Small gaps in the spectrum of tori: asymptotic formulae (Q6046977)

Given \(\alpha=(\alpha_1,\alpha_2,\alpha_3)\in\mathbb{R}^3\) with \(4\alpha_1\alpha_3>\alpha_2^2\) and \(\alpha_1>0\), we denote the corresponding positive binary quadratic form by \(q_{{\alpha}}(m,n)=\alpha_1m^2+\alpha_2mn+\alpha_3n^2\), where \(m,n\in\mathbb{R}\). And without loss of generality if we assume that \(q_{{\alpha}}\) is reduced, then we can restrict \({\alpha}\subseteq\mathcal{D}\) with \[\mathcal{D}=\{(\alpha_1,\alpha_2,\alpha_3)\in\mathbb{R}^3 \,|\ 0\leq\alpha_1\leq\alpha_2\leq\alpha_3 \}.\] We denote by \(0<\Lambda_1\leq\Lambda_2\leq\cdot\cdot\cdot\) the ordered set of values \[\frac{q_{{\alpha}}(m,n)}{D({\alpha})},\] where \(m>0\) or \(m=0\) and \(n>0\), \(D({\alpha})=\frac{1}{\pi}\sqrt{4\alpha_1\alpha_3-\alpha_2^2}\). For an interval \(I\subseteq\mathbb{R}\) and \(N\geq1\), we write \[P({\alpha},N,I)=\frac1N\#\{(j,k)\ |\ \Lambda_j, \Lambda_k\leq N, j\neq k, \Lambda_j-\Lambda_k\in I\},\] and denote by \(\mu(I)\) the length of \(I\). The main results of the paper can be summarized as follows. \begin{itemize} \item[1.] Let \(\eta>0\) be given. There exists a subset \(\mathcal{E}\subseteq\mathcal{D}\) of measure zero with the following property: for all \({\alpha}\in\mathcal{D}\setminus\mathcal{E}\) we have \[P({\alpha},N,[0,\triangle])=(1+o(1))\triangle,\quad N\rightarrow\infty,\] uniformly in \[N^{-1+\eta}\leq\bigtriangleup\leq N^{-\eta}.\] \item[2.] Let \(\eta>0\) be given. There exists a subset \(\mathcal{F}\subseteq\mathcal{D}\) of measure zero with the following property: for all \({\alpha}\in\mathcal{D}\setminus\mathcal{F}\) we have \[P({\alpha},N,[0,\triangle])=(1+o(1))\triangle,\quad N\rightarrow\infty,\] uniformly in \[N^{-1+\eta}\leq\bigtriangleup\leq1.\] \end{itemize}