Generic bound for the Mordell-Lang problem (Q2571057)

Let \(A\) be a semi-abelian variety defined over an algebraically closed field \(K\) of characteristic \(0\), that is, there is an exact sequence of commutative algebraic groups \[ 0\to \mathbb G_{{\mathbf m},K}^t\to A\to A_0, \] where \(A_0\) is an abelian variety defined over \(K\) and \(\mathbb G_{{\mathbf m},K}^t\) the \(t\)-dimensional linear torus. Thus, the group of \(K\)-rational points of \(\mathbb G_{{\mathbf m},K}^t\) is the \(t\)-fold direct product \((K^*)^t\) of the multiplicative group of \(K\). The group operation of \(A\) is denoted additively. The zero element of \(A\) is denoted by \(0_A\). Suppose we are given an immersion \(\iota : A\hookrightarrow \mathbb P_K^N\). For a closed subscheme \(X\) of \(A\), we define the degree \(\text{ deg}_{\iota}(X)\) of \(X\) to be the sum of the degrees of the irreducible components in \(\mathbb P_K^N\) of the Zariski closure of \(\iota (X)\). Further, we let \(Z_X\) denote the union of all translates \(x+B\subseteq X\), where \(x\in X\), and \(B\) runs through all positive dimensional sub-abelian varieties of \(A\). Let \({\mathcal V}(A,\iota , n,D)\) be the collection of all integral closed subschemes \(X\) of \(A\) defined over \(K\) such that \(\dim X =n\), \(\text{ deg}_{\iota}(X)\leq D\) and \(0_A\in X\backslash Z_X\). The author's main result is as follows. In the above exact sequence, let \(g=\dim A_0\). Let \(p\) be a prime number with \(p>D^{n+1}\). For every subgroup \(\Gamma\) of \(A(K)\) of finite rank and every \(D\geq 2\) there is a finite subcollection \({\mathcal E}(\Gamma ,A,\iota , n,D )\) of \({\mathcal V}(A,\iota ,n,D)\) with the following property. For every subvariety \(X\in {\mathcal V}(A,\iota ,n,D) \backslash {\mathcal E}(\Gamma ,A,\iota ,n,D )\) the set \((X\backslash Z_X)(K)\cap\Gamma\) is contained in the union of at most \(p^{2g+t}\) proper closed subschemes of \(A\), each of degree at most \(D^2\). This theorem is deduced from a result by \textit{E. Hrushovski} (the rank \(0\)-case, [Ann. Pure Appl. Logic 112, 43--115 (2001; Zbl 0987.03036)]) and \textit{T. Scanlon} (general case, [Int. Math. Res. Not. 2004, No. 62, 3317--3326 (2004; Zbl 1069.03027)]), proved again in a different way by the author in the paper under review, that \(X(K)\cap \Gamma\) is contained in the union of finitely many translates of sub-semi-abelian varieties contained in \(X\), and that the number of these translates can be bounded above by a finite number depending only on \(A\), \(\Gamma\) and \(\text{ deg}_{\iota}(X)\). We discuss two special cases. In case that \(A\) is an abelian variety over \(K\) and \(n=1\), the author's result implies that for all but finitely many curves \(C\subset A\) over \(K\) of degree \(\leq D\), the number of \(K\)-rational points of \(C\) is at most \(D^{7\dim A}\). The second special case deals with the situation that \(A= \mathbb G_{{\mathbf m},K}^t\), \(D=1\), \(n=t-1\). Then \({\mathcal V}(A,\iota ,n,D)\) consists of all varieties \(X\) given by an equation \(a_1x_1+\cdots +a_tx_t=1\), where \((a_1,\dots ,a_t)\) is a normalized tuple in \(K^t\), i.e., with \(\sum_{i\in I} a_i\not= 0\) for every non-empty subset \(I\) of \(\{ 1,\dots ,t\}\). Further, for each such \(X\), \(Z_X\) consists of all degenerate points, i.e., such that \(\sum_{i\in I} a_ix_i=0\) for some non-empty subset \(I\) of \(\{ 1,\dots ,t\}\). So Rémond's general result gives in this special case, that if \(\Gamma\) is a subgroup of \((K^*)^t\) of finite rank, then for all but finitely many normalized tuples \((a_1,\dots , a_t)\in K^t\), the set of non-degenerate solutions \((x_1,\dots ,x_t)\in\Gamma\) of \(a_1x_1+\cdots +a_tx_t=1\) is contained in the union of at most \(2^t\) proper linear subspaces of \(K^t\). This result was proved earlier by the reviewer [Indag. Math., New Ser. 15, No. 3, 347--355 (2004; Zbl 1086.11019)] and in fact, the author's proof is a generalization of that of the reviewer. In the case \(K=\overline{\mathbb Q}\), the author proves a generalization of his result mentioned above where instead of \(\Gamma\), he considers a ``cylinder'' around \(\Gamma\), i.e., a set of the shape \(\Gamma_{\varepsilon}=\{ x+y: x\in\Gamma\), \(y\in A(\overline{\mathbb Q})\), \(h(y)\leq \varepsilon\}\), where \(h(\cdot )\) is an appropriate canonical height on \(A\). Further, he answers in the negative a question posed by the reviewer, dealing with the ``truncated cone'' \(C(\Gamma ,\varepsilon ):= \{ x+y: x\in \Gamma\), \(y\in A(\overline{\mathbb Q})\), \(h(y)\leq \varepsilon (1+h(x))\}\). The reviewer asked whether there exists \(\varepsilon >0\) independent of \(\Gamma\) such that the set \((X\backslash Z_X)(\overline{\mathbb Q})\cap C(\Gamma ,\varepsilon )\) is finite, and the author gives a counterexample.