Non-Hermitian delocalization in 1D via emergent compactness

Liang-Hong Mo, Zhenyu Xiao, Roderich Moessner, Hongzheng Zhao · 2024

Read Non-Hermitian delocalization in 1D via emergent compactness by Liang-Hong Mo, Zhenyu Xiao, Roderich Moessner, Hongzheng Zhao (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-12490.txt.

Introduction

Liang-Hong Mo, Zhenyu Xiao, Roderich Moessner, and Hongzheng Zhao study waves trapped in a one-dimensional line by disorder, in work from 2024. In a plain line, random bumps trap every wave in place. Adding one-way hops can fight that trap, an effect tied to the skin effect.

The team finds a second route to free the waves. They use disorder in an imaginary part of the wave's energy. Pure random disorder still traps the waves. But a small amount of order in the pattern changes the math: the transfer matrix gains a neat group form. That form makes the trap length grow without bound, so waves can spread. The share of free waves can be tuned by the choice of edge rules.

Catalog record

CREATORLiang-Hong Mo, Zhenyu Xiao, Roderich Moessner, Hongzheng Zhao
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE43 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-12490

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Non-Hermitian delocalization in 1D via emergent compactness Liang-Hong Mo, Zhenyu Xiao, Roderich Moessner, Hongzheng Zhao Abstract Potential disorder in 1D leads to Anderson localization of the entire spectrum. Upon sacrificing Hermiticity by adding non-reciprocal hopping, the non-Hermitian skin effect competes with localization. We find another route for delocalization, which involves imaginary potential disorder. While an entirely random potential generally still leads to localization, imposing minimal spatial structure to the disorder can protect delocalization: it endows the concomitant transfer matrix with an SU ⁡ ( 2 ) \mathrm{SU}(2) structure, whose compactness in turn translates into an infinite localization length. The fraction of delocalized states can be tuned by the choice of boundary conditions. Introduction.— Anderson localization Anderson 1958 ; Evers and Mirlin 2008 is an everlasting subject in condensed matter physics. The presence of quenched disorder significantly suppresses transport, especially in low dimensions. This stimulates decades of studies including, e.g., delocalization phase transitions with or without many-body interaction Nandkishore and Huse 2015 ; Smith et al. 2017 ; Abanin et al. 2019 , and topological phases of matter in disordered systems Li et al. 2009 ; Titum et al. 2016 ; Stützer et al. 2018 . Recently, localization in open systems has attracted considerable attention Hatano and Nelson 1996 ; Feinberg and Zee 1997 ; Feinberg and Zee 1999 ; Regensburger et al. 2013 ; Xu et al. 2016 ; Ding et al. 2016 ; Martinez Alvarez et al. 2018 ; Gong et al. 2018 ; Hamazaki et al. 2019 ; Tzortzakakis et al. 2020 ; Wang and Wang 2020 ; Huang and Shklovskii 2020a ; Borgnia et al. 2020 ; Huang and Shklovskii 2020b ; Kawabata and Ryu 2021 ; Luo et al. 2021a ; Luo et al. 2021b ; Luo et al. 2022 ; Zhu and Gong 2022 ; Li et al. 2022 ; Zhang and Cai 2023 ; Manna and Roy 2023 ; Zhu et al. 2023 ; Thompson et al. 2024 ; Li et al. 2024 ; Xing et al. 2024 ; Halder et al. 2024 . The interplay between non-Hermicity and localization leads to a plethora of novel phenomena, unique in open quantum systems Yao and Wang 2018 ; Yokomizo and Murakami 2019 ; Okuma et al. 2020 ; Ashida et al. 2020 . As shown in Hatano and Nelson’s pioneering work Hatano and Nelson 1996 , in one-dimension (1D) asymmetric hopping can compete with quenched disorder, leading to delocalization, at variance with the scaling theory of localization for potential disorder Evers and Mirlin 2008 . Such delocalization originates from the non-trivial topology Gong et al. 2018 ; Okuma and Sato 2023 , as elucidated by the effective field theory Kawabata and Ryu 2021 ; Chen et al. 2024 . Also, by introducing a deterministic quasi-periodic spatial structure delocalization may also appear Longhi 2019 ; Liu et al. 2021 ; Zhai et al. 2021 ; Chen et al. 2022 , similar to their Hermitian counterpart Aubry and André 1980 ; Liu and Riklund 1987 ; Axel and Peyrière 1989 ; Biddle et al. 2009 ; Wang et al. 2020 . Here, we address the intriguing question of whether 1D delocalized eigenstates can be realized by purely on-site energy gain and loss, without sacrificing reciprocity and spatial randomness. In general, a purely random complex potential localizes the entire spectrum, although prior work demonstrates the possibility of exotic transport dynamics in wave propagation Basiri et al. 2014 ; Yusipov et al. 2017 ; Tzortzakakis et al. 2021 ; Weidemann et al. 2021 ; Li et al. 2024 . We provide an affirmative answer by constructing a concrete model with delocalized eigenstates and analytically tractable mobility edges. The key ingredient involves imposing a minimal structure on a binary potential, which we dub “dipolar disorder”. The corresponding transfer matrix generally belongs to the non-compact \mathrm{SL}(2,\mathbb{C}) group Hall 2013 ; Woit 2017 ; hence, a random product of such matrices leads to a positive Lyapunov exponent, i.e., a finite localization length Evers and Mirlin 2008 . However, the possible existence of an emergent compact \mathrm{SU}(2) structure, which we identify, in the transfer matrix for the real-valued spectrum results in a zero Lyapunov exponent. It, in turn, leads to an infinite localization length, protecting delocalization against disorder. This \mathrm{SU}(2) structure can be revealed by a simple analytical procedure, cf. Fig. 1 (a), allowing the exact determination of a mobility edge, for which the usual numerical approach faces difficulties, especially in the thermodynamic limit. We numerically verify the theoretical prediction by diagonalizing the corresponding lattice model. A natural diagnostic of delocalization is the participation ratio of eigenstates, by which we obtain a localization-delocalization phase diagram at different energies, Fig. 1 (b). The mobility edge precisely matches the analytical prediction. Crucially, the fraction of delocalized eigenstates depends on the boundary conditions, which are tunable by a twisted phase factor, cf. Fig. 1 (c). For complex energy eigenvalues that go beyond this analytical framework, eigenstates are generally localized. However, for a given finite system size, we find strong numerical evidence showing that delocalization can persist as long as the imaginary component of the energy eigenvalue is sufficiently small. The remainder of this account is organized as follows. First, we introduce the model with disordered imaginary potential and the dipolar structure. Then we elaborate on the emergent \mathrm{SU}(2) structure in the transfer matrix. By numerically calculating the participation ratio of each eigenstate, we demonstrate the existence of delocalization and the localization transition. Finally, we summarize and discuss open directions where the emergent compactness in the transfer matrix may play a crucial role. Figure 1: (a) Schematic for the emergent compact structure. The overall gray region denotes the non-compact \mathrm{SL}(2,\mathbb{C}) group, which contains the compact \mathrm{SU}(2) structure (blue). The green region denotes a subset of \mathrm{SL}(2,\mathbb{C}) , inside which a matrix element as well as its conjugation can be simultaneously mapped to \mathrm{SU}(2) matrices, via a similarity transformation PW. The dipolar transfer matrices, \mathcal{M} and \mathcal{N} , belong to this subset and hence a random product of the two leads to zero Lyapunov exponent, i.e., an infinite localization length. (b) Participation ratio (PR) calculated for each eigenstate at different real energies E and disorder strength V . The emergent compactness leads to delocalized eigenstates (blue), which are separated from the localized states (red) by an exact mobility edge (black). This black line is obtained by analyzing the possible existence of a similarity transformation. In numerical simulation, we scan over all possible boundary conditions with a twisted angle K and sample different realizations of the disorder potential. (c) By tuning K , we can control the fraction of real energies (orange) in the entire spectrum and the delocalized eigenstates therein (blue). For periodic and anti-periodic boundary conditions K{=}0,\pi , the fraction of eigenstates is around 10^{-4} , a small but nonzero value. Yet, for other K values, most real energies are delocalized. For numerical simulation, we use V=0.3 , 2L=1024 . We average over 100 random realizations to obtain both (b) and (c). The model.— The Schrödinger equation for 1D tight-binding Hamiltonian H with uniform and reciprocal nearest-neighbor hopping and on-site potential V_{j} reads \displaystyle E\Psi_{j}=\Psi_{j+1}+\Psi_{j-1}+V_{j}\Psi_{j},\quad j\in\mathbb{Z}, (1) where \Psi_{j}\equiv\langle j|\Psi\rangle is the real-space wavefunction on site j

Read the complete text