Diagonalizable Thue equations: revisited (Q6073939)

Let \(F(X,Y)\in\mathbb{Z}[X,Y]\) be a diagonalizable binary form of degree \(r\), i.e., \(F(X,Y)=(\alpha X +\beta Y)^r-(\gamma X+\delta Y)^r\) for certain algebraic numbers \(\alpha ,\beta ,\gamma ,\delta\) with \(j:=\alpha\delta -\beta\gamma\not =0\). One can show that \((\alpha X+\beta Y)(\gamma X+\delta Y)=\chi (AX^2+BXY+CY^2)\) with \(\chi^r\in\mathbb{Q}\) and \(A,B,C\in\mathbb{Z}\). Let \(D:=B^2-4AC\), so that \(j^2=\chi^2D\). Denote by \(\Delta\) the discriminant of \(F\), so \(\Delta =\pm r^rj^{r(r-1)}\). The authors consider so-called diagonalizable Thue inequalities \[ 0<|F(x,y)|\leq h. \] Denote by \(N_F(h)\) the number of primitive solutions of the above inequality, i.e., with \(x,y\in\mathbb{Z}\) and \(\gcd (x,y)=1\), where solutions \((x,y)\), \((-x,-y)\) are considered equal. Under the hypothesis that \(|\Delta|\) exceeds some explicitly given bound in terms of \(h\) and \(r\), \textit{C. L. Siegel} [Nachr. Akad. Wiss. Gött., II. Math.-Phys. Kl. 1970, 169--195 (1970; Zbl 0215.34601)] derived an explicit upper bound for \(N_F(h)\). This bound is linear in terms of \(r\) if \(D<0\) and an absolute constant if \(D>0\). To achieve this, Siegel used his own Diophantine approximation method based on hypergeometric functions, that he developed in the 1920-s and 1930-s and that builds further on work of Thue. \textit{S. Akhtari} et al. [Ramanujan J. 45, No. 2, 521--567 (2018; Zbl 1412.11061)] obtained an improvement of Siegel's result. In the paper under review Saradha and Sharma improve this further, by refining some arguments. Their result reads as follows. \textbf{Theorem}. Assume \(r\geq 7\) and \(|\Delta |\geq a(r)\cdot h^{b(r)}\), where \(a(r):= 2^{r^2-r}r^r\cdot r^{13r^2(r-1)/(r^2-5r-2)}\), \(b(r):=2r-2+4(r-1)(r^2-r+2)/(r^2-5r-2)\). Then (i) \(N_F(h)\leq 2r\) if \(D<0\); (ii) \(N_F(h)\leq 4\) if \(D>0\), \(r\) is even, and \(F\) is indefinite; (iii) \(N_F(h)\leq 2\) if \(D>0\), \(r\) is odd, and \(F\) is indefinite; (iv) \(N_F(h)\leq 1\) if \(D<0\) and \(F\) is definite. In their earlier paper, Akhtari, Saradha and Sharma had the same result, except that in the cases (i), (ii), (iii) they had the upper bounds \(2r+1\), \(5\), \(3\), respectively, under the weaker assumption \(r\geq 6\).