Invariant theory for the elliptic normal quintic. I: Twists of \(X(5)\) (Q2377011)

Let \(V\) and \(W\) be 5-dimensional vector spaces over a perfect field \(K\) (of characteristic 0 or \(\geqslant 7\)) with bases \(\{v_0,\dots,v_4\}\) and \(\{w_0,\dots,w_4\}\), respectively. A \textit{genus one model} of degree 5 is a \(5\times 5\) alternating matrix \(\phi\) of linear forms on \(\mathbb{P}^4\) and has an associated subvariety \(C_\phi\subset \mathbb{P}^4\) defined via the \(4\times 4\) Pfaffians of \(\phi\). There is an identification between genus one models and \(\wedge^2 V\otimes W\) given by \[ (\phi_{i,j}) \leftrightarrow \sum_{i< j}(v_i\wedge v_j)\otimes \phi_{i,j}(w_0,\dots,w_4)\;. \] Computing the invariants for the action of \(\mathrm{SL}(V)\times \mathrm{SL}(W)\) on the polynomial ring \(K[\wedge^2 V\otimes W]\), the author (in [Proc. Lond. Math. Soc. (3) 97, No. 3, 753--782 (2008; Zbl 1221.11135)]) proved that they are generated by polynomials \(c_4\) and \(c_6\) of degree 20 and 30 respectively and that \(C_\phi\) has Jacobian \(y^2=x^3-27c_4(\phi)-54c_6(\phi)\) (if \(C_\phi\) is smooth and of genus 1, i.e., \(\phi\) is \textit{non-singular}). The study of invariants was carried out using covariants like the \textit{Hessian} (see the author [Proc. Lond. Math. Soc. (3) 104, No. 3, 613--648 (2012; Zbl 1286.11088)]). A \textit{covariant} for the rational representation \(\rho_Y\,:\,\mathrm{GL}(V)\times \mathrm{GL}(W) \rightarrow Y\) is a polynomial map \(F: \wedge^2 V\otimes W \rightarrow Y\) such that \[ F\circ g = \rho_Y(g)F\qquad \forall\,g\in \mathrm{SL}(V)\times \mathrm{SL}(W) \;. \] The paper under review deals with covariants \(F\) by studying their restrictions to the Hesse family \[ u : \mathbb{A}^2 \longrightarrow \wedge^2 V\otimes W \] \[ (a,b) \longrightarrow a\sum (v_1\wedge v_4)w_0 + b\sum (v_2\wedge v_3)w_0 \;, \] where the sums are taken over all cyclic permutations of the subscripts modulo 5. Such family represents all non-singular genus one models (modulo the action of \(\mathrm{GL}(V)\times \mathrm{GL}(W)\,\)) and provides a universal family of elliptic curves over \(X(5)\). It turns out that \(F\) is uniquely determined by the \textit{discrete covariant} \(F\circ u\,:\,\mathbb{A}^2 \rightarrow Y^{H_5}\) (where \(H_5\) is the Heisenberg group) and the author also provides a description of the module of discrete covariants for representations \(\pi:\mathrm{SL}_2(\mathbb{Z}/5) \rightarrow \mathrm{GL}(K)\) (to account for the action of the extended Heisenberg group \(H_5^+:=H_5 \ltimes \mathrm{SL}_2(\mathbb{Z}/5)\,\)). \noindent The explicit description of several representations and of their invariants (evaluated at the Hesse family) is applied to the study of 5-congruent elliptic curves (i.e., curves with 5-torsion groups isomorphic as Galois modules) and provides equations for the family of curves 5-congruent to a given \(E\), depending on the effect of the isomorphism on the Weil pairing \(e_5\) (i.e., letting \(\psi:E\rightarrow E'\) be the isomorphism, the equations depend on the \(1\leqslant r\leqslant 4\) such that \(e_5(\psi(S),\psi(T))=e_5(S,T)^r\)). \noindent Another interesting and very nice application is given in Section 6, where the author uses the result on 5-congruent curves and a particular covariant of degree 49 to provide a formula for doubling elements in 5-Selmer groups and presents (as an example) the detailed computations for doubling an element in the Tate-Shafarevic group of \(y^2+xy+y=x^3+x^2-3146x+39049\).