, No. 6, 2017--2023 (2011; Zbl 1215.13007), this bound is optimal. To achieve this result, the authors follow an approach pioneered by \textit{B. Engheta} [J. Algebra 316, No. 2, 715--734 (2007; Zbl 1132.13006); J. Symb. Comput. 45, No. 1, 60--73 (2010; Zbl 1179.13009)], but they need to introduce an array of new ideas.