A Wirsing-type theorem for numerically equivalent divisors (Q6116894)

Roth's Theorem from 1955 concerns the approximation of algebraic numbers by rationals. In 1969, Wirsing extended Roth's theorem and proved a finiteness result dealing with the approximation of a given algebraic number by algebraic numbers of bounded degree. Some years later, Schmidt proved his Subspace Theorem, implying a higher dimensional generalization of Roth's Theorem. Since then, various people, e.g., Locher and Schickewei, Ru and Wang, Levin and Grieve proved versions of Schmidt's Subspace Theorem, where the unknowns ar algebraic points of bounded degree. In the paper under review, the author improves a result of \textit{A. Levin} [Duke Math. J. 163, No. 15, 2841--2885 (2014; Zbl 1321.11073)]. We explain the author's result. Let \(k\) be a number field and \(|\cdot |_v\) (\(v\in M_k)\) the normalized absolute values of \(k\), i.e., \(|\cdot |_v=|\cdot |_p^{[k_v:\mathbb{Q}_p]/[k:\mathbb{Q}]}\) if \(v\) lies above \(p\in\{\infty\}\cup\{ \mathrm{primes}\}\). Let \(X\) be a projective subvariety of \(\mathbb{P}^N\) defined over \(k\). To every Cartier divisor \(D\) of \(X\) we associate a height function \(h_D\), defined up to a bounded quantity, defined on \(X(\overline{k})\) and local Weil functions \(\lambda_{D,v}: X(k)\setminus \mathrm{Supp}\, D\to\mathbb{R}\) satisfying \(h_D(P)=\sum_{v\in M_k} \lambda_{D,v}(P)+O(1)\) for \(P\in X(k)\setminus \mathrm{Supp}\, D\). Similar identities hold for points of \(X\) defined over finite extensions of \(k\) and Weil functions defined over such extensions. Both \(h_D\) and the Weil functions are linear in \(D\) up to a bounded function. If \(D\) is a divisor of \(\mathbb{P}^N\) given by \(f=0\), where \(f\in k[x_0,\ldots ,x_N]\) is a non-zero homogeneous polynomial of degree \(d\) and \(P=(x_0:\cdots :x_N)\in \mathbb{P}^N(k)\) with \(f(P)\not= 0\), then \(\lambda_{D,v}(P)=\log (\| f\|_v\cdot \max_i |x_i|_v^d/|f(P)|_v)\), where \(\|f\|_v=\max |\mathrm{coeff\ of } f|_v\). We say that effective ample Cartier divisors \(D_1,\ldots ,D_q\) of \(X\) are in \(m\)-subgeneral position if for each subset \(I\) of \(\{ 1,\ldots ,q\}\) of cardinality \(|I|\leq m+1\), the dimension of \(\bigcap_{i\in I} \mathrm{Supp}\, D_i\) is at most \(m-|I|\). For divisors \(D_1,D_2\) we write \(D_1\equiv D_2\) if they are numerically equivalent. Then the author's main theorem is as follows: Theorem. Assume the following: let \(X\) be a projective subvariety of \(\mathbb{P}^N\) of dimension \(n\) defined over \(k\), \(m\geq n\), \(D_1,\ldots ,D_q\) effective ample Cartier divisors on \(X\) defined over \(k\) which are in \(m\)-subgeneral position and for which there are a Cartier divisor \(A\) on \(X\) defined over \(k\) and positive integers \(d_1,\ldots d_q\) such that \(D_i\equiv d_iA\) for \(i=1,\ldots ,q\), \(S\) a finite set of places of \(k\) containing the Archimedean places, and \(\delta\) a positive integer. Consider the inequality \[ \sum_{v\in S}\sum_{w\in M_{k(P)}, w|v}\sum_{i=1}^q \frac{\lambda_{D_i,v}(P)}{d_i}\geq Ch_A(P)\tag{1} \] to be solved in points \(P\in X(\overline{k})\setminus\bigcup_{i=1}^q\mathrm{Supp}\, D_i\) with \([k(P):k]\leq \delta\). \\ Then (1) has only finitely many solutions, provided \[ C>C(m,n,\delta ):= \left\{ \begin{array}{ll} (\delta m-\delta n+1)(\delta n+1) &\text{if } n\leq m/2, \\ (1+\delta m/2)^2 &\text{if } n>m/2\ \text{and } \delta m\ \text{is even}, \\ (1+[\delta m/2])(2+[\delta m/2]) &\text{if }n>m/2\ \text{and } \delta m\ \text{is odd}. \end{array}\right. \] Levin's result is of a similar nature, but with a slightly larger lower bound for \(C\). In his proof the author follows Levin, but he inserts recent ideas of \textit{Q. Ji} et al. [Math. Ann. 373, No. 3--4, 1457--1483 (2019; Zbl 1444.32017)]. It is conjectured that a result like above holds with a lower bound for \(C\) linear in \(\delta\). The author deduces a finiteness result for certain Diophantine inequalities, so-called semi-decomposable form inequalities, where the solutions are points of degree \(\leq \delta\) over \(k\) and whose coordinates are integral over the ring of \(S\)-integers \(O_S\) of \(k\).