Constructing one-parameter families of elliptic curves with moderate rank (Q2370190)

Let \(\mathcal E\) be the elliptic \(\mathbb Q(T)\)-curve \[ y^{2}+a_{1}(T)xy+a_{3}(T)y=x^{3}+a_{2}(T)x^{2}+a_{4}(T)x+a_{6}(T), \] where \(a_{i}(T)\in \mathbb Z[T]\) for every \(i\). By a theorem of Silverman, \(\text{rank}_{\mathbb Z}\,\mathcal E_{t}(\mathbb Q)\geq \text{rank}_{\mathbb Z}\,\mathcal E(\mathbb Q(T))\) for sufficiently large \(t\in\mathbb Z\), where \(\mathcal E_{t}\) is the specialization of \(\mathcal E\) at \(t\). Now, for any prime number \(p\), define \[ A_{\mathcal E}(p)=\frac{1}{p}\sum_{t=0}^{p-1}\{p+1-N_{t}(p)\}, \] where \(N_{t}(p)\) equals \(p+1\) if \(\mathcal E_{t}\) has bad reduction at \(p\) and \(\#\mathcal E_{t}(\mathbb F_{p})\) otherwise. \textit{M. Rosen} and \textit{J. Silverman} [Invent. Math. 133, No. 1, 43--67 (1998; Zbl 0905.14019)] have shown that if \(\mathcal E\) is given by \( y^{2}=x^{3}+a_{4}(T)x+a_{6}(T)\) and Tate's conjecture holds for \(\mathcal E\) (this is the case if \(\mathcal E\) is a rational elliptic surface over \(\mathbb Q\)), then \[ \lim_{X\rightarrow\infty}\frac{1}{X}\sum_{p\,\leq X}\,A_{\mathcal E}(p)\,\text{log}\,p^{-1}=\text{rank}_{\mathbb Z}\,\mathcal E(\mathbb Q(T)). \] In this paper the authors construct special rational elliptic surfaces for which they are able to compute \(A_{\mathcal E}(p)\) explicitly. They show that \(A_{\mathcal E}(p)=-r+O(1/p)\) for small values of \(r\) (e.g. \(r=6\)) and conclude that \(\text{rank}_{\mathbb Z}\,\mathcal E(\mathbb Q(T))=r\). The novelty of the approach is that by forcing \(A_{\mathcal E}(p)\) to be essentially constant (provided \(\mathcal E\) is a rational elliptic surface over \(\mathbb Q\)), the authors can immediately calculate the Mordell--Weil rank without having to specialize points and calculate height matrices. Further, they obtain an exact answer for the rank, not just a lower bound. If the degrees of the polynomials defining \(\mathcal E\) are too large, the results of the paper are conditional on Tate's Conjecture provided \(A_{\mathcal E}(p)\) can be evaluated. However, in many cases the authors are unable to evaluate \(A_{\mathcal E}(p)\) to the needed accuracy. Their method does generate candidate points, which upon specialization yield lower bounds for the rank. An example is given where the rank is at least 8.