Shrinking target equidistribution of horocycles in cusps (Q2098206)

The authors consider a shrinking neighborhood of a cusp of the unit tangent bundle of a noncompact hyperbolic surface of finite area. Let the neighborhood shrink into the cusp at a rate of \(T^{-1}\) as \(T\rightarrow \infty\). They define an automorphic function as: \[ \phi_{T, \eta}(z, \theta) :=\phi_{T, \eta}^{(\kappa_{j})}(z, \theta) :=\sum_{\gamma\in \Gamma_{j}\backslash \Gamma} f_{T, \eta}(\sigma_{j}^{-1}\gamma(z, \theta)), \] where \(f_{T, \eta}(z, \theta)\) is a function whose support lies in \([0, 1] \times [T-|\eta|, \infty) \times [0, 2\pi)\) and \(\sigma_{j}\) is a scaling matrix. The authors prove that a closed horocycle whose length \(l\) goes to infinity or even a segment of that horocycle becomes equidistributed on the shrinking neighborhood when normalized by the rate \(T^{-1}\) provided that \(T/{l}\rightarrow 0\) and, for any \(\delta > 0\), the segment remains larger than max \(\{T^{-1/6}, (T/{l})^{1/2}\} (T/{l})^{-\delta}\). Let us mention the following main results: (1) Let \(j\in \{1, \dots, q\}\), \(\delta>0\), \(T>y\), and \(0<Ty<1\). Then, \[ \frac{T}{\beta-\alpha}\int_{\alpha}^{\beta}\phi_{T, 0}^{(\kappa_{j})}(x+iy, 0) dx \rightarrow \frac{1}{\mu(\Gamma\backslash G)} \int_{0}^{2\pi}\int_{0}^{1} h(x, \theta) dx d\theta \] uniformly as \(T\rightarrow \infty\) and \(Ty\rightarrow 0\) for \(\beta-\alpha \geq \text{max} \big\{T^{-1/6}, \big(\frac{Ty}{B_{1}}\big)^{1/2}\big\} \big(\frac{Ty}{B_{1}}\big)^{-\delta}\). (2) Let \(0\leq \alpha < \beta \leq 1\), \(j\in \{1, \dots, q\}\), \(\frac{1}{2}\geq\delta>0\), \(|\eta|\leq \min \big(\frac{B_{1}-B_{0}}{2}, \frac{1}{4}\big)\), \(T-\min \big(\frac{B_{1}-B_{0}}{2}, \frac{1}{4}\big) >y\), and \(0<Ty<\frac{3}{4}\). Then, for \(\eta \neq 0\), \begin{align*} \frac{T}{\beta-\alpha}\int_{\alpha}^{\beta}\phi_{T, \eta}^{(\kappa_{j})}(x+iy, 0) dx= \begin{cases} \langle\phi_{T, \eta}^{(\kappa_{j})}\rangle+E(\alpha, \beta, T, y, B_{1}, s_{1}, s_{1}^{\prime}, \eta) + \text{O}\big(\frac{Ty}{B_{1}}\big)^{\delta/2} & \text{if}\;\; \alpha \neq 0\; \text{or}\; \beta\neq 1 \\ \langle\phi_{T, \eta}^{(\kappa_{j})}\rangle+E(\alpha, \beta, T, y, B_{1}, s_{1}, s_{1}^{\prime}, \eta) + \text{O}\bigg(\big(\frac{Ty}{B_{1}}\big)^{\delta} +\frac{1}{\sqrt{T}}\bigg) & \text{if}\;\; \alpha = 0\; \text{and}\; \beta=1 \end{cases} \end{align*} and, for \(\eta=0\) and any \(0<|\tilde{\eta}|\leq \text{min} \big(\frac{B_{1}-B_{0}}{2}, \frac{1}{4}\big)\), \(\frac{T}{\beta-\alpha}\int_{\alpha}^{\beta}\phi_{T, 0}^{(\kappa_{j})}(x+iy, 0) dx\) \begin{align*} = \begin{cases} \langle\phi_{T, 0}^{(\kappa_{j})}\rangle+E(\alpha, \beta, T, y, B_{1}, s_{1}, s_{1}^{\prime}, \tilde{\eta}) + \text{O}\big(\frac{Ty}{B_{1}}\big)^{\delta/2} & \text{if}\;\; \alpha \neq 0\; \text{or}\; \beta\neq 1 \\ \langle\phi_{T, 0}^{(\kappa_{j})}\rangle+E(\alpha, \beta, T, y, B_{1}, s_{1}, s_{1}^{\prime}, \tilde{\eta}) + \text{O}\bigg(\big(\frac{Ty}{B_{1}}\big)^{\delta} +\frac{1}{\sqrt{T}}\bigg) & \text{if}\;\; \alpha = 0\; \text{and}\; \beta=1 \end{cases} \end{align*} uniformaly as \(T\rightarrow \infty\) and \(Ty\rightarrow 0\) for \(1\geq\beta-\alpha \geq \text{max} \big(T^{4}\big(\frac{Ty}{B_{1}}\big)^{1/2}, T^{-1/6}\big) \big(\frac{Ty}{B_{1}}\big)^{-\delta}\). In the present paper, the authors define and use the following: \begin{align*} \langle\phi_{T, \eta}^{(\kappa_{j})}\rangle &:=\frac{T}{\mu(\Gamma\backslash G)} \int_{0}^{2\pi}\int_{0}^{1}\int_{0}^{\infty} \mathbb{I}_{[T, \infty), \eta}(y) h(x, \theta) \frac{dy dx d\theta}{y^{2}}, \\ \langle\phi_{T, 0}^{(\kappa_{j})}\rangle &:=\frac{1}{\mu(\Gamma\backslash G)} \int_{0}^{2\pi}\int_{0}^{1}h(x, \theta) dx d\theta, \\ E(\alpha, \beta, T, y, B_{1}, s_{1}, s_{1}^{\prime}, .) &:= \text{O}(T^{4}|.|^{-4}) \bigg(\bigg(\frac{Ty}{(\beta-\alpha)^{2} B_{1}}\bigg)^{1/2} \log^{2}\bigg(\frac{(\beta-\alpha) B_{1}}{Ty}\bigg) \\ &\quad + \bigg(\frac{Ty}{(\beta-\alpha)^{2} B_{1}}\bigg)^{1-s_{1}^{\prime}} + \bigg(\frac{Ty}{(\beta-\alpha) B_{1}}\bigg)^{1-s_{1}}\bigg), \\ s_{1} &:= \begin{cases} \langle \frac{1+\sqrt{1-4\lambda_{1}}}{2} & \text{if there exists small eigenvalues} \\ \langle \frac{1}{2} & \text{if there does not exist small eigenvalues,} \end{cases}\\ s_{1}^{\prime} &:= \text{max} \big(s_{1}^{(1)}, \dots, s_{1}^{(q)}\big), \\ s_{1}^{(j)} &:= \begin{cases} \langle \frac{1+\sqrt{1-4\lambda_{1}^{(j)}}}{2} & \text{if}\; \lambda_{1}^{(j)}\; \text{exists} \\ \langle \frac{1}{2} & \text{if}\; \lambda_{1}^{(j)}\; \text{does not exist.} \end{cases} \end{align*} The authors provide proof of these results very rigorously in this paper. To achieve their proofs, they develop many lemmas and several propositions. The authors also define their test and auxiliary functions and develop a useful expansion for both of these via the double coset decomposition. Further, the authors have an effective result for a smaller range of rates of growth of \(T\) and \({l}\). Moreover, they mention a number-theoretic identity involving the Euler totient function follows from their developed technique.