A new approach to Hilbert's theorem on ternary quartics (Q1763504)

From MaRDI portal
scientific article
Language Label Description Also known as
English
A new approach to Hilbert's theorem on ternary quartics
scientific article

    Statements

    A new approach to Hilbert's theorem on ternary quartics (English)
    0 references
    0 references
    0 references
    0 references
    0 references
    22 February 2005
    0 references
    This is a very interesting paper which gives new insight in Hilbert's theorem. The authors prove: Theorem 1.1: Suppose \(f(x,y,z)\) is a nonnegative real quartic form which defines a smooth plane curve \(Q= \{(x: y: z)\in\mathbb{P}^2(\mathbb{C}): f(x,y,z)= 0\}\). Then the inequivalent representations of \(f\) as a sum of three squares (of real quadratic forms) -- modulo the real orthogonal group \(O(\mathbb{R}^3)\) -- are in one-to-one correspondence with the eight 2-torsion points in the non-identity component of \(J(\mathbb{R})\), where \(J\) is the Jacobian of \(Q\). The assumptions imply in particular: \(f\) is irreducible, \(Q(\mathbb{R})= \emptyset\), \(Q\) has genus 3, \(J\) has 63 non-zero complex 2-torsion points. Hilbert's original theorem (i.e. any nonnegative quartic form \(f\) is a sum of three squares in at least one way) follows from the above theorem by continuity arguments. The proof of Theorem 1.1 proceeds in two steps: (1) The non-trivial 2-torsion points of \(J(\mathbb{C})\) are in one-to-one correspondence with the equivalence classes -- modulo \(O(\mathbb{C}^3)\) -- of representations of \(f\) as a sum of three squares of complex quadratic forms. A condensed proof using Weil divisors, the Picard group \(\text{Pic}(Q)\) and the Riemann-Roch Theorem is given in the paper. The result itself goes back to A. B. Coble (1929; JFM 55.0808.02), it was rediscovered by C. T. C. Wall (1991; Zbl 0741.14014). (2) Show that under (1) the non-trivial 2-torsion points of \(J(\mathbb{R})\) correspond to ``signed quadratic representations'' \(f=\pm q^2_1\pm q^2_2\pm q^2_3\) with \(q_i\in\mathbb{R}[x,y,z]\), and the 2-torsion points in the non-identity component of \(J(\mathbb{R})\) correspond to the representations with \(+\) signs. This is the essential new result of the paper. For the proof one needs the following exact sequences \[ \begin{gathered} 0\to \text{Pic}(Q_r)\to \text{Pic}(Q)@>\partial>> \text{Br}(R)\to \text{Br}(Q_r),\\ 0\geq J(\mathbb{R})^0\to J(\mathbb{R})@>\partial>> \text{Br}(\mathbb{R})\to 0,\end{gathered} \] where \(Q_r\) is the curve \(Q\) as a curve over \(\mathbb{R}\) such that \(Q= Q_r\otimes\mathbb{C}\). The first exact sequence follows from Hochschild-Serre spectral sequence for étale cohomology, the second sequence follows from a theorem of G. Weichold (1883; JFM 15.0431.01) reproved by Geyer (1964).
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references