Some remarks on equations defining coincident root loci (Q435939): Difference between revisions

Coincident root loci (CRL) for binary forms are defined in the following way; let \(V = k[x,y]_1\), where \(k\) is algebrically closed and char\(k =0\), and consider \(F\in k[x,y]_d\); if \(\lambda = (\lambda_1 ,\dots ,\lambda_e)\) is a partition of \(d\), we set \[ X_{\lambda} = \{ [F]\in {\mathbb P}(\mathrm{Sym}^d V) | \exists L_1,\dots,L_e \in V: [F] = \prod _{i=1}^e [L_i]^{\lambda _i} \} . \] Better, \(X_\lambda\) can be defined by writing \(\lambda =(1^{e_1},2^{e_2},\dots,d^{e_d})\), where \(e_r := \) ``number of \(\lambda_i = r\)'', and so \(Y_\lambda = \prod _{i=1}^d{\mathbb P}(\mathrm{Sym}^{e_i}(V))\); then a morphism of schemes can be defined: \(f_\lambda : Y_\lambda \longrightarrow {\mathbb P}(\mathrm{Sym}^d(V))\), where \(f_\lambda (G_1,\dots,G_d) = \prod _{r=1}^d G_r^r\), and the ``coincident root locus (with multiplicity \(\lambda\))'' is \(X_\lambda =\) im\(f_\lambda\). In this way \(Y_\lambda\) is the normalization of \(X_\lambda\). \(X_\lambda\) is also related to \(\Gamma _\lambda := X_\lambda \times Y_\lambda \subseteq {\mathbb P}(\mathrm{Sym}^d(V))\times Y_\lambda\), and a resolution for \(\Gamma_\lambda\) is given by an Eagon-Northcott complex in [\textit{J. Y. Chipalkatti}, J. Algebra 267, No. 1, 246--271 (2003; Zbl 1099.13501)]. The proof of such result seems to present a gap: In this paper that gap is filled by showing that \(\Gamma_\lambda\) equals a certain subscheme defined via Fitting ideals (as schemes, not only set-theoretically). Moreover, a nice description of the singular locus of \(X_\lambda\) is given (and a criterion to decide whether \(X_\lambda\) is smooth) and also local equations for \(\Gamma_\lambda\).