Singular continuous spectrum for singular potentials (Q521986)

The authors study Schrödinger operators \[ {(H_{\alpha,\theta}u)_n=u_{n+1}+u_{n-1}+ \frac{g(\theta+n\alpha)}{f(\theta+n\alpha)} u_n} \] acting on \(\ell^2({\mathbb Z})\), where \(\alpha \in {\mathbb R}\backslash {\mathbb Q}\) is the frequency, \({\mathbb T} = {\mathbb R}/{\mathbb Z}\), \(\theta \in {\mathbb T}\) is the phase, \(f\) is an analytic function and \(g\) is Lipschitz. Let \(p_n, q_n\) be the continued fraction approximants of \(\alpha\in {\mathbb R} \backslash {\mathbb Q}\). Assume that \(g/f\) has \(m\) poles \(\theta_i , i = 1,...,m\). Define the index \(\delta\) as follows: \[ \delta( \alpha,\theta) = \limsup_{n\rightarrow \infty} \dfrac{\sum\limits_{i=1}^m \ln\|q_n(\theta-\theta_i) \|_{{\mathbb R} / {\mathbb Z}}+ \ln q_{n+1}} {q_n}, \] where \(\|x\|_{{\mathbb R} / {\mathbb Z}}=\min\limits_{l\in {\mathbb Z}}|x-l|\). Let \(L(E)\) be the Lyapunov exponent. The main results of the paper under review are the following: {\parindent=0.7cm\begin{itemize}\item[(1)] \( H_{\alpha, \theta} \) has no eigenvalues on \(\{ E:L(E)< \delta(\alpha, \theta)\}\). \item[(2)] If \(L(E)>0\) for a.e. \(E\) (in particular, if \(m>0\)), then \( H_{\alpha, \theta} \) has purely singular continuous spectrum on \(\{ E:L(E)< \delta(\alpha, \theta)\}\). \end{itemize}} These results are sharp for the Maryland model (\(f=\sin 2\pi\theta\), \(g=\lambda \cos 2\pi\theta\)). However, it is not sharp for the almost Mathieu operator (\(f\equiv 1\), \(g=\lambda \cos 2\pi\theta\)) other than for \(\alpha\)-Diophantine \(\theta.\)