Temporal Coarse-Graining as the Origin of Macroscopic Friction in Quantum Spin Chains via Data-Driven Liouvillian Extraction

arXiv:2605.05604 · quant-ph, cond-mat.stat-mech · Submitted 2026-05-07 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.

Kai: Today's paper: "Temporal Coarse-Graining as the Origin of Macroscopic Friction in Quantum Spin Chains via Data-Driven Liouvillian Extraction".

Mira: This paper introduces a fully data-driven framework, integrating generalized Extended Dynamic Mode Decomposition (gEDMD) with Mori-Zwanzig projection,

Kai: First, who's behind it and why it matters.

Title and authors: Kai: Now we're moving into a deeper look at the specific findings of "Temporal Coarse-Graining as the Origin of Macroscopic Friction in Quantum Spin Chains via Data-Driven Liouvillian Extraction." They are detailing exactly how they set up their chaotic XXZ spin chain and what kind of initial state they use for their study.

Mira: They start by setting up a non-integrable XXZ spin-one/two chain with nearest and next-nearest neighbor interactions, which is important because it ensures the system exhibits the quantum chaotic behavior necessary for thermalization to occur in a realistic way.

Lev: When you talk about chaotic behavior in a spin chain, that immediately makes me think of how hard it would be to simulate on actual hardware. If we try to run this on a real quantum processor, we have to worry about decoherence messing up the chaotic evolution before we can even get those observables.

Kai: They compute the exact unitary time evolution using the state vector propagation method and prepare an initial state that resembles a Haar random state, which is designed to mimic the infinite-temperature ensemble properties up to small fluctuations.

Mira: That initial setup is clever because it ensures that while we're looking at quantum behavior, we are starting from a point of maximum unbiased thermal fluctuation, reproducing the properties of the infinite-temperature ensemble without needing to actually prepare an impossibly complex state.

Lev: If they can successfully mimic that ensemble property on a finite system, it gives us some hope for testing hydrodynamic models on smaller-scale quantum simulators where we can't achieve perfect infinite temperature.

Kai: The paper then focuses on their data extraction framework, which is the generalized Extended Dynamic Mode Decomposition or gEDMD, and how they use it to find the governing Liouvillian operator without getting stuck with those ambiguous matrix logarithms.

Mira: They detail two modes: one where they use an exact continuous time derivative to preserve microscopic reversibility, and another where they introduce a finite-difference temporal coarse-graining as a first-order estimator of the generator from an observer predicting the future.

Lev: That two-mode approach is interesting; it allows them to keep track of the reversible part and then see how adding that coarse-graining introduces the irreversible features. It’s like separating the microscopic rules from what we observe macroscopically.

Kai: And when they expand their observable dictionary to include both local spin density and spin current, they can finally evaluate mechanical elasticity, friction, and kinematic viscosity independently.

Mira: That's where I see the real payoff; it allows them to prove that when this dictionary is restricted to a finite size of observables—say fifteen macroscopic observables—the reversible information that escapes into higher-order entanglement automatically becomes non-Hermitian components in their extracted operator L.

Lev: So, from a hardware standpoint, this means if we run a simulation with only fifteen key variables, we're not just losing information; we're mathematically encoding the loss of that reversible information into the imaginary parts of our results.

Kai: It’s a sophisticated way to map out where the physics is being projected away in our reduced model, which is really useful for understanding limitations.

Mira: This framework moves beyond simple diffusive models by explicitly tracking current and momentum, showing how those terms are fundamentally linked to the hydrodynamic coefficients derived from this procedure.

Lev: I'm still thinking about the practical execution; if we wanted to implement this on a real quantum hardware system, we’d need to ensure our measurement setup can capture both spin density and spin current simultaneously with high fidelity for these fifteen observables.

Kai: That’s the experimental challenge—the measurement apparatus needs to be robust enough to handle that dictionary size and the necessary temporal resolution shift.

The paper's summary: Kai: Moving on, they discuss what aspects of this framework they suggest improving in "Temporal Coarse-Graining as the Origin of Macroscopic Friction in Quantum Spin Chains via Data-Driven Liouvillian Extraction." They aren't just presenting a finished product; they are pointing out where the limitations lie.

Mira: They point out that the emergent fluid picture, while useful for showing positive friction and viscosity, is inherently transient because it breaks down either when the coarse-graining scale becomes too large, leading to a sinc-filter effect or when finite-size quantum echoes interfere with the bulk dynamics.

Lev: That's a critical point for simulation design; if we want to predict long-term behavior, we can't just run the system forever at one fixed coarse-graining scale because the model itself becomes invalid as time progresses.

Kai: They also emphasize that the sign of any emergent friction is tied directly to the causal, forward-in-time direction of their inference method; this means time-symmetric estimators consistently yield zero net dissipation at any coarse-graining scale.

Mira: They suggest that we need a better way to handle the limitations of dictionary size, implying that if we use too few observables, we risk creating an anti-dissipative mirror image in the extracted operator L.

Lev: That's a direct warning for error correction researchers: if you simplify your model too much by cutting down the observable space, you might end up with a mathematical structure that looks like it has no dissipation when it actually does.

Kai: So, they are basically suggesting that we need to be careful about how much information we keep versus how much projection we perform and what temporal resolution we choose.

Mira: The paper suggests the need for a more sophisticated way to quantify the difference between environmental decoherence dissipation and dissipation caused purely by our observer's choice of temporal coarse-graining.

Lev: That distinction is vital for us, because if we can separate those two sources, we can tell if the physics we are seeing is due to true coupling or just our own modeling choices.

Kai: Ultimately, the paper suggests that stable macroscopic transport requires navigating a specific temporal sweet spot—a narrow window between zero-friction reversibility and artificial over-damping caused by things like the sinc-filter effect at larger scales.

Mira: They suggest that for future work, we should focus on developing better methods to manage these scale dependencies so that the hydrodynamic coefficients remain stable across different coarse-graining scales.

Lev: That leads right into the next part of their plan: building a model that is robust enough to handle those scale changes without completely losing the physical description of how friction and viscosity behave.

The paper's improvements: Kai: So, wrapping up "Temporal Coarse-Graining as the Origin of Macroscopic Friction in Quantum Spin Chains via Data-Driven Liouvillian Extraction," the main point is that macroscopic friction and viscosity only appear when you introduce a finite temporal coarse-graining, and this emergence is driven by the causal direction of inference.

Mira: They successfully showed that genuine irreversible hydrodynamics with positive friction and viscosity only shows up when we average out microscopic coherent oscillations over a timescale where the system enters an intermediate functional regime.

Lev: From my perspective, this means that any future simulation we run on real hardware needs to be aware of this temporal dependency; if we use the exact derivative limit, we’ll just see zero dissipation for friction and viscosity.

Kai: And they conclude by emphasizing that the arrow of time enters the macroscopic description through the act of forward prediction itself, which is a really insightful way to frame it.

Mira: The paper establishes that while spatial projection is a necessary condition for irreversibility, stable transport further demands appropriate temporal coarse-graining.

Lev: For error correction research, this gives us a clear direction on how we need to structure our models: prioritize the causal inference direction when modeling open systems.

Kai: So, in short, this paper provides a data-driven way to extract hydrodynamic coefficients from quantum dynamics based on temporal resolution rather than just assuming they exist.

Mira: It’s a lot of work, but it gives us a much more fundamental understanding of how dissipation arises from the observer's perspective on time.

Lev: I think the path forward is clear: we need to keep pushing for models that respect this causal structure in every step, especially when modeling systems outside of equilibrium.

Kai: We’ve got some really interesting material here, and I think it's worth sharing with everyone listening as we move on to our next discussion.

Conclusion: Kai: So, to wrap up this session on "Temporal Coarse-Graining as the Origin of Macroscopic Friction in Quantum Spin Chains via Data-Driven Liouvillian Extraction," we’ve seen how they’ve shown that macroscopic friction isn't just some arbitrary property but something that emerges strictly from our choice of temporal resolution.

Mira: Exactly, Kai; the core finding is that true irreversible hydrodynamics only shows up when we introduce a finite coarse-graining window, because the exact derivative limit leaves us with zero net dissipation.

Lev: From a hardware standpoint, it’s crucial that we keep in mind this; if we try to build a simulation based on the exact time evolution without any coarse-graining, we won't see those friction terms at all.

Kai: Right, Lev? So the whole point is that the arrow of time enters our macroscopic description through how we choose to predict things forward in time.

Mira: That’s a very specific physical mechanism; they proved that the sign of the emergent friction is inherited directly from that causal inference direction.

Lev: It means if we want to model real-world transport, we have to bake that directional dependency right into our equations, otherwise we end up with models like the exact derivative one which just oscillate without relaxation.

Kai: And they even pointed out how a restricted dictionary size can lead to an "anti-dissipative mirror image," which is a very practical limitation for any experimental setup.

Mira: That’s a necessary caveat; the method doesn't tell us much about what happens when we use far too few observables, so we have to be careful about how much information we project away.

Lev: I think the implication is that for error correction researchers, this tells us that our models need to account for both environmental coupling and observational limits separately if we want accurate predictions on real hardware.

Kai: It’s a complex idea, but the paper shows it’s possible to extract these coefficients systematically using gEDMD and Mori-Zwanzig projection.

Mira: Indeed, and it builds on earlier work by showing how you can keep strict microscopic reversibility while still deriving a physically relevant reduced model.

Lev: It suggests that if we want to build robust quantum fluid models, we have to focus on those intermediate timescales where the reversible fluctuations start averaging out into stable positive friction and viscosity.

Kai: So, while they haven't built a new physical system themselves in this study, the framework they developed is something that can be used across various open quantum systems.

Mira: I think the real impact is on how we interpret simulation results; it gives us a rigorous way to distinguish between dissipation caused by coupling and dissipation caused by our modeling choices.

Lev: It’s a solid piece of theoretical work, and it gives us a clear roadmap for what kind of reduced models we should be aiming for when we eventually translate these ideas to actual quantum simulators.

Kai: Alright team, that covers the main points of "Temporal Coarse-Graining as the Origin of Macroscopic Friction in Quantum Spin Chains via Data-Driven Liouvillian Extraction." We’ve got some really deep stuff to think about.

Mira: It’s a paper that shows how fundamental concepts like time and measurement resolution are intrinsically linked to emergent physics in complex quantum systems.

Lev: Next up, we look at the effect of superconducting fluctuations on nonreciprocal dichroism and gyrotropy, which is another fascinating piece of condensed matter research.

Seiki Saito

Graduate School of Science and Engineering, Yamagata University

quant-ph, cond-mat.stat-mech

Submitted: 2026-05-07

Updated: 2026-07-27

Journal ref: Phys. Rev. Research 8, 033348, 22 September, 2026

DOI: 10.1103/41m6-x2m9

Code: https://github.com/saitos-lab/gedmd-xxz-hydrodynamics

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 86/100

The gist: This paper introduces a fully data-driven framework, integrating generalized Extended Dynamic Mode Decomposition (gEDMD) with Mori-Zwanzig projection, to systematically extract Navier-Stokes

Key concepts

Temporal Coarse-Graining
This involves averaging microscopic coherent oscillations over a timescale where the system enters an intermediate functional regime. It is crucial because true irreversible hydrodynamics, showing positive friction and viscosity, only appears when this finite coarse-graining window is introduced; the exact derivative limit results in zero net dissipation.
Data-Driven Liouvillian Extraction
This framework uses generalized Extended Dynamic Mode Decomposition (gEDMD) and Mori-Zwanzig projection to find the governing Liouvillian operator. It allows researchers to extract hydrodynamic coefficients by analyzing observables like spin density and current, moving beyond simple diffusive models.
Causal Inference Direction
The sign of emergent friction is directly tied to the causal, forward-in-time direction used in the inference method. This means time-symmetric estimators consistently yield zero net dissipation at any coarse-graining scale, highlighting that the arrow of time enters macroscopic descriptions through forward prediction.
Dictionary Size Restriction
Restricting the observable dictionary to a finite size can lead to an 'anti-dissipative mirror image' in the extracted operator L. This warns researchers that simplifying the model too much by cutting down observables might mathematically encode a lack of dissipation when it is physically present.

Terminology

Summary

This paper introduces a fully data-driven framework, integrating generalized Extended Dynamic Mode Decomposition (gEDMD) with Mori-Zwanzig projection, to systematically extract Navier-Stokes hydrodynamic coefficients from the exact unitary dynamics of an isolated quantum many-body system. It addresses the fundamental challenge of reconciling true microscopic reversibility with the emergence of macroscopic irreversible hydrodynamics by demonstrating that genuine dissipation is an emergent phenomenon dictated by the observer's temporal resolution and causal direction of inference.

System and Model Setup

The study investigates a chaotic XXZ spin-1/2 chain under open boundary conditions, defined by a Hamiltonian incorporating nearest-neighbor (NN) and next-nearest-neighbor (NNN) interactions, specifically setting parameters such that the system is non-integrable to ensure quantum chaotic behavior essential for thermalization. The exact unitary time evolution is computed using the state vector propagation method. To ensure unbiased thermal fluctuations, an initial state resembling a Haar random state is prepared, which reproduces the properties of the infinite-temperature ensemble up to small fluctuations.

Data Extraction Framework (gEDMD)

The core methodology employs generalized Extended Dynamic Mode Decomposition (gEDMD) to identify the governing Liouvillian operator without relying on potentially ambiguous matrix logarithms. This framework operates in two modes:

  1. Using an exact continuous time derivative, where the generating operator is identified via the relation X˙ ≈ LX (Equation 3), preserving strict microscopic reversibility.

  2. Utilizing a finite-difference temporal coarse-graining, where the time derivative is replaced by the causal forward difference: X˙ (t) ≈ [X(t + ∆tcg) − X(t)]/∆tcg, which serves as the first-order estimator of a generator from an observer predicting the future.

Observable Dictionary and Dimensionality Reduction

The framework's power lies in its ability to systematically expand the observable dictionary, moving beyond simple spin density to explicitly include both local spin density (Z) and spin current (J). This expansion is crucial because it allows for the independent evaluation of mechanical elasticity, friction, and kinematic viscosity. A key mathematical feature is that when this dictionary is restricted to a finite size (e.g., 15 macroscopic observables), the "information of the reversible unitary evolution that escapes into the orthogonal complement of unobserved higher-order many-body entanglement is mathematically and automatically converted into non-Hermitian components (eigenvalues with negative real parts) in L."

Emergence of Macroscopic Dissipation

The paper demonstrates a clear physical dichotomy based on temporal resolution. The exact-derivative limit preserves strict microscopic reversibility, yielding zero net dissipation for macroscopic friction and viscosity. Genuine irreversible hydrodynamics, characterized by "strictly positive γ > 0 and ν > 0," only emerges when a finite temporal coarse-graining is introduced. This blurring of microscopic coherent oscillations drives the system through a crossover timescale where reversible fluctuations average out into an intermediate functional regime that yields stable, positive friction and viscosity.

Predictive Capability and Causal Inference

The extracted coarse-grained generator constitutes a predictive reduced model requiring only O(N) macroscopic observables. The sign of the emergent friction is inherited from the causal, forward-in-time direction of the coarse-grained inference: time-symmetric estimators yield zero net dissipation at any coarse-graining scale. Furthermore, predictive accuracy is shown to depend on dictionary size; even a minimal hydrodynamic dictionary provides a generator that quantitatively tracks the exact evolution of the bulk spin density over the friction timescale, whereas the exact-derivative generator fails as a predictive model because it continues to oscillate without relaxation.

Conclusion and Physical Interpretation

The findings establish that macroscopic friction is not an absolute property but an emergent phenomenon dictated by temporal resolution. The analysis confirms that while spatial projection (tracing out the environment) is a necessary condition for irreversibility, stable macroscopic transport further demands appropriate temporal coarse-graining. The fluid model survives only within a narrow temporal sweet spot, bounded by zero-friction reversibility at the exact-derivative limit and artificial over-damping due to the sinc-filter effect at larger timescales. Ultimately, the arrow of time enters the macroscopic description through the act of forward prediction itself. Genuine macroscopic diffusion arises from current and momentum dissipation generated during this coarse-graining process.

Improvements for AI systems

Based on this research, here are the specific improvements that could be made to AI systems, categorized by how they leverage the findings:

)Improved AI System Capabilities:

  1. Genie-AI for Emergent Dissipative Modeling (Leveraging Section V & VI):

The system can move beyond purely predictive modeling of unitary evolution and transition into a framework that explicitly models emergent macroscopic friction and viscosity.

  1. Causal Inference Engine (Leveraging Temporal Coarse-Graining Analysis in Section VII):

The AI will be trained to recognize the causal, forward-in-time direction of inference. It can distinguish between time-symmetric estimators (which yield zero net dissipation) and causally directed estimators (which yield positive friction/viscosity).

  1. Adaptive Dictionary Architect (Leveraging Section VI):

The system will dynamically determine the optimal, minimal observation dictionary required to capture macroscopic transport phenomena with high fidelity, rather than relying on pre-defined fixed sets of observables.

  1. Non-Hermitian Dynamics Predictor (Leveraging Section III & IV):

The AI can model systems where information leakage into unobserved degrees of freedom manifests as non-Hermitian components (eigenvalues with negative real parts), allowing it to predict the onset and magnitude of apparent dissipation arising from spatial projection alone.

  1. Scale-Dependent Model Selector (Leveraging Figure 12 & Section VII):

The system can automatically select the appropriate model regime based on the required fidelity: switching between a reversible, zero-dissipation model (exact derivative limit) and a physically relevant, dissipative Navier-Stokes fluid model (intermediate coarse-graining scale).

)Specific Improvements for AI Systems:

  1. Predictive Capability in Open/Noisy Quantum Systems:

The AI system can accurately predict the long-term relaxation trajectory of macroscopic observables (like bulk spin density) not just based on instantaneous unitary evolution, but by incorporating a learned, time-dependent dissipative term that correctly scales with the coarse-graining window size.

  1. Robustness Against Dictionary Underfitting:

When faced with limited computational resources or incomplete observational data (i.e., a restricted dictionary), the AI will automatically assess if the projection is leading to an anti-dissipative mirror image (as seen in Appendix A) and flag this as a failure mode, suggesting an expansion of the observable space (dictionary size).

  1. Quantifying Information Loss vs. Dissipation:

The system can differentiate between dissipation arising from true environmental coupling (decoherence/non-Hermitian components) and dissipation arising purely from the observer's temporal resolution (coarse-graining effects), allowing for a more nuanced understanding of the physical mechanisms driving transport.

  1. Generating Physically Consistent Reduced Models:

Instead of generating arbitrary reduced models, the AI will generate models where the friction and viscosity coefficients are derived directly from the causal structure of its inference mechanism, ensuring that any emergent dissipation is physically grounded in the forward-in-time direction.

Related papers