Lattice point problems and values of quadratic forms (Q1882749): Difference between revisions

Let \(Q(x) := \left< Qx, x \right>\) be the quadratic form stemming from the symmetric matrix \(Q \in \text{ GL} (d,{\mathbb R})\), whose smallest eigenvalue we may assume to be 1. A quadratic form is rational if its associated matrix has only rational entries. If \(Q\) is positive definite, then \[ E = \left\{ x \in {\mathbb R}^d : Q(x) \leq 1 \right\} \] is an ellipsoid. The main theorem in the paper under review states that for \(d \geq 5\) and \(Q\) positive definite, \[ \sup_{ a \in {\mathbb R}^d } \left| {{ \# \left( \left( rE+a \right) \cap {\mathbb Z} \right) - \text{ vol} \left( rE \right) } \over { \text{ vol} \left( rE \right) }} \right| = O \left( r^{ -2 } \right) , \] as \(r \to \infty\), and if, in addition, \(Q\) is irrational, then \[ \sup_{ a \in {\mathbb R}^d } \left| {{ \# \left( \left( rE+a \right) \cap {\mathbb Z} \right) - \text{ vol} \left( rE \right) } \over { \text{ vol} \left( rE \right) }} \right| = o \left( r^{ -2 } \right) . \] The author gives several applications of these estimates, as well as bounds for the errors.