Free resolutions of fat point ideals on \(\mathbb{P}^2\) (Q1380053): Difference between revisions

Given distinct points \(p_1,\dots, p_r\) of the projective plane \(\mathbb{P}^2\) over an algebraically closed field, a fat point subscheme \(Z= m_1p_1 +\cdots+ m_rp_r\) of \(\mathbb{P}^2\) is the subscheme defined by an ideal of the form \(I_Z= {\mathfrak p}_1^{m_1} \cap\cdots\cap {\mathfrak p}_r^{m_r}\), where \({\mathfrak p}_i\) is the prime ideal corresponding to \(p_i\). By allowing infinitely near points, the author extends this notion, and then tries to compute the graded free resolution of \(I_Z\) in case \(p_1,\dots, p_r\) are contained in a plane curve of degree \(\leq 3\). Using the blowing up \(X\) of \(\mathbb{P}^2\) at \(p_1,\dots, p_r\) and the sheaf \({\mathcal F}_d\) corresponding to the divisor \(de_0- m_1e_1-\cdots- m_re_r\) for \(d\in\mathbb{Z}\), the problem amounts to computing \(h^0(X,{\mathcal F}_d)\) and the dimension \(s({\mathcal F}_d, e_0)\) of the cokernel of \(H^0(X, {\mathcal F}_d)\otimes H^0(X, e_0)\to H^0(X,{\mathcal F}_{d+1})\). Thus the problem splits into finding the monoid of effective divisor classes, their Zariski decompositions, and \(h^0(X,{\mathcal F})\) as well as \(s({\mathcal F}, e_0)\) for all \({\mathcal F}\) in the cone of numerically effective classes. In the case of points on a line or conic, the author shows \(s({\mathcal F}, e_0)= 0\) for all numerically effective classes, whereas in the case of points on a cubic his results are less comprehensive and rely on certain restrictions on the classes \({\mathcal F}\) or the surfaces \(X\) under consideration.