Multiple fibres on surfaces: geometry, hyperbolic and arithmetic aspects (Q2388441): Difference between revisions

For surfaces \(S\) of general type the Lang conjectures state that there should be a (reducible) curve \(E\subset S\) which contains the image of every non-constant holomorphic map from the complex line to \(S\). Moreover, for every number field \(K\) over which \(S\) is defined almost all \(K\)-rational points of \(S\) should lie in \(E\). There is also a function field analog of this conjecture (which as usual is more readily accessible than its arithmetic counterpart). For a curve \(B\) as base space it amounts to the following: If \(g:X\to B\) is a surjective proper connected morphism (``fibration'') whose generic fibers are surfaces of general type, then there should exist a subvariety \(D\) such that there are only finitely many ``non-constant'' rational sections \(\sigma:B\to X\) with \(\sigma(B)\not\subset D\). (Rational sections correspond to rational points over the function field of \(B\).) Since the Lang conjectures are hard to tackle, one makes additional assumptions like the existence of a morphism from the surface onto a curve \(C\). This assumption is very useful if this curve is of general type, too. In the spirit of his related work [Ann. Inst. Fourier 54, 499--665 (2004; Zbl 1062.14014)] \textit{F. Campana} regards the base of fibrations as equipped with an orbifold structure. The main purpose of the article under review is to demonstrate in the function field case, that instead of requiring \(C\) to be of general type, it suffices to make the weaker assumption that \(C\) is of general type in the orbifold sense. Théorème 3.20 proves a version of the function field case of the Lang conjecture under such an assumption. In preparation for the proof of this, Campana establishes in Théorème 3.8 an orbifold function field version of the ``Mordell conjecture'' (theorem of Faltings). The easiest examples of fibrations of a surface \(S\) onto a curve \(C\) such that the base is of general type only in the orbifold sense arise as follows: Start with a fibration \(S_0\to C_0\) where \(C_0\) is of general type and such that there is a finite group \(\Gamma\) acting on both \(S_0\) and \(C_0\) such that the projection map is equivariant. If the action is fixed point free on \(S_0\), then the quotient \(S=S_0/\Gamma\) is of the same Kodaira dimension as \(S_0\). If the action on \(C_0\) has fixed points, it is quite likely that \(C_0/\Gamma\) is not of general type in the usual sense, but still of general type in the orbifold sense. Note that by construction \(S\) is not simply-connected. Thus théorème 5.1 of the article shows that there are other such fibrations, namely that there is a simply-connected surface of general type \(S\) which admits a fibration onto a curve. This curve is simply \({\mathbb P}_1\) if one ignores the orbifold structure, but it its of general type in the orbifold sense. In addition, the article discusses certain facts and conjectures related to the analytic und arithmetic orbifold versions of the Lang conjectures.