Khinchin's theorem for approximation by integral points on quadratic varieties (Q540426): Difference between revisions

The main result of the paper under review is a Khintchine type theorem for simultaneous Diophantine approximation by rational points in \(\mathbb{R}^d\) with their homogeneous integer coordinates lying on a rational quadric \(X\) in \(\mathbb{R}^{d+1}\) with \(X\) being given by the equation \(Q(x)=m\), where \(Q\) is a nondegenerate indefinite quadratic form of \(d+1\) variables with coefficients in \(\mathbb{Q}\) and \(m\in\mathbb{Q}\setminus\{0\}\). It is assumed that \(d\geq3\) and \(X(\mathbb{Z})=X\cap\mathbb{Z}^{d+1}\neq\emptyset\). Let \(\psi:(0,\infty)\to(0,\infty)\) satisfy the following condition: there exists \(c>0\) such that \[ \psi(ht)\leq c\psi(t)\qquad\text{for all \(h\in[\tfrac12,2]\) and \(t>0\)}. \] Let \(S^d\) denote the unit sphere in \(\mathbb{R}^{d+1}\). A point \(v\in S^d\) is called \((X,\psi)\)-approximable if \[ \|v-\pi(x)\|<\psi(\|x\|) \] has infinitely many solutions \(x\in X(\mathbb{Z})\). The \(\|\cdot\|\) is any fixed norm on \(\mathbb{R}^{d+1}\). When the norm is Euclidean, the left hand side of the above inequality is the sign of the acute angle between the radii of \(v\) and \(x\) and if often referred to as the projective distance, which is natural to use. If \(\psi\) tends to zero at infinity, then any \((X,\psi)\)-approximable point must lie on the so-called boundary of \(X\) defined by \[ \partial X=\{x\in S^d:Q(x)=0\}. \] The set \(\partial X\) supports the unique probability measure \(\mu_\infty\) which is semi-invariant under the actions of the orthogonal group \(G=O(Q)\). The following Khintchine type result is established in the paper: Theorem: Assuming the above notation and constrains and assuming that \(\psi\) is measurable one has the following: \text{(i)} ~if \(\int_1^\infty t^{d-2}\psi(t)^{d-1}\,dt=\infty\), then \(\mu_\infty\)-almost every point of \(\partial X\) is \((X,\psi)\)-approximable; \text{(ii)} if \(\int_1^\infty t^{d-2}\psi(t)^{d-1}\,dt<\infty\), then \(\mu_\infty\)-almost every point of \(\partial X\) is not \((X,\psi)\)-approximable.