Consider the discrete magnetic Laplacian on $\mathbb Z^2.$

$$(\Delta_{\alpha,\lambda}\psi)(n_1,n_2) = e^{-i \pi \alpha n_2} \psi(n_1+1,n_2) + e^{i\pi \alpha n_2} \psi(n_1-1,n_2) + \lambda \left(e^{i \pi \alpha n_1} \psi(n_1,n_2+1)+e^{-i\pi \alpha n_1} \psi(n_1,n_2-1) \right)$$

We consider $\alpha$ irrational. It is known (to me) that the spectrum of this operator as a set always coincides with the spectrum of the Almost Mathieu operator and that this operator has no point spectrum. But I am curious whether it is known when the spectrum (depending on $\lambda$) of this operator is absolutely continuous/singular continuous. Apparently, it is singular continuous if $\lambda =1$ as the measure of the spectrum is zero in this case.

| cite | improve this question | | | | |

Metal-insulator transition for the almost Mathieu operator (1999) proves that the spectrum is purely absolutely continuous for $\lambda<1$. (They write $\lambda<2$, but their $\lambda$ is different by a factor of two.) The spectrum is singular-continuous if $\lambda>1$ and $\alpha$ is a Liouville number. For a more extensive discussion, see the article by Y. Last (page 102 and following) in Sturm-Liouville theory.

| cite | improve this answer | | | | |
  • 1
    $\begingroup$ These references are about one-dimensional operators and have little to do with the OP's question, which is about a two-dimensional operator. $\endgroup$ – Christian Remling Jan 21 '18 at 18:43
  • $\begingroup$ @ChristianRemling --- correct me if I'm wrong, please, but the papers I referred to address the Almost Mathieu operator on $\mathbb{Z}$ (one-dimensional), which is equivalent to the discrete magnetic Laplacian on $\mathbb{Z}^2$ (two-dimensional). See for example eqs. 1.1 and 1.2 from this paper. This is why the Mathieu eq. is used to study the spectrum of electrons in a two-dimensional lattice in a perpendicular magnetic field (Hofstadter problem). The parameter $\alpha$ in the Mathieu operator is the magnetic flux through a unit cell. $\endgroup$ – Carlo Beenakker Jan 21 '18 at 19:00
  • $\begingroup$ Thanks, I wasn't aware of this, this is relevant. However, my original comment still seems correct because the correspondence between the two problems (apparently) is not very strong: it just concerns the spectrum as a set and the DOS; the papers you quote prove much more refined results on the type of the spectrum of Almost Mathieu, which will not give anything on the 2d problem. $\endgroup$ – Christian Remling Jan 21 '18 at 20:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.