Algebraic hyperbolicity of ramified covers of \(\mathbb{G}^2_m\) (and integral points on affine subsets of \(\mathbb{P}_2\)) (Q360075): Difference between revisions

The authors study the affine curves with bounded genus and number of points at infinity on smooth affine surface \(X\), where \(\pi: X \to \mathbb{G}_{m}^{2}\) is a finite morphism, and obtain bounds for their degree in terms of Euler characteristic \(\chi_{\mathcal{C}} = 2g - 2 + \sharp S\). Here \(g\) is the genus of the smooth projective curve \(\tilde \mathcal{C}\), \(S\) is a finite nonempty subset of \(\tilde \mathcal{C} \) of the cardinality \(\sharp S\), \(\mathcal{C} = {\tilde \mathcal{C}} \backslash S\). A typical example where these bounds hold is represented by the complement of a three-component curve in the projective plane of total degree at least 4. The corresponding results may be interpreted as bounding the height of integral points on \(X\) over a functional fields. In the language of Diophantine Equations, authors results may be rephrased in terms of bounding the height of the solutions of \(f(u,v,y) = 0\), with \(u, v, y\) over a functional field, \(u, v\) \(S\)-units. It turns out that all of this contain some cases of strong version of a conjecture by \textit{P. Vojta} [Diophantine approximations and value distribution theory. Berlin etc.: Springer-Verlag (1987; Zbl 0609.14011)] over function fields in the split case. Moreover, the authors' methods would apply also to the nonsplit case. The authors of the paper under review remark that special cases of their results in the holomorphic context were studied by M. Green already in the seventies, and recently in greater generality by Noguchi, Winkelmann and Yamanoi [\textit{J. Noguchi} et al., J. Math. Pures Appl. (9) 88, No. 3, 293--306 (2007; Zbl 1135.32018)]; however, the algebraic context was left open and seems not to fall in the existing techniques. Let us now describe contents and results of this paper in more detail. Let \(\kappa\) be an algebraically closed field of characteristic zero, let \(\mathcal{O}_{S}= \kappa [\mathcal{C}]\) be the ring of regular functions on the affine curve \(\mathcal{C}\). The integral points \(X(\mathcal{O}_{S})\) on the surface \(X\) correspond to curves \(Y\) (not necessary smooth) lying on \(X\), parameterized by the smooth curve \(\mathcal{C}\), and authors put \(\chi_{Y} = \chi_{\mathcal{C}}\). Theorem 1. (Voita's conjecture for \({\mathbb P}_{2}\) minus three divisors). Let \(D = D_{1} + \dots + D_{r} \subset {\mathbb P}_{2}\) be a curve whose components \(D_{i}\) meet transversally at each point of intersection. Suppose \(r \geq 3\) and \(\deg(D) \geq 4\). Then there exists a number \(C_{1} = C_{1}(D)\) such that for every curve \(Y \subset {\mathbb P}_{2} \backslash D \), we have the following bound: \(\deg(Y) \leq C_{1}\cdot \max\{1, \chi_{Y}\}\). Let \(\tilde X\) be a smooth complete surface, \(K_{\tilde X }\) be a canonical divisor on \(\tilde X, X = \tilde X \backslash D.\) Recall follow to \textit{O. Debarre} [Higher-dimensional algebraic geometry. New York, NY: Springer (2001; Zbl 0978.14001)] that \(X\) is log-general type if \(h^{0}(n(D + K)) \gg n^2.\) Theorem 2. Let \(Z \subset X\) be the ramification divisor of the finite map \(\pi: X \to \mathbb{G}_{m}^{2}.\) Assume that the closure of \(\pi(Z)\) in \({\mathbb P}_{2}\) does not intersect the set of singular points of the boundary of \(\mathbb{G}_{m}^{2}\) in \({\mathbb P}_{2}\) -- i.e., \((0:0:1), (0:1:0), (1:0:0).\) If \(X\) is of log-general type, then there exists a number \(C_2 = C_2(X, \pi)\) such that for every curve \(Y \subset X\) of Euler characteristic \(\chi_{Y}\) the following inequality holds: \(\deg(Y) \leq C_{2}\cdot \max\{1, \chi_{Y}\}\). The surface \(X\) is defined in \(\mathbb{G}_{m}^{2} \times \mathbb{A}^1\) up to birationality, an equation of the form \(f(u, v, y) = 0.\) Theorem 3. Let \(f(U, V, Y) \in \kappa[U, V, Y] \) be an irreducible polynomial, monic in \(Y.\) Suppose that the discriminant \(\Delta(U, V) \in \kappa[U, V] \) of \(f\) with respect to \(Y\) has no multiple nonmonomial factors. Then, for an affine curve \(\mathcal{C}\) as above, one of the following cases occurs: (a) There exist numbers \(C_{3}, C_{4},\) effectively computable in terms of \(f\) and \(\chi = \chi_{\mathcal{C}} \), such that the solution \((u, v, y) \in \mathcal{O}_{S}^{*} \times \mathcal{O}_{S}^{*} \times \mathcal{O}_{S}\) of the equation \(f(u, v, y) = 0\) satisfy either \[ \max \{\deg(u), \deg(v), \deg(y)\} \leq C_3 \] or a multiplicative dependence relation \(u^a = \lambda \cdot v^b\) for a pair \((a, b) \in \mathbb{Z} \backslash \{(0, 0) \}\) with \(\max \{ |a|, |b|\} \leq C_4\) and \(\lambda \in \kappa^*. \) (b) After an automorphism of \(\mathbb{G}_{m}^{2},\) there is a \(Q \in \kappa[U]\) such that \(\Delta(U, V) = Q(U)V^a\) and \(f(U, V, Y) = V^{l}P(U, V^{m}Y + A(U, V)),\) where \(P \in \kappa[U^{\pm 1}, W], A(U, V) \in \kappa[U^{\pm 1}, V^{\pm 1}, a, l, m \in \mathbb{Z}, a \geq 0.\) Also, either \(f(U, V, Y) = (Y + A(U, V))^d - bV^n U^p,\) where \(b \in \kappa^* , d, n, p\) are positive integers, or \(\deg(u) \leq C_5,\) and the number of possible \(u\) is finite, bounded only in terms of \(\deg f\) and \(\chi.\) In Section 2 the authors of the paper under review are concerned with the connections between ``geometric'' and ``Diophantine'' languages, present preliminaries on heights and the proof of Theorem 2. The main result on the ramification divisor \(Z \subset X\) of \(\pi: X \to \mathbb{G}_{m}^{2}\) which is needed in the proof of Theorem 2 is as following: Proposition 1. Let \(X, Z\) be as above; for every \(\epsilon > 0\) and every integer \(\chi \) there exists a number \(C = C(X, \pi, \epsilon, \chi) = C'(X, \pi, \epsilon) \cdot \max(1, \chi)\) such that for every morphism \(\varphi : \mathcal{C} \to X\) (with \( \chi_{\mathcal{C}} = \chi\)) of height \(H > C,\) with \(\varphi(\mathcal{C}) \not \subset Z,\) the degree of the divisor \(\varphi^{*}(Z) \) satisfies \(\deg (\varphi^{*}(Z)) \leq \epsilon H.\) In particular, the number of points \(p \in \mathcal{C}\) such that \(\varphi(p) \in Z\) is bounded by \(\epsilon H.\) The proof of Theorem 2 rests also on the \(\mathrm{gcd}\) estimates of the authors [J. Algebr. Geom. 17, No. 2, 295--333 (2008); addendum Asian J. Math. 14, No. 4, 581--584 (2010; Zbl 1221.11146)]. In Sections 3 and 4 they present correspondingly proofs of Theorem 1 and Theorem 3. Finally, Section 5 contains interesting examples as well as a counter-example, showing that the condition on the normal crossing of the of the divisor \( D_{1} + \dots + D_{r}\) in Theorem 1 cannot be omitted. The paper should be read by everybody interested in algebraic hyperbolicity and Vojta's conjecture, not only because of the plenitude of information that it contains but also because of the examples and counter-example at the end.