On some extensions of the Ailon-Rudnick theorem (Q328599): Difference between revisions

The polynomials \(F_1,\ldots,F_s\in {\mathbb C}[X_1,\ldots,X_l]\) are called {\textit{multiplicatively independent}} in the group \({\mathbb C}(X_1, \ldots,X_l)^{\ast}/{\mathbb C}^{\ast}\) if there exists no nonzero vector \((\nu_1,\ldots,\nu_s)\in{\mathbb Z}^s\) and \(a\in{\mathbb C}^{\ast}\) such that \(F_1^{\nu_1}\cdots F_s^{\nu_s}=a\). The Ailon-Rudnick theorem [\textit{N. Ailon} and \textit{Z. Rudnick}, Acta Arith. 113, No. 1, 31--38 (2004; Zbl 1057.11018)] establishes that if \(f,g\in{\mathbb C}[T]\) are multiplicatively independent polynomials, then there exists \(h\in{\mathbb C}[T]\) such that \(\gcd(f^n-1,g^n-1)\mid h\) for all \(n\in{\mathbb N}\). The argument of Ailon and Rudnick can be easily generalized to that there exists \(h_1\in{\mathbb C}[T]\) such that \(\gcd(f^n-1,g^m-1)\mid h_1\) for all \(n,m\in{\mathbb N}\). The purpose of this paper is to present extensions of the result of Ailon and Rudnick to both, the univariate and the multivariate cases. The main results are the following. If \(f,g,h_1,h_2\in{\mathbb C}[T]\) satisfy that \(f\) and \(g\) are multiplicatively independent in \({\mathbb C}(T)^{\ast}/{\mathbb C}^{\ast}\), then for all \(n,m\in{\mathbb N}\), we have \(\deg\gcd(h_1(f^n),h_2(g^m))\leq d_{h_1}d_{h_2} (11 d_{\ast}(d_f+d_g)^2)^{d_{\ast}}\), where \(d_{\ast}= \min\{d_f,d_g\}\) and \(d_P\) is the degree of the polynomial \(P\). Similarly, for the multivariate case we have the following result. Let \(h_1,h_2\in{\mathbb C}[T]\), \(F,G\in{\mathbb C}[X_1,\ldots,X_l]\) and \(D=\max_{i=1,\ldots,l}\{\deg_{X_i} F,\deg_{X_i} G\}\). If \(F,G\) are multiplicatively independent in \({\mathbb C}(X_1,\ldots, X_l)^{\ast}/{\mathbb C}^{\ast}\), then for all \(n,m\in{\mathbb N}\), we have \[ \deg\gcd(h_1(F^n), h_2(G^m))\leq d_{h_1}d_{h_2}(44 (D+1)^{2l})^{(D+1)^l}. \] The other main results in both the univariate and the multivariate cases are the following. Let \(Z(f)\) denote the set of zeros of \(f\) in \({\mathbb C}\). Consider \(f_1,\ldots,f_l,\varphi_1,\ldots, \varphi_k,g_1,\ldots,g_r,\psi_1,\ldots,\psi_s\in{\mathbb C}[T]\) multiplicatively independent polynomials such that \(Z(f_1\cdots f_l)\cap Z(\varphi_1\cdots\varphi_k)=\emptyset\), \(Z(g_1\cdots g_r)\cap Z(\psi_1\cdots \psi_s)=\emptyset\). Then for all \(n_1,\ldots,n_l,\nu_1,\ldots,\nu_k,m_1,\ldots,m_r,\mu_1,\ldots, \mu_s\in{\mathbb N}\cup\{0\}\), there exists a polynomial \(h\in {\mathbb C}[T]\) such that \[ \gcd\big(\prod_{i=1}^l f_i^{n_i}-\prod_{i=1}^k\varphi_i^{\nu_i}, \prod_{i=1}^r g_i^{m_i}-\prod_{i=1}^s \psi_i^{\mu_i}\big)\mid h. \] Similarly, for the multivariate case. One of the main tools used in the paper is a result on the finiteness of the number of points on the intersection of a curve in \({\mathbb G}_m^k\) with algebraic subgroups of codimension at least \(2\). Properties of Hilbertian fields and Hilbert's irreducible theorem are also used. The hypersurface case of the description of algebraic subgroups containing the torsion points on algebraic varieties is important in this paper.