Combinatorial bounds on Hilbert functions of fat points in projective space (Q538065): Difference between revisions

Given a finite subset \(\{p_1,\dots,p_r \}\subset {\mathbb P}^N\) , how to determine the Hilbert function of schemes of fat points \(A = a_1p_1+\dots+a_rp_r\), \(a_i\in \mathbb N\)? Here \(A\) is defined by the ideal \(I_A = I_{p_1}^{a_1}\cap \dots\cap I_{p_r}^{a_r}\). This is a problem which has received much attention in the last 30 years, since it has appeared in connection to many other problems in Math (someone even talked about ``ubiquity of fat points''). In this paper algorithms are given to get upper and lower bounds for the Hilbert function of schemes \(A\) in the plane, depending only on the \(a_i\)'s and on which subsets of the points are linearly dependent. The idea is to repeatedly use exact sequences of type: \[ 0 \rightarrow {\mathcal I}_{A:L} (t-1) \rightarrow {\mathcal I}_{A} (t) \rightarrow {\mathcal I}_{A\cap L,L }(t) \rightarrow 0 \] where \(L\) is a hyperplane defined by \(F=0\) an the residual scheme \(A:L\) is defined by the ideal \(I_A:(F)\). Consider hyperplanes \(L_1,\dots,L_n\) such that, by iterating this procedure with \( A_1 = A:L_1, A_2=A_1:L_2,\dots,A_n=A_{n-1}:L_n\), we get \(A_n=\emptyset\). Then one gets the required bounds for the Hilbert function of \(A\) from the cohomology sequences associated to the exact sequences of ideal sheaves. In the second part of the paper a condition is given in order to get that the lower and upper bounds obtained for the Hilbert function coincide (it is a numerical condition on the degrees which are ``cut off'' from \(A\) at each step, by using the \(L_i\)'s). Conditions are given to get that the bounds on the Hilbert function also give the graded Betti numbers of a minimal resolution of \(I_A\), and many examples of families of fat points are studied.