Equations in three singular moduli: the equal exponent case (Q2109407)

Let \(\mathbb{H}\) be the complex upper half plane. Let \(j\) be the modular \(j\)-function. If \(\tau\in \mathbb{H}\) generates a quadratic field over \(\mathbb{Q}\), we call the complex number \(j(\tau)\) a singular modulus. For an integer \(n\geq0\), a CM point in \(\mathbb{C}^n\) is a point with all coordinates given by singular modulus. A special subvariety in \(\mathbb{C}^n\) is an irreducible component of some subvariety consisting of points which have coordinates given by singular modulus and satisfy modular polynomial. For \(V\subset \mathbb{C}^n\) a subvariety, the André-Oort conjecture proven by Pila asserts that \(V\) contains only finitely many maximal special subvarieties. In particular, \(V\) contains only finitely many CM-points which do not lie on positive dimension special subvariety of \(V\). However, Pila's method is not effective. For \(n=2\), there are some effective version of the André-Oort conjecture made by Kuhne, Bilu, Masser and Zannier. The aim of the article is to classify explicitly all the CM-points in two families of algebraic surface in \(\mathbb{C}^3\). For a positive integer \(a\in\mathbb{Z}\), and let \(\varepsilon_i\in\{\pm1\}(i=1,2,3)\). The author classifies the singular modulus \(x_1,x_2,x_3\) satisfying \(\varepsilon_1x^a_1+\varepsilon_2x^a_2+\varepsilon_3x^a_3\in\mathbb{Q}\) (resp. \(x^{\varepsilon_1}_1x^{\varepsilon_2}_2x^{\varepsilon_3}_3\in\mathbb{Q}^\times\)). There are some additional condition for the second case. As a corollary, the author deduces a ``Fermat's last theorem'' for singular moduli from the first case, i.e., if \(x_i(i=1,2,3)\) are singular moduli, and satisfy either \(x^a_1+x^a_2+x^a_3=0\) or \(x^a_1+x^a_2=x^a_3\), then \(x_1x_2x_3=0\). Except for a computational style in the proof, the main ingredient is a primitive element theorem for singular moduli which generalises a result of \textit{B. Faye} and \textit{A. Riffaut} [J. Number Theory 192, 37--46 (2018; Zbl 1444.11135)].