Spectral and quantum dynamical properties of the weakly coupled Fibonacci Hamiltonian (Q634625)

(I) The authors treat here the Fibonacci Hamiltonian: \[ [H_{V,\omega}\psi](n)= \psi(n+ 1)+\psi(n-1)+ V\chi_{[1-\alpha,1)}(n\alpha+ \omega\text{\,mod\,}1)\psi(n), \] in \(\ell^2(Z)\), where \(V>0\), \(\alpha= (5^{1/2}- 1)/2\), and \(\omega\in T= \mathbb{R}/\mathbb{Z}\). They consider the spectrum \(\sigma(H_{V,\omega})\) for small values of \(V\). \(\sigma(H_{V,\omega})=\Sigma_V\), \(V> 0\), for every \(\omega\in T\). It was shown that \(\Sigma_V\) is a zero-measure Cantor set, and that the Hausdorff dimension of \(\Sigma_V\) lies between 0 and 1. Theorem 1. There are \(C_1,C_2> 0\) such that \(I-C_1 V\leq\dim\Sigma_V\leq 1- C_2V\), for \(V>0\) sufficientry small. Let \(C\subset\mathbb{R}\) be a Cantor set. I: convex hull of \(C\), and any connected component of \(I\setminus C:= \{U_i\}={\mathcal U}\) is called a gap of \(C\). Let \(u\) be a boundary point of a gap \(U_n\) of \(C\), and \(K\): a connected component of \(I\setminus\bigcup_{1\leq i\leq n}U_i\), which contains \(u\). By using \(\tau(C,{\mathcal U},u)= |K|/|U_n|\), the thickness \(\tau= \sup_{{\mathcal U}}\text{inf}_u\tau(C,{\mathcal U},u)\) and the denseness \(\theta= \text{inf}_{{\mathcal U}}\sup_u\tau(C,{\mathcal U},u)\) of the Cantor set \(C\subset\mathbb{R}\) are defined. Theorem 2. \(\lim_{V\to 0}\tau(\Sigma_V)=\infty\). (II) Next they treat the square Fibonacci Hamiltonian: \[ \begin{multlined} [H^{(2)}_V\psi](m, n)= \psi(m+1, n)+ \psi(m-1, n)+\psi(m,n+1)+ \psi(m, n-1)+\\ V(\chi_{[1-\alpha, 1)}(m\alpha\text{\,mod\,}1)+ \chi_{[1-\alpha,1)}(n\alpha\text{\, mod\,}1))\psi(m,n).\end{multlined} \] Theorem 3. \(\sigma(H^{(2)}_V)= \Sigma_V+ \Sigma_V\), for \(V> 0\) sufficiently small. (III) Denote the restriction of \(H_{V,\omega}\) to finite interval \(\Lambda\subset\mathbb{Z}\) with Dirichlet boundary condition by \(H^\Lambda_{V,\omega}\). \(N(E,\omega, V,\Lambda)\): the number of eigenvalues of \(H^\Lambda_{V,\omega}\), less than or equal to \(E\). \(N(E,V)= \lim_{n\to\infty} N(E,\omega, V,[1,n])/n\), \(\{m\alpha\}\): fractional part of \(m\alpha\). Theorem 4. There is \(V_0> 0\) such that for every \(V\in(0,V_0]\), \[ \{N(E,V): E\in\mathbb{R}\setminus \Sigma_V\}= \{\{m\alpha\}: m\in\mathbb{Z}\}\cup \{1\}, \] holds. For \(m\in\mathbb{Z}\setminus\{0\}\), \(U_m(V)\): the gap \((\ni E)\) of \(\Sigma_V\), where \(N(E, V)= \{m\alpha\}\) holds (Theorem 5): (i) Given any continuous family \(\{U_V\}\) of gaps of \(\Sigma_V\), \(\lim_{V\to 0}|U_V|/|V|\) \((\in(0,\infty))\) exists. (ii) There is a finite \(C^*\) such that \(\lim_{V\to 0}|U_m(V)|/|V|= C_m/|m|\), for a suitable \(C_m\in (0,C^*)\).