On the enumerative geometry of the resultant variety (Q2655185)

Let \(k\) be a normal ring, \(A=(a_1,\ldots, a_n)\) be a sequence of positive integers and \(X_1,\ldots,X_n\) be variables of weights \(a_i\). Let \(B=(b_1,\ldots,b_n)\) be another sequence of positive integers and \(F_1,\ldots,F_n\) be homogeneous polynomials of respective degree \(b_j\), \(j=1,\ldots,n\), of the form \(F_j=\sum U_{j\nu}X^\nu\), where the coefficients are generic and \(\nu\) varies in a subset \(N_j\) of \(N_{b_j}(A)=\{ \nu\in\mathbb N^n\cup\{0\}\mid\sum\nu_ia_i=b_j\}\). The ideal \(\alpha\) generated by \(F_1,\ldots,F_n\) in \(Q[X_1,\ldots,X_n]\), where \(Q=\otimes k[U_{j\nu},\nu\in N_j]\), defines an algebraic set \(X\), whose natural projection in \(k^n\) is defined by a single polynomial \(\mathrm{Res}_B^A\), called the anisotropic resultant of \(F_1,\ldots,F_n\). In the paper under review a characterization is given of the linear subvarieties of the resultant variety \(V(Res_B^A)\). This relies on the admissibility conditions for the pair \((B,A)\) found by Friedlander--Halperin, and Scheja--Storch [see \textit{J. B. Friedlander, S. Halperin}, Invent. Math. 53, 117--133 (1979; Zbl 0396.55010); \textit{G. Scheja, U. Storch}, Regular sequences and resultants. Research Notes in Mathematics. 8. Natick, MA: A K Peters. (2001; Zbl 1012.13004)]. In the isotropic case, i.e. when \(A=(1,\ldots,1)\), the main result is that for all \(s\), \(1\leq s\leq n-1\), and for all \(I,J\) subsets of \(\{1,\ldots,n\}\) of cardinality resp. \(s\) and \(n-s+1\), there exists a linear subspace of codimension \(c(I,J)\) contained in \(V(Res_B^A)\) and maximal among its linear subspaces. Moreover for every maximal linear subspace an explicit description is given. Here \(c(I,J)\) denotes \(\sum_{j\in J}\pi(b_j)\), where \(\pi(b_j)\) is the partition number of \(b_j\). In the general case, an analogous description is given.