Unlikely, likely and impossible intersections without algebraic groups (Q2637217): Difference between revisions

Let \(G\) be a semiabelian variety defined over \({\mathbb C}\) and \(V\) an irreducible algebraic subvariety of \(G\) that is not a translate of an algebraic subgroup of \(G\) by a torsion point. For \(m\geq 0\), denote by \(G^{[m]}\) the union of all algebraic subgroups of \(G\) of codimension at least \(m\). A special case of the Zilber-Pink conjecture asserts that \(V\cap G^{[\dim V+1]}\) is not Zariski-dense in \(V\). The authors consider a function field analogue, where instead of a semi-abelian variety they consider the affine variety \(X={\mathbb A}^n\) and where instead of translates of algebraic subgroups by torsion points they take subvarieties defined over the constant field of the underlying function field. Thus, let \(k\) be any algebraically closed field (the constant field), and \(K\) a finitely generated extension of \(k\). A \textit{constant} subvariety of \(X\) is one which is defined over \(k\). It is possible to endow \(X(K)\) with a canonical height which is zero precisely on the points from \(X(k)\). This height is extended to \(X(\overline{K})\). For \(m\geq 0\), let \(X^{(m)}\) denote the union of all constant subvarieties of \(X\) of codimension at least \(m\). The main theorem of the paper under review, which may be viewed as an analogue of the Zilber-Pink conjecture, and which is proved by means of elementary algebraic geometry only, is as follows: {Theorem.} Let \(V\) be a proper absolutely irreducible subvariety of \(X\) defined over \(K\) that is not contained in a proper, constant subvariety of \(X\). Then \(V\cap X^{(\dim V +1)}\) is not Zariski-dense in \(V\). In fact, \(V\cap X^{(\dim V+1)}\) is contained in a proper, constant subvariety of \(X\). The authors conjecture a refinement of this theorem as follows. Call a subvariety \(W\) of \(V\) \textit{constant-anomalous} if there is a constant subvariety \(Y\) of \(X\) such that \(W\subseteq V\cap Y\) and \(\dim W>\max (0,\) \(\dim V+\dim Y-n)\) (i.e., the dimension of \(W\) is larger than expected). Define the constant-anomalous part \(Z_V^0\) of \(V\) in \(X\) to be the union of all constant-anomalous subvarieties \(W\) of \(V\). { Conjecture.} \(Z_V^0\) is Zariski closed in \(V\), and \(F_V^0:=(V\setminus Z_V^0)\cap X^{(\dim V+1)}\) is finite. The authors give more precise conjectures on the cardinality of \(F_V^0\) and the heights of the points in \(F_V^0\). The authors prove the finiteness of \(F_V^0\) in the case that \(V\) is a curve in \(X\) and \(K\) is purely transcendental over \(k\). The authors also propose some conjectures implying that certain subsets of subvarieties \(V\) of \(X\) have bounded height. \textit{E. Bombieri} et al. [Int. Math. Res. Not. 1999, No. 20, 1119--1140 (1999; Zbl 0938.11031)] proved that if \(V\) is a curve over \(\overline{{\mathbb Q}}\) in the linear torus \(G={\mathbb G}_m^n\), which is not a translate of a proper algebraic subgroup of \(G\), then the \(\overline{\mathbb{Q}}\)-rational points of \(V\cap G^{[1]}\) have bounded height. The authors of the paper under review argue, that the appropriate analogue of a translate of a proper algebraic subgroup in their setting is a quasi-constant variety, where a subvariety of \(X\) is called quasi-constant if it is defined over a field of transcendence degree at most \(1\) over \(k\). This leads to the following conjectural analogue to the result of Bombieri et al. [loc. cit.]: {Conjecture.} Let \(V\subset X\) be an absolutely irreducible curve that is not contained in a proper, quasi-constant subvariety of \(X\). Then the \(\overline{K}\)-rational points of \(V\cap X^{(1)}\) have bounded height. The authors also formulate a generalization of this conjecture, and prove their conjecture and its generalization for certain concrete examples.