Asymptotic behavior of small solutions of quadratic congruences in three variables modulo prime powers (Q2164885)

In this well-written and very well-thought paper, the authors give an asymptotic for the quantity \[ \sum_{\substack{(x_1,x_2,x_3)\in\mathbb{Z}^3\\ (x_1x_2x_3,p)=1\\ \alpha_1x_1^2+\alpha_2x_2^2+\alpha_3x_3^2\equiv0[p^n]}} \Phi\biggl(\frac{x_1}{N}\biggr) \Phi\biggl(\frac{x_2}{N}\biggr) \Phi\biggl(\frac{x_3}{N}\biggr) \] where \(p\) is some fixed prime number, \(\Phi\) is a function of Schwartz class, \(n\) goes to infinity and either \(\alpha_1\), \(\alpha_2\) and \(\alpha_3\) are fixed and \(N>p^{n/2+\varepsilon}\), or \(\alpha_1\), \(\alpha_2\) and \(\alpha_3\) are allowed to vary with \(N\) and \(N>p^{11n/18+\varepsilon}\), improving on earlier contributions. The proof combines rational parametrization of the curve \(\alpha_1x_1^2+\alpha_2x_2^2=-\alpha_3\) in \(\mathbb{Q}_p\), exponential sum results by \textit{T. Cochrane} and \textit{Z. Zheng} [Acta Arith. 95, No. 1, 67--95 (2000; Zbl 0956.11018)] and Poisson summation formula in the two variables \(x_1\) and \(x_2\), then on the variable \(x_3\). The three pages long first section contains a nice introduction to the problem, its history and related questions.