On the rate of quantum ergodicity for quantised maps (Q1002373)

Let \(e(x):= e^{2\pi ix}\), \(e_N(x):= e(x/N)\), \(\Phi: T^2\to T^2\) (torus), \(\overline a:= \int_{T^2} a(x)\,dx\), and \[ C[a](t):= \int_{T^2} a^*(x) a(\Phi^t(x))\,dx- |\overline a|^2,\;t\in\mathbb{Z}. \] The map \(\Phi\) is ergodic if for \(\forall a\in L^2(T^2)\) one has \[ \lim_{T\to\infty} T^{-1} \sum_{t=0\sim T-1} C[a](t)= 0, \] and it is weak mixing if, for every \(\theta\in R\) and \(a\in L^2(T^2)\), \[ \lim_{T\to\infty} T^{-1} \sum_{t=0\sim T-1} e(\theta t)C[a](t)= 0. \] \(\{U_N: H_N\to H_N\}_{N\in\mathbb{N}}\) is a quantisation of \(\Phi\) if for \(\forall a\in C^\infty(T^2)\) we have \(\lim_{N\to\infty}\| U_N* 0p_N[a]U_N- 0p_N[a\cdot\phi]\| 0\). The author studies the variance of transition amplitude for quantised maps. Theorem 1. Let \(\psi^N_j\), \(j= 1,\dots, N\) be an orthonormal basis of eigenfunctions with eigenvalues \(e(\theta^N_j)\) of \(U_N\). \(\Phi\) is ergodic \(\rightleftarrows\) \[ \lim_{\delta\to 0}\limsup_{N\to\infty} N^{-1} \sum_{I(i,j,0; N)\leq\delta}|\langle\psi^N_1, 0p_N[a]\psi^N_j\rangle- \delta_{ij}\overline a|^2= 0 \] for \(\forall a\in C^\infty(T^2)\). Here \(I(i,j,\theta; N)= \min_{k\in\mathbb{Z}}|\theta^N_1- \theta^N_j- \theta+|k|\). The equivalent condition for weak mixing is given by using \(I(i,j,\theta;N)\leq\delta\) (for \(\forall\theta\in\mathbb{R}\)) instead of \(I(i,j,0;N)\leq\delta\). Next let \(\Psi_\alpha: T^1\to T^2\) be \({^t(p, q)\to^t(p+\alpha, q+ 2p)}\), \(\alpha\): irrational \[ \| U_N^{-1} 0p_N[a]U_N^t- 0p_N[a\cdot\psi^t_\alpha]\|\leq C_\alpha|t|/N, \] holds for \(t\in\mathbb{Z}\). Theorem 2. Suppose that there exist \(C\), \(\gamma> 0\) such that \(|k\alpha-\ell|\geq C/|k|^\gamma\) holds for all \(k,\ell\in\mathbb{Z}-\{0\}\). Then we have \[ N^{-1}\sum_{j= 1\sim N}|\langle\psi^N_j, 0p_N[a]\psi^N_j\rangle-\overline a|^2\leq C_\alpha/N^{1/2}. \] The similar estimations for Hyperbolic maps are also given.