Geometric Bogomolov conjecture for abelian varieties and some results for those with some degeneration (with an appendix by Walter Gubler: the minimal dimension of a canonical measure) (Q382237)

Let \(K\) be a number field and let \(\overline{K}\) be a fixed algebraic closure of \(K\). Suppose that \(A\) is an abelian variety over \(\overline{K}\) and \(L\) is a symmetric ample line bundle on \(A\). The Néron-Tate height \(\widehat{h}_L\) defines a semi-positive definite quadratic form on \(A(\overline{K})\). Let \(X\) be a closed subvariety of \(A\). Set \(X(\varepsilon;L):=\{x\in X(\overline{K}) \mid \widehat{h}_L(x)\leq\varepsilon\}\) for any positive real number \(\varepsilon>0\), the Bogomolov conjecture for abelian varieties proved by Shou-wu Zhang states that there exists an \(\varepsilon>0\) such that \(X(\varepsilon;L)\) is not Zariski dense in \(X\), unless \(X\) is a torsion subvariety. The paper under review concerns the similar conjecture for abelian varieties over function field \(K\) over an algebraically closed field \(k\) and the height is the classical geometric height. Precisely, the author gave a precise formulation of the geometric version of the Bogomolov conjecture for arbitrary abelian varieties and gave partial answer to it under a certain condition on degeneration. In this geometric setting, let \(X\) be an irreducible closed subvariety of \(A\). Let \((A^{\overline{K}/k},\text{Tr}_A^{\overline{K}/k})\) denote the \(\overline{K}/k\)-trace of \(A\). Then \(X\) is said to be special if there are a torsion point \(\tau\in A(\overline{K})\) and a closed subvariety \(X'\) of \(A^{\overline{K}/k}\) such that \[ X=G_X+\text{Tr}_A^{\overline{K}/k}(X_{\overline{K}}')+\tau \] where \(G_X\) is the stabilizer of \(X\). Then the geometric Bogomolov conjecture states that there exists an \(\varepsilon>0\) such that \(X(\varepsilon;L)\) is not Zariski dense in \(X\) unless \(X\) is a special subvariety. For an irreducible closed subvariety \(X\subset A\) and a place \(v\) of \(\overline{K}\), the author defined an integer \(b(X_v)\), called the dimension of abelian part of \(X_v\) and he showed that if there is a place \(v\) with \(\text{dim}(X/{G_X})>b((X/{G_X})_v)\), then \(X\) is not a special subvariety. The main theorem of the paper under review is the following: Theorem. Assume that there exists a place \(v\) such that \(\text{dim}(X/{G_X})>b((X/{G_X})_v)\). Then \(X(\varepsilon;L)\) is not Zariski dense for some \(\varepsilon>0\). This theorem can obviously be seen as an evidence of the geometric Bogomolov conjecture.