Deformation of integral sections in the Mordell-Weil group (Q2694003)

In the paper under review, the authors study the infinitesimal structure of the moduli space of integral sections in the Mordell-Weil group. Let \(\mathbb{K}\) be an algebraically closed filed of characteristic zero and \(f : S \rightarrow \mathbb{P}^{1}_{\mathbb{K}}\) be a relatively minimal elliptic surface fibration with a fixed identity section \(O\). Let \(K = \mathbb{K}(t)\) be the function field of \(\mathbb{P}^{1}_{\mathbb{K}}\) and \(E:=S \times_{\mathbb{P}^{1}_{\mathbb{K}}}K\) be the generic fibre of \(f\). Sections of \(f\) form a group \(\mathrm{MW}(f)\) known as the Mordell-Weil group of \(f\). In \(\mathrm{MW}(f)\), those sections of \(f\) that are disjoint from \(O\) are called integral sections and we denote by \(\mathcal{P} \subset\mathrm{MW}(f)\) the subset of integral sections. Take \(W\) to be the minimal Weierstrass model of \(S\), which is defined by a certain Weierstrass equation: \[ y^{2}=x^{3}+a(t)x+b(t), \quad a(t),b(t) \in \mathbb{K}[t]\tag{\(\star\)} \] over the standard affine subset \(\mathbb{A}^{1}_{\mathbb{K}}=\mathrm{Spec}(\mathbb{K}[t])\). \textit{T. Shioda} [Proc. Japan Acad., Ser. A 86, No. 2, 21--26 (2010; Zbl 1183.14050)] introduced the closed subscheme \(Z\subset \mathbb{A}^{5\chi +2}_{\mathbb{K}} = \mathrm{Spec}(\mathbb{K}[u_{0},u_{1}, \dots, u_{2\chi},v_{0},v_{1}, \dots, v_{3\chi}])\), where \(\chi = \chi(S)\), so that the polynomials \(x=u_{2\chi}t^{2\chi} + \dots + u_{0}\), \(y = v_{3\chi}t^{3\chi} + \dots + v_{0}\) determine a solution of \((\star)\) if and only if \((u_{0}, \dots u_{2\chi}, v_{0}, \dots, v_{3\chi}) \in Z\). It turns out that there is a canonical one to one correspondence between \(\mathcal{P}\) and \(Z(\mathbb{K})\) and hence \(Z\) can be considered as the moduli space of integral section. We assume additionally that \(f\) is non-trivial. Under this assumption \(\chi = \chi(S)\) is positive. This implies \(\# \mathcal{P} < + \infty\) and \(Z\) is finite over \(\mathbb{K}\). Shioda asked how the infinitesimal neighborhood of \([P] \in Z(\mathbb{K})\) looks like. Denote by \([P]\) the associated \(\mathbb{K}\)-point in \(Z\) for \(P \in \mathcal{P}\). We have the following two questions: Question A. Assume that an integral section \(P\) has height \(\langle P, P \rangle =2\chi\). Do we have that \(\dim_{\mathbb{K}}\, \mathcal{O}_{Z, [P]}=1\)? Question B. Assume that there are no reducible fibers for \(f\). Then is it true that \(\dim_{\mathbb{K}} \, \mathcal{O}_{Z, [P]}=1\) for every \(P\)? In order to formulate the main result, we need to recall the following definition. Let \(\pi : S \rightarrow W\) be the contraction map and \(P'=\pi(P)\), then \[ T_{Z,[P]} = H^{0}(P', \mathcal{N}_{P'/W}), \] where \(\mathcal{N}_{P'/W}\) is the normal sheaf of \(P'\) in \(W\). The main result of the paper under review tells us the following. Main Theorem. If an integral section \(P\) has \(\langle P,P \rangle = 2\chi\), then \[ \dim_{\mathbb{K}}T_{Z,[P]}=0. \] In particular, this implies that Questions A and B have affirmative answers.