Anderson transition at complex energies in one-dimensional parity-time-symmetric disordered systems

Wei Wang, Xulong Wang, Guancong MaarXiv 2024 · DOI · arXiv

TL;DR

Researchers have discovered that in 1D non-Hermitian disordered systems, waves can become localized even when their energies are complex—a phenomenon previously thought impossible. By combining theoretical modeling with an active mechanical lattice experiment, the team demonstrated that these "Complex-Energy Localized Modes" (CELMs) emerge from a fundamental link between spectral topology and non-Bloch parity-time ($\mathcal{PT}$) symmetry.

What Problem Does This Paper Address?

In the classical understanding of physics (Hermitian systems), the "scaling theory of localization" suggests that in one dimension, any amount of disorder will eventually cause all wave modes to become localized—a phenomenon known as Anderson localization. However, non-Hermitian systems—where energy can be exchanged with the environment through gain, loss, or non-reciprocal "chiral" hopping—break these rules.

In the late 90s, the Hatano-Nelson model showed that 1D non-Hermitian rings could support both localized and extended (delocalized) modes. For decades, the consensus was that localized modes always possessed real energies, while complex energies were the exclusive domain of extended modes. This paper challenges that dichotomy, asking: Can complex-energy modes also undergo an Anderson transition and become localized?

Methods

The authors employed a two-pronged approach:

  1. Theoretical Modeling: They designed a 1D non-Hermitian Hamiltonian (Class AI) featuring nearest-neighbor and second-neighbor chiral hopping. They analyzed the system using the Biorthogonal Inverse Participation Ratio (BIPR), a mathematical tool used to distinguish between localized and extended states in non-Hermitian frameworks.
  2. Experimental Validation: To move beyond theory, they built an active mechanical lattice consisting of 60 rotational oscillators. By using motors to control the coupling between oscillators, they could precisely simulate non-reciprocal hopping and introduce controlled "disorder" (random onsite potential). They then measured the system's response to single-site excitations to map out the wave behavior.

Datasets

As this is a physics research paper involving original experimental data and numerical simulations, it does not use standard machine learning datasets. The "data" consists of:

  • Numerical spectra and wavefunctions generated for a 400-site lattice model.
  • Experimental response maps from a 60-site physical mechanical lattice across 25 independent configurations of random disorder.

Key Insights and Ideas

  • Breaking the Real-Energy Rule: The most significant insight is that the association between real energy and localization is not absolute. By increasing the strength of second-neighbor chiral hopping ($|\delta_2|$), the system enters a regime where localized modes with complex energies (CELMs) emerge.
  • The Role of the Pristine System: The birth of CELMs isn't random; it is pre-determined by the "pristine" (non-disordered) version of the system. Specifically, CELMs appear when the pristine system under Open Boundary Conditions (OBC) undergoes a non-Bloch $\mathcal{PT}$ transition.
  • DOS as a Predictor: The Density of States (DOS) of the pristine system under Periodic Boundary Conditions (PBC) acts as a roadmap. Localization tends to emerge first at the energies where the DOS is at its maximum.
  • Triple Coexistence: The authors show that a single lattice can simultaneously host three distinct types of modes: Real-Energy Localized Modes (RELMs), Complex-Energy Localized Modes (CELMs), and extended modes.

Key Results

  • Critical Threshold: The study identified a specific mathematical threshold for the emergence of CELMs: $|\delta_2| > |t/3\sqrt{3}|$. When the non-reciprocity exceeds this value, the system's "skin modes" collapse or transition in a way that allows for complex-energy localization.
  • Experimental Observation: Using the mechanical lattice, the team successfully visualized CELMs. At 10.8 Hz (a complex-energy regime), the wave response was strongly confined to the excitation point—clear evidence of localization. At other frequencies (e.g., 9.2 Hz), the waves propagated freely, confirming the coexistence of different transport behaviors.
  • Statistical Confirmation: The level spacings of the CELMs were found to follow a Poisson distribution, which is a classic "smoking gun" signature of Anderson localization.

Limitations and Future Work

  • Lattice Size Constraints: The experimental lattice (60 sites) is significantly smaller than the theoretical models (400 sites). This required the use of larger disorder strengths to clearly observe the localization effects, as predicted by scaling theory.
  • Stability and Gain/Loss: Because CELMs involve complex energies, half of them are technically "temporally divergent gain modes." While this is useful for applications like lasers, it presents challenges for maintaining long-term system stability in certain physical realizations.
  • Future Directions: The authors suggest that CELMs could open new doors for controlling wave transport. Potential applications include high-sensitivity sensors, novel random lasers (utilizing gain modes), and "total absorption" devices (utilizing loss modes). Exploring these transitions in higher dimensions (2D or 3D) remains an open challenge.

Summary updated

Abstract

The presence of disorder can severely impede wave transport, resulting in the famous Anderson localization. Previous theoretical studies found that Anderson transition can exist in one-dimensional (1D) non-Hermitian disordered rings with chiral hopping, defying the scaling theory of localization for Hermitian systems. In these systems, localized (extended) modes are associated with real (complex) energies. Here, we report that Anderson localized modes with complex energies can also exist in such systems. The emergence of the complex-energy localized modes (CELMs) directly ties to the properties of the corresponding pristine non-Hermitian system. Specifically, the density of states of the complex spectrum under the periodic boundary condition and the non-Bloch parity-time transition of the open-boundary chain both play critical roles in the emergence of the CELMs. The coexistence of extended modes, real-energy localized modes (RELMs), and CELMs should be a generic phenomenon for 1D non-Hermitian disordered systems under class AI. Our work shows that the interplay between Anderson mechanism and non-Hermitian physics enriches the properties of disordered media and opens new possibilities for controlling wave transport.
BibTeX
@misc{wang2024anderson,
  author = {Wei Wang and Xulong Wang and Guancong Ma},
  title = {{Anderson transition at complex energies in one-dimensional parity-time-symmetric disordered systems}},
  year = {2024},
  eprint = {2412.06232},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2412.06232},
}
Figure