Classical and Quantum Theory of Fluctuations for Many-Particle Systems out of Equilibrium

This paper extends Yu.L. Klimontovich's classical fluctuation approach to quantum many-particle systems, demonstrating that reformulating kinetic equations in terms of fluctuations enables accurate, time-linear scaling simulations of nonequilibrium phenomena—such as the GW approximation—while overcoming the high memory costs associated with traditional nonequilibrium Green function methods.

The Gist

Imagine you are trying to predict the weather. You have a massive crowd of people (particles) moving around, bumping into each other, and reacting to the wind (external forces).

If you try to track every single person's exact location and speed at every moment, you would need a supercomputer the size of a city just to do the math for a few seconds. This is the problem physicists face when studying quantum systems (like electrons in a computer chip or atoms in a laser-cooled gas). The math gets so heavy that it becomes impossible to run simulations for long periods or large groups.

This paper introduces a clever new way to solve this problem by changing how we look at the crowd. Instead of tracking the "average" weather, they track the fluctuations—the tiny, random deviations from the average.

Here is the breakdown of their idea using simple analogies:

1. The Old Way: The "Perfect Librarian" vs. The "Chaos"

• The Problem: Traditional methods (like the G1-G2 scheme) are like a librarian who tries to catalog every single book in a library, including every page every time a book is opened. It's accurate, but the library grows so fast that the librarian runs out of space (memory) and time (CPU power).
• The Bottleneck: To predict how the crowd moves, you usually need to know how pairs of people interact. Storing the data for every possible pair interaction is like trying to write down every possible conversation between every two people in a stadium. It's too much data.

2. The New Idea: The "Fluctuation Detective"

The authors, inspired by a Soviet physicist named Yu.L. Klimontovich (who turned 100 this year), suggest a different approach.

Instead of trying to predict the exact path of every person, they ask: "How much does the crowd wiggle away from the average?"

• The Analogy: Imagine a calm lake (the average state). The water is still. But if you throw a stone, you get ripples (fluctuations).
• The Trick: Instead of simulating the whole lake, the authors simulate the ripples. They realized that if you understand how the ripples behave, you can figure out how the whole lake moves without needing to track every single water molecule.

3. The Quantum Twist: The "Ghost" and the "Real"

In the quantum world, things are weird. Particles can be in two places at once, and they don't like to be in the same spot (Pauli Exclusion Principle).

The authors developed a Quantum Fluctuation Theory. They treat these quantum "wiggles" as if they were random, chaotic events.

• The "Stochastic" Method: They use a technique called Stochastic Mean-Field Theory. Imagine you have a thousand different "what-if" scenarios. In one scenario, the wind blows left; in another, it blows right. You run a simple simulation for each scenario (which is fast) and then take the average of all of them.
• The Result: This average gives you the same accuracy as the super-heavy "perfect librarian" method, but it runs on a standard laptop because you aren't storing the massive pair-data anymore. You are just running many small, fast simulations.

4. The "Multiple Ensembles" (The Magic Trick for Two-Time Data)

There was one catch: The old "average" method couldn't easily tell you about response times (e.g., "If I push the crowd at 2:00 PM, how does it react at 2:05 PM?"). It was good for "what is happening now," but bad for "what happens next."

To fix this, they invented the Multiple Ensembles (ME) approach.

• The Analogy: Imagine you have two groups of actors. Group A represents the "push," and Group B represents the "reaction."
• Normally, in quantum mechanics, you can't just multiply these groups because they are "ghosts" that don't follow normal rules.
• The authors found a way to run Group A and Group B separately and then combine their results mathematically. This allows them to calculate how the system reacts over time (like a shockwave moving through the crowd) without needing the heavy memory of the old methods.

5. Why This Matters

• Speed: They can simulate systems with hundreds of sites (like a long chain of atoms) that were previously impossible to simulate.
• Accuracy: They proved that their "fast and fuzzy" method gives the exact same results as the "slow and perfect" method for weakly interacting systems.
• Future: This opens the door to simulating complex materials, better batteries, and new quantum computers, because we can finally model how these materials behave when they are pushed out of balance (like when a laser hits them).

Summary

The paper is about trading memory for speed.

• Old Way: Store everything about every pair of particles. (Accurate but impossible for big systems).
• New Way: Run thousands of small, random simulations of "wiggles" and average them out. (Fast, accurate, and scalable).

It's like realizing you don't need to know the exact path of every grain of sand in an hourglass to know how fast the sand is flowing; you just need to understand the flow of the "ripples" in the sand.

Technical Summary

1. Problem Statement

The accurate theoretical description of correlated classical and quantum many-particle systems out of equilibrium (e.g., dense plasmas, correlated solids, ultracold atoms) is computationally challenging.

• Classical Systems: While exact simulations via molecular dynamics are possible, they are often too expensive for large systems. Kinetic theories (like Boltzmann) or hierarchy approaches (BBGKY) are alternatives but require approximations.
• Quantum Systems: Exact simulations (e.g., Quantum Monte Carlo) are limited by the "sign problem" or exponential scaling. The Non-Equilibrium Green Functions (NEGF) method offers a rigorous description but suffers from unfavorable cubic scaling ($N_t^3$) with the number of time steps ($N_t$).
• The G1–G2 Scheme: A recent advancement, the G1–G2 scheme, achieves linear scaling ($N_t$) by propagating single-particle ($G_1$) and two-particle ($G_2$) Green functions simultaneously. However, this comes at a massive cost: the memory and CPU time scale as $N_b^6$ (where $N_b$ is the basis size) due to the need to store and compute all matrix elements of the two-particle correlation function.
• The Gap: There is a need for an approach that retains the high accuracy of NEGF/G1–G2 methods (specifically the $GW$ approximation) but reduces the computational cost to that of mean-field theories, enabling the study of large systems and long simulation times.

2. Methodology

The authors propose a Quantum Fluctuations Approach, extending the classical theory developed by Yu.L. Klimontovich to quantum systems.

A. Theoretical Framework

• Classical Foundation: The paper revisits Klimontovich's theory, where the microscopic phase space density $N(x,t)$ is decomposed into a mean distribution $f(x,t)$ and fluctuations $\delta N(x,t)$. The dynamics are governed by a hierarchy of equations for these fluctuations. The Polarization Approximation (PA) is identified as a key method that neglects three-particle correlations while retaining two-particle fluctuations, leading to the Balescu-Lenard kinetic equation.
• Quantum Generalization: The authors define quantum fluctuations using the lesser component of the single-particle Green function operator, $\hat{G}^<$.
  • Fluctuations are defined as $\delta \hat{G} = \hat{G}^< - \langle \hat{G}^< \rangle$.
  • Two-particle fluctuations are defined as correlation functions of these single-particle fluctuations: $L = \langle \delta \hat{G} \delta \hat{G} \rangle$.
  • This formulation establishes a direct link between quantum fluctuations and the exchange-correlation function ($L$) in NEGF theory.
• Quantum Polarization Approximation (QPA): By assuming weak coupling (where two-particle correlations are small compared to fluctuations and three-particle correlations are negligible), the authors derive the QPA.
  • Equivalence: The QPA is proven to be equivalent to the $GW$ approximation within the G1–G2 scheme (with additional exchange contributions) in the weak coupling limit.
  • Key Insight: The equation of motion for the two-particle fluctuation $L$ can be reformulated entirely in terms of the single-particle fluctuation $\delta \hat{G}$. This eliminates the need to propagate the full two-particle matrix, which is the primary bottleneck in G1–G2.

B. Stochastic Mean-Field (SMF) Implementation

To solve the operator equations for $\delta \hat{G}$, the authors employ a Stochastic Mean-Field (SMF) approach:

• Random Realizations: The quantum operator $\delta \hat{G}$ is replaced by an ensemble of random classical variables $\Delta G^\lambda$.
• Initial Conditions: The ensemble is constructed to reproduce the first two moments of the quantum initial state (mean and variance).
  • Stochastic Sampling: Uses complex Gaussian or four-point distributions to generate random realizations.
  • Deterministic Sampling: Solves a system of nonlinear equations to find a specific set of realizations that exactly satisfy the first two moments, reducing the number of required samples.
• Multiple Ensembles (ME) Approach: To overcome the limitation of standard SMF (which cannot compute non-commuting observables like response functions due to the commutativity of classical random variables), the authors introduce two independent ensembles ($\Delta G^{(1)}$ and $\Delta G^{(2)}$).
  • Products of operators are mapped to products of variables from different ensembles (e.g., $\delta \hat{G}_1 \delta \hat{G}_2 \to \Delta G^{(1)} \Delta G^{(2)}$).
  • This allows the calculation of two-time correlation functions and response functions (e.g., density response $\chi^R$) while maintaining the computational structure of mean-field equations.

3. Key Contributions

1. Extension of Klimontovich's Theory: Successfully generalized the classical fluctuation theory to quantum many-body systems, establishing a rigorous link between fluctuations and NEGF.
2. Derivation of QPA: Demonstrated that the Quantum Polarization Approximation is equivalent to the $GW$ approximation in the weak coupling limit, providing a physical justification for the method.
3. Computational Efficiency: Developed the Stochastic Polarization Approximation (SPA). By replacing the two-particle propagation with stochastic sampling of single-particle fluctuations, the method reduces the scaling:
   • CPU Time: From $O(N_b^6 N_t)$ (G1–G2/$GW$) to $O(N_s N_b^4 N_t)$ (SPA), where $N_s$ is the number of samples. With constant $N_s$, this effectively becomes $O(N_b^4 N_t)$, similar to mean-field scaling.
   • Memory: From $O(N_b^4)$ to $O(N_b^2)$.
4. Access to Spectral Properties: Introduced the SPA-ME (Multiple Ensembles) variant, enabling the calculation of dynamic response functions and structure factors for systems far from equilibrium, which were previously inaccessible with time-local stochastic methods.

4. Results

The authors validated the theory using the Fermi-Hubbard model on 1D lattice chains:

• Sampling Comparison: For weak coupling ($U=0.1J$), both stochastic (Gaussian/4-point) and deterministic sampling methods reproduced the QPA results with high accuracy (relative deviations $< 10^{-3}$). Deterministic sampling showed superior convergence for stronger coupling ($U=1.0J$).
• Comparison with $GW$:
  • The SPA results for density dynamics following a "confinement quench" showed excellent agreement with the $GW$ approximation for weak coupling.
  • Deviations appeared at stronger coupling or longer times, primarily affecting oscillation amplitudes rather than frequencies, consistent with the weak-coupling assumption of the approximation.
• Response Functions:
  • Ground State: The SPA-ME calculated the dynamic structure factor $S(q, \omega)$, showing excellent agreement with the Random Phase Approximation (RPA) for weak coupling.
  • Nonequilibrium: The SPA-ME successfully computed the retarded density response function $\chi^R(t, t')$ for systems driven out of equilibrium, matching the two-time QPA and $GW$ results.
• Scalability: The method demonstrated that large systems (up to 50 sites) could be simulated efficiently, whereas standard G1–G2 is currently limited to much smaller systems due to memory constraints.

5. Significance

• Bridging the Gap: This work provides a computationally efficient alternative to the G1–G2 scheme, making high-accuracy many-body simulations (equivalent to $GW$) feasible for large-scale systems (hundreds of sites) and long simulation times.
• New Capabilities: It enables the study of dynamic response properties (spectral functions, structure factors) in nonequilibrium scenarios, a task that was previously restricted to equilibrium or required prohibitive computational resources.
• Theoretical Unification: It unifies classical fluctuation theory (Klimontovich) with modern quantum many-body theory (NEGF/$GW$), offering a fresh perspective on how to handle correlations and fluctuations.
• Future Applications: The approach is highly promising for studying warm dense matter, ultracold atoms in optical lattices, and correlated solids, particularly where spatially uniform systems require large memory that currently limits G1–G2 implementations.

In summary, the paper presents a robust, scalable, and theoretically sound framework for simulating correlated quantum systems out of equilibrium, overcoming the memory and time bottlenecks of traditional NEGF methods while retaining their accuracy in the weak-to-moderate coupling regime.