Intersecting subvarieties of \({G_m^n}\) with algebraic subgroups (Q943363)

The author continues the research, initiated by \textit{E. Bombieri, D. Masser} and \textit{U. Zannier} [Int. Math. Res. Not. 1999, No. 20, 1119--1140 (1999; Zbl 0938.11031)], on the intersections of subvarieties of the \(n\)-dimensional linear torus \({\mathbb G}_m^n\) with algebraic subgroups of the linear torus. Denote by \({\mathcal H}_m\) the union of all algebraic subgroups of \({\mathbb G}_m^n\) of dimension at most \(m\). Further, denote by \(h\) the logarithmic Weil height on \({\mathbb G}_m^n(\overline{{\mathbb Q}})\), and then define \(C({\mathcal H}_m,\epsilon )\) to be the set of \(p\in {\mathbb G}_m^n(\overline{{\mathbb Q}})\) for which there exist \(a\in {\mathcal H}_m\), \(b\in {\mathbb G}_m^n(\overline{{\mathbb Q}})\) such that \(p=a\cdot b\) and \(h(b)<\epsilon (1+h(a))\). Let \(X\) be a subvariety of the \(n\)-dimensional linear torus \({\mathbb G}_m^n\), defined over \(\overline{{\mathbb Q}}\). A subvariety \(Y\) of \(X\) is called \textit{anomalous}, if it is an irreducible component of an improper intersection of \(X\) with a translate of an algebraic subgroup of \({\mathbb G}_m^n\), i.e., if there is a translate \(H\) of an algebraic subgroup of \({\mathbb G}_m^n\) such that \(Y\subset H\) and \(\text{codim} Y\leq\min \big(n-1 ,\text{codim} X+\text{codim} Y-1\big)\). Denote by \(X^{\text{oa}}\) the subset of \(X\), obtained be removing the union of all its anomalous subvarieties defined over \(\overline{{\mathbb Q}}\). It is known that if \(U\) is a Zariski open subset of \(X\) defined over \(\overline{{\mathbb Q}}\), then the set of \(\overline{{\mathbb Q}}\)-points of \(\big(X\setminus U\big)\cap {\mathcal H}_m\) has unbounded height if \(m\geq n-\dim X+1\); the author recalls the simple proof. Bombieri, Masser and Zannier proved that \(X^{\text{oa}}\) is Zariski open in \(X\), and stated the conjecture that the set of \(\overline{{\mathbb Q}}\)-rational points of \(X^{\text{oa}}\cap{\mathcal H}_m\) has bounded height if \(m\leq n-\dim X\). This conjecture has been proved in the cases \(\dim X=1,2\) (Bombieri, Masser and Zannier) and \(\dim X =n-1\) (Bombieri, Zannier). The author formulates the stronger conjecture that there exists \(\epsilon >0\) such that \(X^{\text{oa}}\cap C({\mathcal H}_m,\epsilon )\) has bounded height if \(m\leq n-\dim X\). Further, he proves among other things, the following explicit version of a special case of his conjecture; \(\deg X\) and \(h(X)\) denote the degree and height of \(X\): Put \(s:=\dim X\), \(c(n):= (2n)^{30n^2}\), and let \(m\) be an integer with \(m<n/s\). Then if \(0\leq \epsilon < c(n)^{-1}\cdot (\deg X)^{-n/(n-ms)}\) and \(p\in X^{\text{oa}}\cap C({\mathcal H}_m,\epsilon )\), we have \(h(p)\leq c(n)(\deg X)^{ms/(n-ms)}\cdot (h(X)+\deg (X))\). In his proof, the author uses geometry of numbers and the arithmetic Bézout Theorem.