Higher discriminants of binary forms (Q2480108): Difference between revisions

This article addresses an old problem in classical invariant theory, viz. to find algebraic conditions for a binary form to have linear factors with assigned multiplicities. Any binary form \(P_{k}(x,y)=\sum_{i=0}^{k}a_{k-i}x^{k-i}y^{i}\) of degree \(k\) over \(\mathbb C\) splits into linear factors which define \(1\)-dimensional subspaces called the \textit{roots} of \(P_{k}\). For a partition \(k=k_{1}+\cdots +k_{p}\), let \(S^{k}_{k_{1},\dots ,k_{p}}\) denote the set of \(k\)-forms with \(k_{1},\dots ,k_{p}\) coincident roots, where \(k_{i}\) is omitted if it is \(1\), and the \(p\) groups of coincident roots are not assumed to be distinct. (For example, \(k=2+1+\cdots +1\) corresponds to the space \(S^{k}_{2}\) of all \textit{degenerate} forms.) \(\text{SL}(2,\mathbb C)\) acts on the manifold \(S^{k}\) of binary forms of degree \(k\), and the submanifold \(S^{k}_{k_{1},\dots ,k_{p}}\) is invariant under this action. \(S^{k}_{2}\subset S^{k}\) is defined by one equation (the discriminant \(D=0\)), but the \(S^{k}_{k_{1},\dots ,k_{p}}\) with \(k_{1}>2\) are defined by systems of several polynomial equations in the \(a_{i}\), called ''higher discriminants''. The aim is to find these explicitly. \textit{J. V. Chipalkatti} [J. Algebra 267, No. 1, 246--271 (2003; Zbl 1099.13501)] calls \(S^{k}_{k_{1},\dots ,k_{p}}\) a \textit{coincident root locus}, and uses representations of \(\text{SL}(2,\mathbb C)\) to find the algebraic conditions. The article under review proposes a more elementary method. Select the chart \(y\neq 0\) on \(\mathbb C P^{1}\) and put \(z=\frac{x}{y}\), so the roots \(\Lambda_{i}\) of \(P(x,y)\) correspond to roots \(\lambda_{i}\) of \(P(z)\). Then apply formal logic to transform statements about coincidence of roots (which define the \(S^{k}_{k_{1},\dots ,k_{p}}\)) into an equivalent system of equations in the \(\lambda_{i}\). These can be symmetrized, and then the Vieta formulae are used to obtain equations in the coefficients \(a_{i}\). The method does not always yield a system which is \(\text{SL}(2)\)-invariant (e.g. \(S^{3}_{3}\)). In general, the exposition is detailed and clear. The relation between the \(\text{SL}(2)\)-orbits and the polynomials in general remains an open question, as does the generalization of this relation to dimension \(n>2\). We point out that the notion of nondegeneracy used here is usually called nonsingularity in the algebraic theory of higher degree forms (and in algebraic geometry), where nondegeneracy refers to a weaker notion.