Thue inequalities with a small number of primitive solutions (Q5948346)

For a binary form \(F(X,Y)=\prod_{i=1}^n (\alpha_iX-\beta_iY)\) with \(\alpha_i,\beta_i\in{\mathbb C}\), we define the Mahler height \(M(F):= \prod_{i=1}^n\max (|\alpha_i|,|\beta_i|)\) and the discriminant \(D(F)=\prod_{1\leq i<j\leq n} (\alpha_i\beta_j-\alpha_j\beta_i)^2\). For \(A=\Big( {a\atop c} {b\atop d}\Big)\in SL_2({\mathbb Z})\), set \(F_A(X,Y):= F(aX+bY,cX+dY)\) and \(M^*(F):=\inf_{A\in SL_2({\mathbb Z})}M(F_A)\). The author considers the so-called Thue inequality (*) \(|F(x,y)|\leq m\) in solutions \(x,y\in{\mathbb Z}\), where \(F\) is a binary form in \({\mathbb Z}[X,Y]\) of degree \(n\geq 3\) without multiple factors and \(m\) is a positive integer. A solution \((x,y)\) of this inequality is called primitive if \(x,y\) are coprime. Solutions \((x,y)\), \((-x,-y)\) are identified. \textit{E. Bombieri} and \textit{W. M. Schmidt} [Invent. Math. 88, 69-81 (1987; Zbl 0614.10018)] showed that the equation \(|F(x,y)|=1\) has at most \(cn\) solutions in \(x,y\in{\mathbb Z}\), where \(c\) is a constant. Modifying the ideas of Bombieri and Schmidt, the author proves the following results: (*) has at most \(25n\) primitive solutions \((x,y)\) with \[ \max (|x|,|y|)\geq (2^{n+2}n^{n/2}M(F)^n/|D(F)|^{1/2})^{\nu (n)}, \] where \(\nu (n)={1\over n- 2}+{1\over n^2}\). If \(M^*(F)\geq (3.5^nm^2)^{{1\over 1-\theta}}\) with \(0<\theta <1\) then (*) has at most \(25n+{\theta +8\over 4\theta}\cdot (n+2)\) primitive solutions. This can be improved to \(5n+\theta^{-1}(n+2)\) for \(n\) sufficiently large. We mention that earlier \textit{J. H. Evertse} and \textit{K. Győry} [The mathematical heritage of C. F. Gauss, World Scientific, Singapore, 204-224 (1991; Zbl 0767.11018)], by combining the ideas of Bombieri and Schmidt with linear forms in logarithms estimates, showed that the Thue inequality (*) has at most \(6n\) primitive solutions provided that \(n\geq 400\) and that \(M^*(F)>\exp\{ c(n)m^{10(n-1)^4}\}\), where \(c(n)\) is some effectively computable number depending only on \(n\). This bound is somewhat better than Győry's new result, on the other hand Győry's lower bound for \(M^*(F)\) is much better and moreover Győry doesn't use linear forms in logarithms estimates.