Lyapunov Exponents as Duality-Invariant Signatures of Critical States

This paper establishes a rigorous, duality-invariant definition of critical states based on the simultaneous absence of exponential localization in both real and momentum space (the Liu–Xia condition), transforming it from a phenomenological criterion into an exact solvability principle that predicts critical lines and surfaces in diverse quasiperiodic and non-Hermitian models.

The Gist

Imagine you are trying to describe a very strange, complex object. Usually, scientists look at this object from just one angle—say, from the front. They might measure how "spread out" it is or how "clumped" it looks. If it's neither fully clumped nor fully spread out, they call it a "critical state." It's like a cloud that is neither a solid rock nor a thin mist, but something in between.

However, the problem with looking from just one angle is that your description might change if you walk around the object. What looks like a "cloud" from the front might look like a "rock" from the side. This paper argues that we need a better way to identify these special states—one that doesn't depend on which angle you are looking from.

Here is the simple breakdown of what the authors, Tong Liu and Gao Xianlong, discovered:

1. The "Two-Sided Coin" Rule (The Exclusion Principle)

The authors start with a fundamental rule about how waves (like the waves describing electrons in a material) behave. They prove a "Fourier exclusion principle."

Think of a wave as having two sides:

• Side A (Real Space): Where the wave is located physically (like a person standing in a specific room).
• Side B (Momentum Space): How the wave is moving or vibrating (like the person's speed and direction).

The rule is simple: A wave cannot be tightly packed in both places at once.

• If the wave is tightly squeezed into a small room (localized in real space), it must be spread out and messy when you look at its movement (momentum space).
• If it is tightly squeezed in its movement, it must be spread out in the room.

It's like trying to hold a balloon: if you squeeze it tight in your hand, it puffs out elsewhere. You can't have it tight everywhere.

2. The "Critical State" is the Perfect Balance

So, what is a "critical state"?

• A Localized State is like a person huddled in a corner (tight in the room, messy in movement).
• An Extended State is like a person filling the whole room evenly (spread out in the room, tight in movement).
• A Critical State is the "Goldilocks" zone. It is the only state where the wave is not tightly squeezed in the room, AND not tightly squeezed in its movement.

The authors call this the Liu-Xia Condition. They say: "A critical state is the only time when the 'tightness' (or localization) is zero in both views simultaneously."

3. Why This is a Big Deal (The "Magic Map")

Before this paper, scientists had to look at a wave, measure its shape, and guess if it was critical. It was like trying to find a hidden treasure by looking at a blurry map.

This paper turns the Liu-Xia condition into a magic map. Because the rule about "tightness in both views" is so strict, the authors show that you can use it to predict exactly where these critical states will appear in different types of materials, without having to simulate the whole thing first.

They tested this on three different types of "materials" (mathematical models):

1. The Generalized Map: They found that the critical states form specific lines that depend on the energy of the particle.
2. The Decorated Chain: They found a whole "region" (a safe zone) where critical states exist, plus a specific line where they also exist.
3. The Weird Non-Hermitian Model: They even found a complex, 3D "surface" of critical states in a model that doesn't follow standard symmetry rules.

The Takeaway

The authors aren't just giving a new way to spot these critical states after they are found. They are providing a rulebook that tells you exactly where to find them before you even start looking.

By realizing that criticality is defined by the absence of tightness in two different worlds at the same time, they have created a tool that works across different microscopic structures. It's like realizing that the only way to be a "perfectly balanced" object is to be loose in two different dimensions simultaneously, and using that fact to find those objects anywhere in the universe of physics.

Technical Summary

Technical Summary: Lyapunov Exponents as Duality-Invariant Signatures of Critical States

Problem Statement
The identification of critical eigenstates in disordered and quasiperiodic systems has traditionally relied on wave-function geometry within a single representation (typically real space). Standard diagnostics include participation ratios, multifractal spectra, and finite-size scaling. However, these quantities are basis-dependent; a state appearing critical in one basis may acquire a different interpretation in another. The fundamental physical question is whether criticality is an intrinsic property that survives the comparison between conjugate representations (e.g., real space and momentum space), rather than an artifact of a chosen basis. While the Liu–Xia condition ($\gamma_x(E) = \gamma_m(E) = 0$) has been used phenomenologically to characterize criticality, its rigorous theoretical foundation and predictive power across diverse microscopic structures required clarification.

Methodology
The authors reformulate criticality as a dual-space Lyapunov property. The core methodological advance is the proof of a Fourier exclusion principle: a wave function that is exponentially localized in one representation cannot be exponentially localized in its Fourier-dual representation. This is established by analyzing the discrete Fourier transform of a state confined to a finite localization length $\xi$, demonstrating that such a state must spread over $O(L)$ components in the dual space, precluding exponential localization there.

The authors employ the transfer-matrix approach to define Lyapunov exponents ($\gamma_x$ and $\gamma_m$) for real- and momentum-space problems, respectively. Unlike inverse participation ratios, these exponents are shown to be invariant under bounded local gauge transformations of the transfer matrix. The critical state is rigorously defined by the simultaneous vanishing of these exponents:
$$\gamma_x(E) = \gamma_m(E) = 0$$
This condition implies the absence of exponential confinement (finite localization length) in both conjugate spaces. The authors apply this criterion to three distinct analytically tractable models:

1. A generalized self-dual Aubry–André model with energy-dependent potentials.
2. A class-dual decorated-chain model where real- and momentum-space equations belong to the same Möbius–cosine nonlinear-eigenvalue class (but lack strict Aubry self-duality).
3. A non-Hermitian model with complex one-sided modulation, which lacks any Aubry-type self-duality.

In these models, the Lyapunov exponents are derived exactly (often using the Thouless formula, Avila's complexified phase theory, or Jensen's formula for ergodic averages) to solve for the critical manifolds.

Key Contributions and Results

• Rigorous Foundation of the Liu–Xia Condition: The paper proves that the condition $\gamma_x(E) = \gamma_m(E) = 0$ is not a conjecture but a rigorous consequence of the Fourier exclusion principle. It establishes criticality as a dual-space length-scale statement: the simultaneous disappearance of characteristic exponential length scales in both real and momentum spaces.
• Compatibility with Single-Space Diagnostics: The authors clarify that while multifractal scaling and divergent correlation lengths remain essential single-space probes, the dual-space Lyapunov condition isolates the invariant core shared by these signatures. The Lyapunov condition asserts the absence of exponential cutoffs, while multifractal exponents describe the internal geometry of the state within a chosen basis.
• Exact Predictive Power in Generalized Models:
  • Generalized Self-Dual Model: For a quasiperiodic model with energy-dependent effective potential $V_{\text{eff}}(E) = V - \mu E$, the condition yields closed-form critical lines $E_c^\pm = V \pm 1/\mu$. This extends the standard Aubry–André result (a single transition point) to spectrum-dependent critical branches.
  • Class-Dual Decorated Chain: In a model where real and dual equations are related by "class duality" rather than strict self-duality, the criterion predicts an exact finite critical region defined by $|E - t| \le 2|V|$ and $|E| \le 2|J|$, alongside an exact critical branch $E = Jt/(J - V)$.
  • Non-Hermitian Non-Self-Dual Model: For a model with complex potentials and unidirectional hopping, the authors derive a complex critical surface in the $(E, V)$ plane defined by the simultaneous vanishing of two independently solved Lyapunov drifts ($G_x(E) = 0, G_m(E) = 0$). This demonstrates the criterion's applicability even in the absence of Aubry self-duality.

Significance
The paper claims that the Liu–Xia condition provides more than a post-hoc diagnostic; it serves as an exact solvability principle for locating critical sets across models with distinct microscopic structures. By separating criticality from basis-dependent wave-function morphology, the work suggests that the essential object of criticality is the absence of exponential length scales in a pair of dual descriptions.

The authors posit that this dual-space Lyapunov framework offers a natural route for extending critical-state diagnostics to systems where conventional single-particle notions of real and momentum space are insufficient, such as interacting quasiperiodic systems and many-body localization transitions. In these contexts, the relevant dual spaces may be configuration space and its spectral or graph-dual representation, where a generalized Liu–Xia criterion could provide an invariant method for identifying many-body critical states.