Unlikely intersections for curves in products of Carlitz modules (Q2163296): Difference between revisions

The paper being reviewed continues the research on the Zilber-Pink problem, which generalizes the results of Manin-Mumford and Mordell-Lang. This problem was originally studied over zero characteristic, and the present paper focuses on its study in positive characteristic. The second author of the present paper initiated this research in [\textit{D. Masser}, Q. J. Math. 65, No. 2, 505--515 (2014; Zbl 1317.11067)], focusing on curves in multiplicative groups. The study was further developed by the authors of the current paper in [Proc. Am. Math. Soc. 145, No. 11, 4617--4627 (2017; Zbl 1401.11110)], this time examining curves in additive groups equipped with an ``\(F\)-structure'', where ``\(F\)'' denotes the action of the Frobenius map. The present paper extends this exploration by considering curves in additive groups equipped with the Carlitz action. Let \(p\) be a fixed prime number, and let \(K\) be an algebraically closed field containing \({\mathbb F}_{p}(t)\). Let \(C\) be an irreducible curve in \({\mathbf G}_{a}^{n}\) defined over \(K.\) The problem is to understand how \(C\) intersects with algebraic subgroups of \({\mathbf G}_{a}^{n}\) with suitable codimensions. Over positive characteristic, there are well-known counterexamples to the naive analogues of the conjectures due to \textit{B. Zilber} [J. Lond. Math. Soc., II. Ser. 65, No. 1, 27--44 (2002; Zbl 1030.11073)] and to \textit{R. Pink} [``A common generalization of the conjectures of André-Oort, Manin-Mumford, and Mordell-Lang'', Preprint, \url{https://people.math.ethz.ch/~pink/ftp/AOMMML.pdf}] for situations over zero characteristic. The results of the two papers mentioned in the first paragraph suggest that extra structures on \({\mathbf G}_{a}^{n}\) are needed to have an analogue parallel to the situation over zero characteristic. Recall that the additive group \({\mathbf G}_{a}(K)\) has the structure of a Carlitz module under the action of the ring \(R := {\mathbb F}_{p}[{\mathcal C}]\), where \({\mathcal C} x = t x + x^{p}\) for \(x\in {\mathbf G}_{a}(K)\). The action of \(R\) naturally extends to \({\mathbf G}_{a}^{n}(K)\) by acting diagonally, i.e., \(\alpha (x_{1},\ldots, x_{n}) = (\alpha x_{1}, \ldots , \alpha x_{n})\) for \(\alpha \in R\). Here, an algebraic subgroup of \({\mathbf G}_{a}^{n}\) is an \(R\)-submodules defined by several equations of the form \(\alpha_{1} x_{1} + \cdots + \alpha_{n} x_{n} = 0\) where \(\alpha_{1}, \ldots, \alpha_{n}\) are in \(R.\) In this setting, the authors formulate a conjecture that is parallel to the situation over zero characteristic. Conjecture: assume for any non-zero \((\rho_{1},\ldots, \rho_{n})\) in \(R^{n}\) that the form \(\rho_{1}x_{1} + \cdots + \rho_{n} x_{n}\) is not identically zero on \(C\). Then there are at most finitely many \((\xi_{1},\ldots,\xi_{n})\) in \(C(K)\) for which there exist linearly independent \((\alpha_{1}, \ldots, \alpha_{n}), (\beta_{1}, \ldots, \beta_{n}\)) in \(R^{n}\) such that \(\alpha_{1} \xi_{1} + \cdots + \alpha_{n} \xi_{n} = \alpha_{1} \xi_{1} + \cdots + \alpha_{n} \xi_{n} = 0\). The main result of this paper is a proof for the case when \(n=3\). It is worth noting that there is nothing to prove if \(n=1\), and the case when \(n=2\) has already been proven by \textit{T. Scanlon} in the general context of Drinfeld modules, using techniques from model theory [J. Number Theory 97, No. 1, 10--25 (2002; Zbl 1055.11037)]. The proof of the main result closely follows the strategy presented in [\textit{E. Bombieri} et al., Int. Math. Res. Not. 1999, No. 20, 1119--1140 (1999; Zbl 0938.11031)]. The main ingredient is a new relative version of Dobrowolski's classical lower bound for canonical heights in the Carlitz context, along with a suitable upper bound. For higher-dimensional cases, the authors suggest possible approaches to tackle the problem.