Values of random polynomials at integer points (Q1787202)

One can begin with authors' abstract: ``Using classical results of \textit{C. A. Rogers} [Proc. London Math. Soc. 6, 305--320 (1956; Zbl 0071.27403), Theorem 1] bounding the \(L^2\)-norm of Siegel transforms, we give bounds on the heights of approximate integral solutions of quadratic equations and error terms in the quantitative Oppenheim theorem of \textit{A. Eskin} et al. [Ann. Math. 147, No. 1, 93--141 (1998; Zbl 0906.11035)] for almost every quadratic form. Further applications yield quantitative information on the distribution of values of random polynomials at integral points.'' The authors note that the novelty of results presented in this research `` includes the elementary nature of the methods, using only classical results from the geometry of numbers, in particular, mean and variance estimates of certain counting functions on the space of lattices.'' Let us note that some main results of this research are related to variance estimates for lattice point counting as we integrate over the space of unimodular lattices. It is noted that in [Trans. Amer. Math. Soc. 95, 516--529, (1960; Zbl 0101.27904)], \textit{W. Schmidt} obtained similar counting results for almost all general (not unimodular) lattices. Schmidt's results give better bounds when the geometry of the set is restricted, and results of the present paper hold for more general sets. In addition, the attention is given to estimates of certain integrals on the space of lattices, the values of fixed homogeneous polynomials at the points of a random unimodular lattice, and to the size of solutions to the inequality \(|Q(x)|\varepsilon\), where \(x\in\mathbb Z^n\setminus \{0\}\), for almost every quadratic form \(Q\). In the survey of this paper, the main attention is given to connections between the present and known results.