Table of Contents
Fetching ...

Arithmetic spectral transition for the unitary almost Mathieu operator

Fan Yang

TL;DR

This work analyzes the unitary almost Mathieu operator (UAMO) in the positive Lyapunov exponent regime, establishing a sharp arithmetic localization threshold: Anderson localization holds for irrational frequencies with $\beta(\omega)<L$ and non-resonant phases $\tilde{\gamma}(\omega,\theta)=0$. By embedding the model into extended CMV matrices and exploiting Szegő cocycles, the authors perform a scale-by-scale Green's function analysis along continued-fraction scales $q_n$, balancing Lyapunov growth against small divisors. They distinguish strong and weak Liouville scales and treat non-resonant and resonant regimes accordingly, proving exponential decay of generalized eigenfunctions and hence pure point spectrum with exponentially decaying eigenfunctions. The results extend prior Diophantine-frequency localization to a sharp arithmetic threshold, with potential implications for sharp spectral transitions in unitary quasi-periodic systems and quantum walks in magnetic fields.

Abstract

We study the unitary almost Mathieu operator (UAMO), a one-dimensional quasi-periodic unitary operator arising from a two-dimensional discrete-time quantum walk on $\mathbb Z^2$ in a homogeneous magnetic field. In the positive Lyapunov exponent regime $0\le λ_1<λ_2\le 1$, we establish an arithmetic localization statement governed by the frequency exponent $β(ω)$. More precisely, for every irrational $ω$ with $β(ω)<L$, where $L>0$ denotes the Lyapunov exponent, and every non-resonant phase $θ$, we prove Anderson localization, i.e. pure point spectrum with exponentially decaying eigenfunctions. This extends our previous arithmetic localization result for Diophantine frequencies (for which $β(ω)=0$) to a sharp threshold in frequency.

Arithmetic spectral transition for the unitary almost Mathieu operator

TL;DR

This work analyzes the unitary almost Mathieu operator (UAMO) in the positive Lyapunov exponent regime, establishing a sharp arithmetic localization threshold: Anderson localization holds for irrational frequencies with and non-resonant phases . By embedding the model into extended CMV matrices and exploiting Szegő cocycles, the authors perform a scale-by-scale Green's function analysis along continued-fraction scales , balancing Lyapunov growth against small divisors. They distinguish strong and weak Liouville scales and treat non-resonant and resonant regimes accordingly, proving exponential decay of generalized eigenfunctions and hence pure point spectrum with exponentially decaying eigenfunctions. The results extend prior Diophantine-frequency localization to a sharp arithmetic threshold, with potential implications for sharp spectral transitions in unitary quasi-periodic systems and quantum walks in magnetic fields.

Abstract

We study the unitary almost Mathieu operator (UAMO), a one-dimensional quasi-periodic unitary operator arising from a two-dimensional discrete-time quantum walk on in a homogeneous magnetic field. In the positive Lyapunov exponent regime , we establish an arithmetic localization statement governed by the frequency exponent . More precisely, for every irrational with , where denotes the Lyapunov exponent, and every non-resonant phase , we prove Anderson localization, i.e. pure point spectrum with exponentially decaying eigenfunctions. This extends our previous arithmetic localization result for Diophantine frequencies (for which ) to a sharp threshold in frequency.
Paper Structure (26 sections, 32 theorems, 187 equations)

This paper contains 26 sections, 32 theorems, 187 equations.

Key Result

Theorem 1.1

Assume eq:supercritical_regime_intro. Let $\omega\in\mathbb R\setminus\mathbb Q$ satisfy $\beta(\omega)<L$, and let $\theta\in\mathbb T$ be non-resonant, i.e. $\tilde{\gamma}(\omega,\theta)=0$. Then the unitary almost Mathieu operator $W_{\lambda_1,\lambda_2,\omega,\theta}$ has Anderson localization

Theorems & Definitions (53)

  • Theorem 1.1: Arithmetic Anderson localization
  • Remark 1.2
  • Theorem 2.1: Corollary 2.8 of CFO and (3.25) of FanUAMO
  • Lemma 2.2
  • Corollary 2.3
  • Lemma 3.1: Lagrange interpolation
  • Lemma 3.2
  • Lemma 3.3
  • Lemma 3.4
  • Theorem 5.1
  • ...and 43 more