Counterexamples to the Kotani-Last conjecture for continuum Schrödinger operators via character-automorphic Hardy spaces (Q261202)

Given a probability space \((\Omega, \mathcal{F},\mu)\) with a one-parameter group \(\{T_x\}_{x\in\mathbb R}\) of \(\mu\)-ergodic transformations of \(\Omega\), the Schrödinger operators with ergodic potentials are \[ [L_{q_\omega}\,y](x)=-y''(x)+q_\omega(x)y(x), \qquad q_\omega(x)=f(T_x\omega), \] \(f\) is a bounded, real-valued function on \(\Omega\). This class of potentials contains many important special cases, such as periodic, almost-periodic, and random potentials. It is well known that all periodic potentials lead to the absolutely continuous spectrum of the corresponding operator, whereas the random ones lead to the pure point spectrum. The existence of absolute continuity beyond periodicity can be observed for certain classes of limit-periodic potentials (ones which can be uniformly approximated by periodic potentials) or some quasi-periodic potentials. Both of them belong to the class of almost periodic potentials, where in the above description, \(\Omega\) is a compact abelian group, \(\mu\) is the Haar measure on \(\Omega\), \(\{T_x\}\) is a translation flow, and \(f\) is continuous. A conjecture of Kotani and Last claims that the potentials \(\{q_\omega\}\) are almost periodic as long as the family \(L_{q_\omega}\) has a nonempty absolutely continuous spectrum. The goal of the paper is to disprove this conjecture.