Generalized greatest common divisors for orbits under rational functions (Q2287883)

The article under review studies the greatest common divisor of iterates of polynomials evaluated at integers. This line of results originated by the following result obtained by Bugeaud, Corvaja and Zannier in [\textit{Y. Bugeaud} et al., Math. Z. 243, No. 1, 79--84 (2003; Zbl 1021.11001)]: Given two multiplicatively independent integers \(a,b \geq 2\) and \(\varepsilon > 0\) there exists \(N = N(a,b,\varepsilon)\) such that \[ \gcd(a^n - 1, b^n -1) \leq \exp(\varepsilon n) \] for all \(n \geq N\). \textit{J. H. Silverman} in [Monatsh. Math. 145, No. 4, 333--350 (2005; Zbl 1197.11070)] observed that \(\gcd\) result as the one above are strictly linked to Vojta's conjecture. In particular he showed that the \(\gcd\) in Bugeaud, Corvaja and Zannier's result [loc. cit.], can be interpreted as a height function with respect to the exceptional divisor on the blow-up of the affine plane at the origin. Furthermore, he showed how, assuming Vojta's conjecture on suitable blowups, one can obtain similar estimate for the gcd in more general contexts. The author consider the situation in which the power map is replaced by a more general iteration of polynomials. He proves an analogue of Silverman estimates for the gcd of the iterates under suitable genericity hypotheses, conditional on Vojta's conjecture on a specific blow up of \(\mathbb{P}^1 \times \mathbb{P}^1\). The main result of the paper is the following Theorem. Assume Vojta's Conjecture for blowups of \(\mathbb{P}^1 \times \mathbb{P}^1\). Suppose \(a, b, \alpha, \beta \in \mathbb{Z}\), and that \(f(x), g(x) \in \mathbb{Z}[x]\) are polynomials of degree \(d \geq 2\). Assume that \(\alpha,\beta\) are not exceptional for \(f , g\) respectively. Assume that the sequence \(( f^{\circ n} (a), g^{\circ n} (b))_n\) is generic in \(\mathbb{Q}^2\) (i.e. is not contained in a proper closed subvariety). Then given \(\varepsilon > 0\), there exists a constant \(C = C(\varepsilon,a,b,\alpha,\beta, f,g) > 0\), such that for all \(n \geq 1\), we have \[ \gcd( f^{\circ n}(a) - \alpha, g^{\circ n}(b) - \beta) \leq C \cdot \exp(\varepsilon \cdot d^n). \] The idea behind the proof is to follow Silverman's approach; however there are three main difficulties that appear in the case of the previous theorem. The first is that the blow up involved is centered at a Zariski closed subset, instead of a subvariety like in Silverman's results, which depends on \(\varepsilon\). The second is that for general rational function iterates, the numerator of the iterates might not be the iterate of a polynomial. Finally one needs to control the degree of ramification of the maps.