Non-Hermitian Disordered Systems

✨

This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to understand how a city works. In the old days, physicists thought of the city as a perfectly balanced, closed system. If you pushed a car, it would roll forever unless friction stopped it. This is the world of Hermitian physics: systems that are closed, energy-conserving, and predictable. In this world, we have a famous map called the "10-fold way" that helps us categorize every possible type of city based on its symmetries (like whether it has a mirror image or if time runs backward).

But real life isn't a closed city. It's messy. Cars lose fuel, people get tired, and weather changes. This is the world of Non-Hermitian physics: systems that are open, losing energy (dissipation), or gaining it from outside.

Now, add disorder to the mix. Imagine the city has potholes, random traffic jams, and unpredictable construction. This is a Non-Hermitian Disordered System.

This paper is a grand tour of this messy, open, chaotic world. Here is the breakdown in simple terms:

1. The New Map: From 10 to 38

In the old "closed city" (Hermitian), we had 10 types of symmetry rules to categorize how things behave.
In the new "open city" (Non-Hermitian), things get complicated. Because energy can flow in and out, and because the rules for "time running backward" split into two different versions (one for the system itself, one for its mirror image), the number of categories explodes.

The Analogy: Imagine you have 10 colors of paint. In the old world, you could only mix them to make 10 specific shades. In the new world, you have a magical prism that splits every color into three distinct versions, and then you can mix them in new ways. Suddenly, you have 38 distinct categories. The authors built this new "38-fold map" to help us navigate this complex landscape.

2. The Dice Game: Random Matrix Theory

To understand these messy systems, physicists use a tool called Random Matrix Theory. Think of it as rolling a giant, complex die to simulate a system.

The Old Way (Hermitian): When you roll the die, the numbers (eigenvalues) land on a straight line (the real number line). It's like rolling a die on a flat table.
The New Way (Non-Hermitian): When you roll the die, the numbers don't just land on a line; they scatter all over a 2D plane (like a dartboard).
The Discovery: The authors explain how to measure the "distance" between these scattered darts. In the old world, we just looked at how close numbers were on the line. In the new world, we have to look at the whole 2D area. They found that even though the darts are scattered, they still follow strict rules based on the 38 categories. Some categories push the darts apart (level repulsion), while others let them clump together.

3. The Chaos Detector: Is the System "Chaotic" or "Ordered"?

In physics, we love to know if a system is Chaos (unpredictable, like a storm) or Integrable (orderly, like a clock).

The Old Rule: In closed systems, if the system is chaotic, the "darts" on our line follow a specific pattern (like the Ginibre ensemble). If it's orderly, they follow a random pattern (Poisson).
The New Rule: In open systems (like a quantum computer losing information to the environment), the authors show that we can still use these "dart patterns" to detect chaos. They tested this on models of quantum spins (tiny magnets) and found that even when the system is losing energy, the "darts" still reveal if the underlying physics is chaotic or orderly.
The Catch: Sometimes, the old rules break down. A system might look chaotic in its "dart pattern" but behave orderly in its actual motion. The paper warns us that we need to be careful and develop new ways to define chaos in these open worlds.

4. The Magic Carpet: Anderson Transitions

One of the most famous phenomena in physics is Anderson Localization. Imagine a crowd of people trying to walk through a city.

In a normal city (Hermitian): If the city is full of random obstacles (disorder), the people eventually get stuck. They can't move past a certain point. This is "localization." In 1D (a straight line), even a tiny bit of disorder stops everyone.
In the Non-Hermitian city: The authors discuss the Hatano-Nelson model. Here, the "wind" blows in one direction (non-reciprocity).
The Surprise: Even with a huge amount of disorder, the people don't get stuck! The "wind" (non-Hermiticity) creates a topological effect that keeps them moving. It's like a magic carpet that refuses to let the passengers fall off, even when the ground is full of holes. This breaks the old "rules of localization" and creates a new kind of transition where things go from "stuck" to "free" in ways we never saw before.

5. The Blueprint: Nonlinear Sigma Models

Finally, the paper talks about Nonlinear Sigma Models.

The Analogy: If the "darts" and "cities" are the specific examples, the Sigma Model is the architect's blueprint. It's a mathematical formula that describes the shape of the space where these systems live.
The authors show how to rewrite these blueprints for the new 38 categories. They found that the "shape" of the space changes depending on whether the system is Hermitian or not, which explains why the new rules (like the magic carpet effect) happen.

Summary

This paper is a guidebook for a new frontier in physics. It tells us that when we open a system up to the world (adding dissipation) and make it messy (adding disorder), the old rules don't just get tweaked—they get rewritten.

We went from 10 categories to 38.
We went from lines to planes.
We discovered that disorder doesn't always stop movement; sometimes, the "openness" of the system creates a new kind of freedom.

It's a reminder that in the real, messy, open world, the laws of physics are far more colorful, complex, and surprising than they are in the perfect, closed labs of the past.

1. Problem Statement

The paper addresses the theoretical framework required to understand non-Hermitian disordered systems, a regime where two fundamental ingredients of modern physics intersect:

Non-Hermiticity: Arising from open quantum systems, dissipation, gain/loss, and non-reciprocity, leading to complex eigenvalues.
Disorder: Arising from impurities and inhomogeneities, traditionally studied via Anderson localization.

Key Challenges Identified:

Spectral Complexity: Unlike Hermitian systems where eigenvalues are real, non-Hermitian eigenvalues populate the 2D complex plane. Standard tools for real spectra (e.g., nearest-neighbor level spacing) fail or require redefinition.
Symmetry Classification: The celebrated "10-fold way" (Altland-Zirnbauer classification) for Hermitian systems is insufficient. Non-Hermiticity splits symmetries (e.g., Time-Reversal Symmetry vs. its conjugate counterpart), necessitating a broader classification.
Universality and Chaos: It is unclear how disorder-induced criticality (Anderson transitions) and quantum chaos manifest in open systems where unitarity is broken.
Effective Field Theory: Existing nonlinear sigma models (NLSM) for Hermitian systems must be generalized to handle complex spectra and the resulting topological terms.

2. Methodology

The authors employ a multi-faceted theoretical approach combining random matrix theory (RMT), symmetry analysis, and effective field theory:

Symmetry Classification: They systematically analyze antiunitary and unitary symmetries acting on non-Hermitian operators ( $H$ ) and their Hermitian conjugates ( $H^\dagger$ ). This involves distinguishing between symmetries like Time-Reversal Symmetry (TRS) and TRS $^\dagger$ , and Particle-Hole Symmetry (PHS) and PHS $^\dagger$ .
Random Matrix Theory (RMT):
- Utilization of Ginibre ensembles (complex Gaussian matrices) as the baseline for generic non-Hermitian spectra.
- Development of new diagnostics for complex spectra, such as complex level-spacing ratios ( $z_n$ ) and dissipative spectral form factors, to replace standard 1D statistics.
Effective Field Theory: Derivation of Nonlinear Sigma Models (NLSM) using the replica method. Crucially, they adapt the replica formalism for complex spectra by "Hermitizing" the system (doubling the Hilbert space) to define a valid partition function, revealing new topological terms.
Numerical Simulations: Analysis of specific models (e.g., Hatano-Nelson, Lindblad-driven Ising models) to extract critical exponents and spectral statistics, comparing them against theoretical predictions.

3. Key Contributions

A. The 38-Fold Symmetry Classification

The paper establishes that non-Hermitian systems are governed by a 38-fold symmetry classification, a significant expansion of the Hermitian 10-fold way.

Mechanism: In Hermitian systems, TRS and PHS are unique. In non-Hermitian systems, constraints on $H$ and $H^\dagger$ are distinct, creating new symmetry classes (e.g., TRS vs. TRS $^\dagger$ ).
Structure: The classification includes:
- 10 classes based on TRS, PHS, and Chiral Symmetry (CS).
- 10 classes based on TRS $^\dagger$ , PHS $^\dagger$ , and CS.
- Additional classes arising from Sublattice Symmetry (SLS) and its interplay with the above.
Physical Implication: Different symmetries dictate distinct spectral statistics in the bulk, on the real/imaginary axes, and at the origin.

B. Diagnostics of Complex Spectral Statistics

The authors define robust tools to characterize chaos and integrability in open systems:

Level-Spacing Ratios ( $z_n$ ): Defined as the ratio of nearest-neighbor to next-to-nearest-neighbor eigenvalue differences in the complex plane. This avoids the need for spectral unfolding.
Dissipative Spectral Form Factor: A complex-time generalization of the standard form factor, capturing spectral rigidity and the "dip-ramp-plateau" structure characteristic of chaos.
Eigenvector Overlaps: Quantifying the non-orthogonality of left and right eigenstates, which is a unique feature of non-Hermitian systems (related to the Petermann factor).

C. Non-Hermitian Random Matrix Universality

Ginibre Ensembles: Confirmed that generic non-Hermitian matrices follow the circular law (uniform density in a disk).
Level Repulsion: While Hermitian TRS leads to linear ( $s^1$ ), quadratic ( $s^2$ ), or quartic ( $s^4$ ) level repulsion, non-Hermitian TRS $^\dagger$ leads to cubic level repulsion ( $P(s) \propto s^3$ ) in the bulk, regardless of the specific symmetry class (with logarithmic corrections for some classes).
Symmetry-Specific Behavior:
- Bulk: Governed by TRS $^\dagger$ .
- Real/Imaginary Axes: Governed by TRS/PHS $^\dagger$ and Pseudo-Hermiticity, leading to suppressed spectral density and subextensive numbers of real eigenvalues.
- Origin: Governed by PHS and SLS, affecting the density of states near zero energy.

D. Disorder-Induced Criticality and Anderson Transitions

Hatano-Nelson Model: The paper revisits this 1D non-reciprocal model, showing that disorder induces a transition where delocalized states (complex eigenvalues) coexist with localized states (real eigenvalues).
Breakdown of One-Parameter Scaling: Unlike Hermitian 1D systems (where any disorder localizes states), non-Hermitian systems can support delocalized states due to point-gap topology (winding numbers).
Two-Parameter Scaling: The transition is described by a renormalization group flow involving both the coupling constant ( $g$ ) and a topological term ( $W$ ), analogous to the Quantum Hall transition.
Critical Exponents: Numerical results show that non-Hermitian Anderson transitions have different critical exponents ( $\nu$ ) compared to their Hermitian counterparts, even within the same symmetry class (e.g., Class AII vs. AII $^\dagger$ ).

E. Effective Field Theory (NLSM)

The authors derive NLSMs for non-Hermitian disordered systems.

Hermitization: They show that the replica partition function for non-Hermitian $H$ is equivalent to that of a Hermitized system $\tilde{H}$ with an additional chiral symmetry.
Target Manifolds: This leads to different target manifolds (e.g., $(G \times G)/G$ ) compared to Hermitian cases, explaining the distinct universality classes.
Topological Terms: The NLSM action includes a topological term (winding number) that protects delocalization in 1D, explaining the breakdown of the standard scaling hypothesis.

4. Results

Universality Classes: The 38-fold classification successfully organizes all possible non-Hermitian disordered systems, predicting distinct spectral statistics for each class.
Chaos in Open Systems: The paper validates the Grobe-Haake-Sommers conjecture (that chaotic open systems follow Ginibre statistics) for many-body Lindblad systems, though it notes recent exceptions where semiclassical chaos and random matrix statistics may decouple.
Critical Exponents: Table 7 presents numerical values for critical exponents in 2D and 3D, demonstrating that non-Hermiticity fundamentally alters the universality of Anderson transitions.
Skin Effect Connection: The non-trivial point-gap topology identified in the NLSM is linked to the non-Hermitian skin effect, where eigenstates accumulate at boundaries.

5. Significance

This review is a foundational text for the emerging field of non-Hermitian physics. Its significance lies in:

Unification: It provides a unified framework connecting random matrix theory, symmetry classification, and effective field theory for open and disordered systems.
Experimental Relevance: The theoretical predictions (e.g., complex spectral statistics, skin effects, modified Anderson transitions) are directly testable in photonic crystals, electrical circuits, cold atoms, and superconducting qubits.
Beyond Physics: The principles extend to non-physical systems like ecological networks, neural networks, and population dynamics, where dissipation and stochasticity are inherent.
Theoretical Advancement: It resolves long-standing questions about how disorder affects non-unitary dynamics and establishes that "universality" in the non-Hermitian regime is richer and more complex than previously thought, requiring a 38-fold rather than 10-fold classification.

In summary, Kawabata and Ryu demonstrate that the interplay of non-Hermiticity and disorder creates a new landscape of critical phenomena and chaotic dynamics, governed by a rigorous 38-fold symmetry structure and distinct from the Hermitian paradigm.