Quartic index form equations and monogenizations of quartic orders (Q6044274)

\textit{K. Györy} [Publ. Math. Debr. 23, 141--165 (1976; Zbl 0354.10041)] proved in an effective form that if \(O\) is an order in a number field \(K\), then up to equivalence there are only finitely many \(\alpha\) with \(O=\mathbb{Z}[\alpha ]\). Here, \(\alpha\) and \(\beta\) are called equivalent if \(\beta =\pm\alpha +c\) for some \(c\in\mathbb{Z}\). An equivalence class of \(\alpha\) with \(\mathbb{Z} [\alpha ]=O\) is called a monogenization of \(O\). Letting \(\{ 1,\omega_2,\ldots , \omega_n\}\) be a \(\mathbb{Z}\)-basis of \(O\), there is a homogeneous polynomial \(I\in\mathbb{Z}[X_2,\ldots , X_n]\) of degree \(n(n-1)/2\), the index form associated with \(\omega_2,\ldots ,\omega_n\) such that \(\alpha =x_1+x_2\omega_2+\cdots +x_n\omega_n\) with \(x_i\in\mathbb{Z}\) satisfies \(O=\mathbb{Z}[\alpha ]\) if and only if \[ |I(x_2,\ldots ,x_n)|= 1.\tag{*} \] Thus, the number of monogenizations of \(O\) is equal to the number of integer solutions of (*), where solutions \(\textbf{x}\) and \(-\textbf{x}\) are identified. \textit{J. H. Evertse} and \textit{K. Györy} [J. Reine Angew. Math. 358, 6--19 (1985; Zbl 0552.10010)] obtained a uniform upper bound \(c(n)\) in terms of only \(n=[K:\mathbb{Q}]\) for the number of monogenizations of \(O\) or solutions of (*). \textit{J.-H. Evertse} [Publ. Math. Debr. 79, No. 3--4, 411--422 (2011; Zbl 1249.11102)] improved this to \(2^{4(n+4)(n-2)}\). For \(n=3\), eq. (*) is a cubic Thue equation, which by a result of \textit{M. A. Bennett} [Trans. Am. Math. Soc. 353, No. 4, 1507--1534 (2001; Zbl 0972.11014)] has at most \(10\) solutions. So \(O\) has at most \(10\) monogenizations. Recently, \textit{M. Bhargava} [Publ. Math. Debr. 100, No. 3--4, 513--531 (2022; Zbl 1499.11328)], proved that an order \(O\) of a number field of degree \(4\) has at most \(2760\) monogenizations, and at most \(182\) solutions if the absolute value of the discriminant of \(O\) is sufficiently large. Bhargava proved this by making a reduction to cubic, as well as quartic Thue equations and using upper bounds for the number of solutions of both. For quartic Thue equations he used a recent result of Akhtari, proved in an appendix of Bhargava's paper. Bhargava's reduction is more algebraic and uses different parametrizations of quartic rings. In the paper under review, Akhtari gives another proof of Bhargava's result. Her approach also reduces the problem of estimating the number of monogenizations of a quartic order \(O\) to estimating the number of solutions of certain cubic and quartic Thue equations, but instead of Bhargava's algebraic approach, she uses the more elementary approach of \textit{I. Gaál} et al. [J. Number Theory 57, No. 1, 90--104 (1996; Zbl 0853.11023)]. In fact, following Gaál et. al. she shows that there are ternary quadratic forms \(Q_1,Q_2\in \mathbb{Z}[x_1,x_2,x_3]\) and a cubic binary form \(F\in\mathbb{Z}[U,V]\) such that if \((x_2,x_3,x_4)\in\mathbb{Z}^3\) then \(|F(u,v)|= 1\), where \(u=Q_1(x_2,x_3,x_4)\), \(v=Q_2(x_2,x_3,x_4)\). Further, she shows that if \(x_2,x_3,x_4\) satisfy the lattter equations for some \(u,v\), then \(x=x_i(p,q)\) for \(i=1,2,3\), \(G_{u,v}(p,q)=1\) for certain integers \(p,q\), where the \(x_i\) are binary quadratic forms and \(G_{u,v}\) is a quartic binary form with integer coefficients, depending on \(u,v\). Using existing uniform upper bounds for the number of solutions of cubic, respectively quartic Thue equations one arrives at an upper bound for the number of solutions of (*).