Variations on a theme of Runge: effective determination of integral points on certain varieties (Q1011962)

The theorem of Runge referred to in the title is the following: \textbf{Theorem~1.1 (Runge).} Let \(f \in {\mathbb Q}[x,y]\) be an absolutely irreducible polynomial of total degree \(n\). Let \(f_0\) denote the leading form of \(f\), i.e., the sum of the terms of total degree \(n\) in \(f\). Suppose that \(f_0\) factors as \(f_0 = g_0 h_0\), where \(g_0, h_0 \in {\mathbb Q}[x,y]\) are nonconstant relatively prime polynomials. Then the set of solutions to \[ f(x,y)=0, x,y \in {\mathbb Z} \] is finite and can be effectively determined. This article describes various generalisations and applications of the above theorem. After defining an appropriate notion of integral point (or, more generally, \(S\)-integral point), Theorem~2.1 describes integral points on \(X \setminus D\), where \(X\) is any smooth projective variety and \(D\) an effective divisor on \(X\); the behaviour varies according to the Kodaira--Iitaka dimension of the prime components of \(D\), and the combinatorial structure of how these prime components intersect. Section~3 investigates applying Runge's method to an unramified covering \(X \to C\) of a curve \(C\), when \(C\) itself does not satisfy the required hypotheses. The method applies, roughly, when there is a large-rank rational torsion subgroup in the Jacobian of \(C\), and \(C\) has relatively few places of bad reduction; a precise criterion is given in Theorem~3.1. In particular, this method can be applied to studying the \(S\)-integral points of superelliptic curves, and Theorem~4.1 describes this application, with an explicit height bound making it feasible to find all \(S\)-integral points on such a curve. Section~5 gives specific applications to perfect powers in arithmetic progressions.