Constructing curves with high rank via symmetry (Q4397440)

Let \(C\) be an algebraic curve of genus \(g\) defined over a function field \(K:=\mathbb{Q}(t_1,\dots,t_N)\). By definition the rank of \(C\) is the Mordell-Weil rank of its Jacobian variety \(A\). Note that \(A(K)\) is not necessarily finitely generated, so the definition of the Mordell-Weil rank of \(A\) is more complicated than in the classical case: It is the rank of the finitely generated group \(A(K)/ \tau B(k)\), where \((\tau,B)\) is the \(K/k\)-trace of \(A/K\) [see \textit{S. Lang}, ``Fundamentals of diophantine geometry'' (1983; Zbl 0528.14013)]. It is an important question to produce curves over \(K\) of large rank because (by a spezialization argument) there are infinite families of curves of high rank over \(\mathbb{Q}\). In the reviewed paper the author proves that for any \(g>1\), there exists a curve of genus \(g\) over \(\mathbb{Q}(t_1, \dots, t_N)\) (with \(N=2g+4)\) with rank at least \(4g+7\). In particular this result improves the bound \(3g+7\) given by \textit{A. Néron} [Proc. int. Congr. Math., Amsterdam 1954, Vol. 3, 481-488 (1956; Zbl 0074.15901)]. It follows from the specialization theorem of Néron-Silverman-Tate that there exists an infinite family of (non constant) curves of genus \(g\) over \(\mathbb{Q}\) with rank at least \(4g+7\). Here are some details about the construction. Set \(u_i=t^2_i\in K\) \((1\leq i\leq N)\). Consider the polynomial \(f(x)\) of degree at most \(n-1\) defined by the identity: \(\prod^N_{i=1} (x-u_i)= h(x)^2-f(x)\) where \(h(x)\) is a monic polynomial of degree \(n\). Then the \(K\)-curve \(\widetilde\Gamma: Y^2=f(X^2)\) is of genus \(g\) and has rank at least \(2N-1=4g+7\). -- The use of such curves as \(\widetilde\Gamma\) is not completely new (they were used by Mestre and by the author to construct curves of small genus with high rank over \(\mathbb{Q}(t))\), but the paper under review contains a new tool to check the independence of some rational points over \(K=\mathbb{Q}(t_1, \dots, t_N)\): It consists of the use of the canonical height of Néron-Tate for abelian varieties defined over \(K\). By contrast with the case \(N=1\), there is no explicit formula for the height pairing. Therefore one has to use the Galois action of a large symmetry group (e.g. \({\mathcal S}_N)\) to show the independence of \(2N-1\) rational points on \(\widetilde \Gamma\).