🔢 mathematics

Static features from mixing in short- and long-range Lindbladians: Markov property and correlations

This paper establishes that the static correlation properties of mixed-state phases in long-range Lindbladian systems, specifically the polynomial decay of mutual and conditional mutual information, arise naturally from dynamical features like rapid mixing and frustration-freeness, thereby extending Markov property results to long-range regimes relevant for experimental platforms.

Original authors: Paul Rosa-Ruiz, Matteo Scandi, Ángela Capel, Álvaro M. Alhambra

Published 2026-06-29

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Paul Rosa-Ruiz, Matteo Scandi, Ángela Capel, Álvaro M. Alhambra

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ 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 a vast, crowded room full of people (particles) who are constantly talking to each other. In physics, we often want to know: How much do two people, standing far apart, actually "know" about each other?

In a quiet room (a short-range system), if you whisper to someone next to you, the person on the other side of the room hears nothing. The "information" dies out very quickly, like a whisper fading into silence. This is called exponential decay.

But what if the room is noisy, or the people have a special ability to shout across the room (long-range interactions)? Does the information still fade away, or does it linger? And if it lingers, does it matter for the "state" of the room?

This paper, titled "Static features from mixing in short- and long-range Lindbladians," investigates exactly this. It looks at how information spreads and fades in complex quantum systems that are open to the environment (noisy systems).

Here is the breakdown using simple analogies:

1. The Setup: The Noisy Room (Lindbladians)

Think of the system as a room where people are constantly changing their minds due to outside noise (the environment). In physics, this is modeled by something called a Lindbladian.

The Fixed Point: Eventually, the room settles down. People stop changing their minds randomly and reach a stable "mood." This stable state is the fixed point.
The Goal: The authors want to know: In this stable mood, how much does Person A (on the left) correlate with Person C (on the right), if there is a buffer zone of people (Person B) in between?

2. The Two Types of "Knowing"

The paper measures two types of connections:

Mutual Information (MI): How much do A and C know about each other directly?
Conditional Mutual Information (CMI): If we already know everything about the buffer zone (B), how much extra does A know about C?
- Analogy: If A and C are both talking to B, and B tells you everything they heard, does A still have any secret code with C that B doesn't know? If the answer is "no" (or very little), the system has the Markov property. This is a sign of a healthy, stable phase.

3. The Big Discovery: Short vs. Long Range

The paper compares two scenarios:

Scenario A: The Short-Range Room (Normal Interactions)

People only talk to their immediate neighbors.
Result: The connection between A and C dies out exponentially. It's like a whisper that vanishes after a few meters. The "Markov length" (the distance where they stop knowing each other) is short and finite. This is the standard behavior expected in most stable phases.

Scenario B: The Long-Range Room (Power-Law Interactions)

People can shout across the room, though the shout gets quieter the further it travels (following a power law, like $1/distance^\alpha$ ).
Result: The connection does not vanish exponentially. Instead, it fades away polynomially (like a slow, steady decline).
The Metaphor: Imagine a rumor in a short-range room dies out in 10 seconds. In a long-range room, the rumor might still be faintly audible after 100 seconds, 1,000 seconds, and so on. It never fully disappears instantly; it just gets very, very quiet.
The Finding: The authors prove that even with these long-range shouts, the system still settles into a stable state where the "extra" knowledge (CMI) between A and C eventually becomes negligible, provided the shouting isn't too strong (the decay rate $\alpha$ is high enough).

4. The Ingredients for Stability

The paper identifies two "secret ingredients" that guarantee this stable behavior:

Rapid Mixing: The room must settle down quickly. If the noise keeps the room in a chaotic frenzy forever, no stable pattern emerges.
Frustration-Freeness: This is a fancy way of saying the rules of the room are "consistent." Everyone agrees on the local rules, so there are no conflicting demands that prevent the room from settling. (Think of it as everyone agreeing on the temperature of the room, rather than some wanting it hot and others cold).

If you have Rapid Mixing + Frustration-Freeness, the "long-range shouting" still results in a stable room where distant people don't share deep secrets.

5. The "Gibbs State" (The Thermal Room)

The authors also looked at a specific type of room: one that is in thermal equilibrium (like a hot cup of coffee cooling down).

They proved that even for these complex, non-commuting quantum systems with long-range interactions, the "Markov property" holds.
Translation: Even in a hot, noisy, long-range quantum system, if you look far enough away, the parts of the system become effectively independent. You can reconstruct the whole picture just by looking at local pieces.

6. The Computer Simulation (The Experiment)

To prove this wasn't just math on paper, the authors simulated a Long-Range Ising Model (a classic model of magnets).

Without a magnetic field: They saw the "polynomial decay" (the slow fade) predicted by their math.
With a strong magnetic field (Transverse Field): They saw the decay speed up, looking more like the "exponential decay" of a short-range system.
Takeaway: The math works. The system behaves exactly as the equations predicted: long-range interactions lead to a slower, polynomial fade of correlations, unless something (like a strong field) forces it to behave like a short-range system.

Summary

This paper tells us that even in noisy, long-range quantum systems, order can still exist.

If the system settles down quickly (rapid mixing) and follows consistent local rules (frustration-free), then distant parts of the system stop sharing information, just like in normal systems. The only difference is the speed of the fade:

Short-range: The connection vanishes like a light switch being turned off (Exponential).
Long-range: The connection fades like a dimming light bulb that takes a long time to go out (Polynomial).

This helps physicists understand how to classify different "phases" of matter in the real world, where long-range interactions (like gravity or magnetic forces) are common.

Technical Summary: Static Features from Mixing in Short- and Long-Range Lindbladians

1. Problem Statement and Motivation

The classification of mixed-state phases in quantum many-body systems requires criteria that go beyond the decay of two-point correlation functions. Recent work has established that information-theoretic quantities, specifically the Mutual Information (MI) and the Conditional Mutual Information (CMI), are essential for characterizing phase stability and detecting phenomena such as strong-to-weak spontaneous symmetry breaking. A key concept in this context is the Markov length, which quantifies the decay rate of the CMI; a finite Markov length (exponential decay) is a crucial stability condition for mixed-state phases in short-range systems.

However, many experimental platforms (e.g., Rydberg atom arrays, trapped ions, magnetic dipolar gases) realize systems with long-range interactions that decay as a power law. It remains unclear how the static properties of the fixed points of dissipative processes (Lindbladians) in these long-range regimes relate to their dynamical features. Specifically, it is unknown whether rapid mixing and frustration-freeness in long-range Lindbladians imply the decay of MI and CMI, and if so, whether this decay remains exponential or transitions to a different scaling (e.g., polynomial).

2. Methodology

The paper investigates the relationship between the dynamical properties of local Lindbladian generators and the static information-theoretic properties of their fixed points. The authors employ the following methodological framework:

Dynamical Assumptions: The analysis relies on two primary dynamical conditions:
1. Rapid Mixing: The system converges to its fixed point exponentially fast in time. The authors distinguish between global rapid mixing (convergence for the whole system) and local rapid mixing (convergence for local observables).
2. Frustration-Freeness: A property where every local term of the Lindbladian annihilates the global fixed point individually, not just their sum.
Mathematical Tools:
- Lieb-Robinson (LR) Bounds: The authors utilize LR bounds adapted for open dissipative systems and long-range Hamiltonians to control the propagation of information and the locality of time evolution.
- Recovery Maps: To bound the CMI, the authors construct local recovery maps. They utilize the Fawzi-Renner bound, which relates the CMI to the fidelity of a state reconstructed from its marginals via a recovery channel.
- Perturbation Analysis: To bound the MI, the authors analyze the stability of the fixed point under local perturbations (specifically, removing interaction terms crossing the boundary between subsystems).
- CKG Lindbladian: For thermal states of non-commuting Hamiltonians, the authors utilize the Chen–Kastoryano–Gilyén (CKG) Gibbs sampler, a quasi-local Lindbladian whose fixed point is the Gibbs state.

3. Key Contributions and Results

A. Fixed Points of Local Lindbladians (General and Long-Range)

The paper establishes that dynamical mixing properties imply static decay of information-theoretic quantities for the fixed points of local Lindbladians.

Decay of Conditional Mutual Information (CMI):
- Theorem 1.1: For a local Lindbladian satisfying global rapid mixing and frustration-freeness, the fixed point exhibits a global Markov property. The CMI decays with the shielding distance $R = d(A, C)$ as:
  $I(A : C | B)_\rho \leq p(|A|, |C|) R^{nD/2} (F(R))^{\frac{\lambda}{2(\lambda+v)}}$
  where $p$ is a polynomial, $F$ is the decay function of the interaction, $\lambda$ is the mixing rate, and $v$ is the Lieb-Robinson velocity.
- Corollary 1.2 (Long-Range): For long-range interactions decaying as $F_\alpha(r) \sim (1+r)^{-\alpha}$ with sufficiently high $\alpha$ , the CMI decays polynomially rather than exponentially. This contrasts with short-range systems where exponential decay (finite Markov length) is expected.
Decay of Mutual Information (MI):
- Theorem 1.3: For a local Lindbladian satisfying local rapid mixing and primitivity (uniqueness of the fixed point), the fixed point satisfies a global decay of MI. Notably, frustration-freeness is not required for this result.
- The decay is polynomial for long-range interactions: $I(A : C)_\rho \leq k(|A|, |C|) F_{\alpha'}(d(A, C))$ , where the effective decay rate $\alpha'$ depends on the interaction decay $\alpha$ , the mixing rate, and the Lieb-Robinson velocity.

B. Gibbs States of Long-Range Non-Commuting Hamiltonians

The authors extend the analysis to thermal states of non-commuting Hamiltonians, which are fixed points of quasi-local (KMS) Lindbladians rather than strictly local ones.

Theorem 1.4: The Gibbs state of a non-commuting long-range Hamiltonian satisfies a local Markov property at any temperature.
- The proof extends the results of Chen–Rouzé (2025) for short-range systems to the long-range regime.
- The CMI decays as $\tilde{O}(|C| e^{\mu|A|} R^{-\alpha'})$ , where the decay rate $\alpha'$ is determined by the interaction decay $\alpha$ and the temperature.
- This result relies on the quasi-locality of the CKG Lindbladian and the construction of a time-averaged recovery map.

C. Numerical Verification

The authors provide numerical evidence using the long-range Ising model:

Classical Long-Range Ising Model: Simulations confirm that the CMI of the Gibbs state exhibits a polynomial decay ( $I \sim \ell^{-q}$ ) with the buffer size $\ell$ . The exponent $q$ grows approximately linearly with the interaction decay exponent $\alpha$ .
Transverse-Field Ising Model: Adding a transverse field or increasing the decay rate $\alpha$ (making interactions effectively short-range) appears to shift the decay behavior from polynomial towards exponential, consistent with the expectation that short-range systems possess a finite Markov length.

4. Significance and Claims

The paper claims to provide a rigorous bridge between the dynamical properties of dissipative systems (mixing rates, frustration-freeness) and their static information-theoretic structure (decay of MI and CMI).

Extension to Long-Range Regimes: The primary significance is the demonstration that while rapid mixing ensures the decay of correlations in long-range systems, the nature of this decay changes from exponential (short-range) to polynomial. This implies that the classification of mixed-state phases must accommodate this qualitatively different regime, where the "Markov length" is effectively infinite, but the CMI still decays sufficiently fast to allow for local reconstruction.
Generalization of Markov Properties: The work generalizes the concept of Markov properties (pairwise, local, global) to long-range open systems and Gibbs states of non-commuting Hamiltonians, proving that these systems satisfy specific Markov conditions under natural physical assumptions.
Robustness of Phase Characterization: By establishing that rapid mixing implies the decay of MI and CMI, the results narrow down the conditions under which Lindbladian fixed points belong to the mixed-state analogue of "gapped phases," suggesting that these systems do not generate long-range order in the sense relevant for phase classification.

The authors remain modest regarding the tightness of their bounds, noting that the derived decay exponents may not be optimal and that the transition from polynomial to exponential decay in the presence of transverse fields requires further investigation. They also identify the removal of the frustration-freeness assumption and the determination of sharp thresholds for rapid mixing as open problems.