Integral points close to a space curve (Q2314814): Difference between revisions

The paper begins with a brief account of integral points on a curve, centred on the work of \textit{E. Bombieri} and \textit{J. Pila} [Duke Math. J. 59, No. 2, 337--357 (1989; Zbl 0718.11048)]. Amongst other results, they proved that if \(\Gamma: y=f(x)\) is a strictly convex arc, the number of integral points on the dilation \(q\Gamma\) by a factor \(q\ge 1\) satisfies \(\#(q\Gamma\cap\mathbb{Z}^2)\ll_{\Gamma,\varepsilon}q^{\frac{1}{2}+\varepsilon}\) for \(f\in C^\infty[0,1]\). After making some remarks on the problem of counting integral points near the dilation of a proper submanifold in \(\mathbb{R}^n\), the paper focuses on space curves in \(\mathbb{R}^3\). Let \(\mathcal{C}\) be a compact \(C^3\) curve in \(\mathbb{R}^3\) with nonvanishing torsion. After dilation and translation, the curve can be parametrised by \(\{(x,f_1(x),f_2(x)):x\in[0,1]\}\). Let \[A(q,\delta)=\#\{a\in[0,q]\cap\mathbb{Z}:||qf_1(a/q)||<\delta, ||qf_2(a/q)||<\delta\}.\] Then \[A(q,\delta)\ll_C \delta^2q+q^{\frac{3}{5}}(\log q)^{\frac{4}{5}}\quad\text{for }0<\delta<1/2\text{ and }q\ge1. \] The author gives several conjectures on improvements that might be possible. The approach taken is to reduce the problem for space curves to counting irrational points near plane curves. If \(I=[\xi,\eta]\), \(f\in C^2(I)\) satisfies \(0<c_1\le|f''(x)|\le c_2\), \(\mu(j_1,j_2,\lambda)=\{x\in I: ||j_1x+j_2f(x)||<\lambda\}\), \(J\) is a positive integer and \(0<\lambda<1/2\), then \[\sum_{\substack{|j_1|,|j_2|\le J \\ (j_1,j_2)\ne(0,0)}} |\mu(j_1,j_2,\lambda)| \ll \lambda J^2+\lambda^{\frac{1}{2}}J^{\frac{3}{2}}\log J\] which is essentially best possible. The idea then is based on an argument of \textit{V. G. Sprindzhuk} [Metric theory of Diophantine approximations. New York etc.: John Wiley \& Sons (1979; Zbl 0482.10047)], where the space curve is approximated by short line segments.