Pseudo-Néron model and restriction of sections (Q2167785)

Let \(k\) be an algebraically closed field and let \(u_0: S\to\mathbb{P}^n_k\) be a generically finite and generically unramified morphism. One defines \begin{itemize} \item[(1)] a \textit{\(u_0\)-line} to be a curve in \(S\) of the form \(S\times_{\mathbb{P}^n_k}L\) where \(L\) is a line in \(\mathbb{P}_k^n\), \item[(2)] a \textit{\(u_0\)-conic} to be a curve in \(S\) of the form \(S\times_{\mathbb{P}^n_k}C\) where \(C\) is a plane conic in \(\mathbb{P}^n_k\), \item[(3)] a \textit{\(u_0\)-line pair} to be a curve in \(S\) of the form \(S\times_{\mathbb{P}^n_k}L\) where \(L=L_1\cup L_2\) is a pair of incident lines in \(\mathbb{P}^n_k\), \item[(4)] a \textit{\(u_0\)-smooth curve} to be an irreducible smooth curve in \(S\) of the form \(S\times_{\mathbb{P}^n_k}C_0\) where \(C_0\) is a smooth curve in \(\mathbb{P}^n_k\), \item[(5)] a \textit{\(u_0\)-curve pair} of degree \((d+2)\) to be a connected curve in \(S\) of the form \(S\times_{\mathbb{P}^n_k}C\) where \(C=C_0\cup C_1\) is a pair of curves in \(\mathbb{P}^n_k\) intersecting transversally at a single closed point such that \(C_0\) is a genus zero smooth curve of degree \(d\), and \(C_1\) is a smooth conic, \item[(6)] a \textit{\(u_0\)-planar surface} to be a surface in \(S\) of the form \(S\times_{\mathbb{P}^n_k}\Sigma\) for a \(2\)--plane \(\Sigma\) in \(\mathbb{P}^n_k\). \end{itemize} The article is motivated by the work [\textit{T. Graber} and \textit{J. M. Starr}, Clay Math. Proc. 18, 311--327 (2013; Zbl 1327.14201)] that proves a theorem on restrictions of sections for families of abelian varieties. More precisely, let \(k\) be an uncountable algebraically closed field, and let \(S\) be an integral normal quasi-projective \(k\)-scheme of dimension \(b\geq 2\). Let \(A\) be an abelian scheme over \(S\). Then the theorem of Graber and Starr says that for a very general line pair \(C\) in \(S\), the restriction map of sections Sections\((A/S)\to\mathrm{Sections}(A_C/C)\) is a bijection. Moreover, the same holds when \(C\) is a very general planar surface in \(S\). Further, if char\((k)=0\), then the same holds with \(C\) a very general conic in \(S\). The proof uses Néron models in the case of conics. The first main result of the article shows that there exists a broader class of varieties for which analogues of the above hold. Turning to details, let \(k\) and \(S\) be as above with char\((k)=0\). Let \(X\) be a smooth \(S\)-scheme admitting a finite morphism \(f:X \to A\) to an abelian scheme \(A\) over \(S\). Let \(e\) be the fiber dimension of Iso\((A)\), where Iso(A) is the isotrivial factor of A. Let \(d \geq 0\) be an even integer. Then Theorem 1.3 of this paper asserts that for \(d >2e - 2\), every section of \(X_C\) over a very general genus-\(0\) and degree-\((d +2)\) curve pair or a very general genus-\(0\) degree-\((d +2)\) smooth curve \(C\) is the restriction of a unique global section of \(X\) over \(S\). Importantly, instead of working with the Néron model, this article introduces pseudo-Néron models which is a weaker version of Néron models. It is shown that if a smooth variety \(X_K\) admits a finite morphism to an abelian variety \(A_K\), then \(X_K\) admits a pseudo-Néron model. This naturally leads to the question regarding the existence of other classes of varieties admitting pseudo-Néron models. The second main result of this paper answers this question in the affirmative. As before, let \(k\) be an uncountable algebraically closed field. Let \(S\) be a Dedekind scheme of finite type over \(k\), i.e., an irreducible, Noetherian, and normal scheme of dimension one, with function field \(K\). Let \(d\) be an integer prime to char\((k)\). Let \(H \subset \mathbb{P}^n_k\) be a very general smooth hypersurface of degree \(d \geq 2n - 1\), and let \(X_K\) be a smooth \(K\)-variety admitting a finite morphism to \(H \times_{\mathrm{Spec} k} \mathrm{Spec} K \subset \mathbb{P}^n_K\). Then Theorem 1.11 of this paper asserts that \(X_K\) has a pseudo-Néron model over \(S\). For more details, we refer to the introduction of the paper, which also contains a review of the theorems on sections, as well as the idea of the proof. The appendix written by Jason Michael Starr contains several important lemmas including the bigon lemma that have been used in the article.