Exceptionally Slow Relaxation from Micro-canonical to Canonical Ensembles in Quasi-one-dimensional Quantum Gases

This paper reports the experimental observation and theoretical modeling of exceptionally slow thermalization in highly excited quasi-one-dimensional quantum gases, demonstrating that near-integrable effects delay the relaxation from a microcanonical to a canonical ensemble over several seconds.

The Gist

The Big Idea: The "Super-Slow" Coffee Stir

Imagine you have a cup of coffee. If you drop a sugar cube in, it eventually dissolves and spreads evenly. If you stir it, the cream mixes with the coffee. This is thermalization: a system moving from a messy, organized state to a calm, mixed-up state where everything is the same temperature.

Usually, this happens fast. But in this experiment, scientists watched a cloud of atoms take several seconds to do something that should take milliseconds. It's like watching a drop of ink in water take an hour to spread out.

Why? Because these atoms were trapped in a "one-lane highway" where the rules of physics are weirdly strict, making them incredibly stubborn about mixing.

---

1. The Setup: The One-Lane Highway

The scientists used a special laser trap to squeeze thousands of Rubidium atoms into a very thin tube. Think of this as a one-lane highway where cars (atoms) can only move forward or backward. They can't change lanes.

In a normal 3D room, if two cars crash, they can bounce off in any direction. But on a one-lane highway, if two cars crash, they can only do two things:

1. Keep going exactly as they were.
2. Swap places (Car A becomes Car B, and Car B becomes Car A).

This is called Integrability. It's like a perfect dance where the partners never actually change their rhythm; they just swap partners. Because of this, the "energy" of the system stays stuck in a specific pattern for a very long time. The atoms refuse to "thermalize" (mix up).

2. The Experiment: The "High-Energy Slide"

The scientists wanted to see what happens when these atoms are very excited (high energy), not just sitting quietly.

• The Slide: They pushed the atoms to the very edge of the trap (like pushing a sled to the top of a huge hill).
• The Release: They let the atoms slide down. As they slid, they gained massive speed (kinetic energy).
• The Catch: They used a very faint "bumpy road" (a weak optical lattice) to make the atoms slide down in a specific way. This created a group of atoms that all had almost exactly the same speed.

In physics terms, they created a Microcanonical Ensemble. Imagine a crowd of runners who all start a race at the exact same moment, running at the exact same speed. They are perfectly synchronized.

3. The Mystery: Why So Slow?

Once the atoms were sliding, the scientists watched them.

• Expectation: They expected the atoms to crash into each other, swap speeds, and eventually mix into a random, "thermal" cloud (like a crowd of people milling about randomly).
• Reality: The atoms stayed in that synchronized, "perfect speed" state for several seconds.

It was like watching a line of perfectly synchronized swimmers who, despite bumping into each other, refused to get out of formation for minutes on end. This is a "high-energy version" of the famous Newton's Cradle (the desk toy with swinging balls), where the balls keep swinging without losing energy.

4. The Solution: The "Leaky" Highway

Why did they eventually mix? Why didn't they stay perfect forever?

The scientists realized the "one-lane highway" wasn't perfectly one-dimensional. It was a quasi-one-dimensional tube. It was thin, but it had a tiny bit of width.

• The Analogy: Imagine the highway is actually a very narrow bridge. Most of the time, cars stay on the road. But occasionally, a car hits a bump and jumps slightly off the bridge into the air (the "transverse" direction) before landing back on the road.
• The Result: When an atom jumps off the road and comes back, it loses a tiny bit of its forward speed. This tiny "leak" of energy breaks the perfect rules of the one-lane highway.

This "leak" is what allows the atoms to finally mix. But because the leak is so small, the mixing process is exceptionally slow.

5. How They Saw It: The "Magic Camera"

The atoms were moving too fast and were too small to see their individual paths. The scientists used a clever trick:

1. They took a photo of the cloud's density (how crowded the atoms were in different spots).
2. They used a Machine Learning AI (like a super-smart detective) to work backward from the photo.
3. The AI reconstructed a "ghost map" (called a Wigner Function) that showed not just where the atoms were, but how fast they were moving.

What they saw on the map:

• Start: A thin, bright ring. This meant all atoms had the same energy (perfectly organized).
• End: A fuzzy, round blob. This meant the atoms had mixed up and lost their perfect organization (thermalized).
• The Process: The ring slowly shrank and turned into a blob over several seconds.

6. The Theory: The "Leaky" Math

The scientists wrote a new math equation (a modified Boltzmann Equation) to explain this.

• Old Math: Said collisions in 1D are perfect; energy never changes.
• New Math: Says collisions are almost perfect, but there's a tiny chance energy leaks out to the "side" (the width of the tube).

They tested this math against their experiment, and it matched perfectly. They even calculated exactly how "leaky" the tube was based on how strongly the atoms bumped into each other.

Summary

• The Problem: In 1D, atoms usually don't mix because they can't change lanes.
• The Discovery: Even when atoms are super-fast and high-energy, they still refuse to mix for a surprisingly long time.
• The Cause: A tiny "leak" of energy into the sideways direction breaks the perfect rules, but it happens so slowly that the atoms stay organized for seconds.
• The Takeaway: Nature has a way of keeping things organized even when you think they should be chaotic. And sometimes, the smallest "leaks" in the system are the only things that let chaos win.

This experiment bridges the gap between the perfect, frozen world of quantum mechanics and the messy, mixed-up world of everyday thermodynamics, showing us just how stubborn quantum particles can be.

Technical Summary

1. Problem Statement

The paper addresses the fundamental question of quantum thermalization in isolated systems. While the Eigenstate Thermalization Hypothesis (ETH) suggests that isolated quantum systems generally thermalize, integrability in one-dimensional (1D) systems prevents this. In strictly 1D, elastic collisions between particles only allow for momentum exchange or retention, preserving the energy distribution and preventing thermalization (the "Newton's Cradle" effect).

Previous research has extensively studied this phenomenon in low-energy states, leading to frameworks like Generalized Hydrodynamics (GHD) and the Generalized Gibbs Ensemble (GGE). However, the behavior of highly excited states in near-integrable systems remained unexplored. Specifically, it was unclear how a system prepared in a high-energy microcanonical ensemble (where all atoms share nearly identical energy) would relax toward a canonical ensemble (thermal equilibrium) when weak integrability-breaking mechanisms (such as transverse mode coupling) are present. The challenge lies in preparing such a high-energy microcanonical state and observing the relaxation dynamics, which are predicted to be exceptionally slow.

2. Methodology

Experimental Protocol

The researchers utilized a Bose-Einstein Condensate (BEC) of $^{85}$Rb atoms confined in a quasi-1D harmonic trap created by a 1560-nm laser.

• State Preparation:
  1. A BEC was initially prepared in a tight 1064-nm dipole trap, displaced by $x_0 = 120\,\mu\text{m}$ from the center of the main harmonic trap.
  2. The 1064-nm trap was switched off, allowing the BEC to slide down the potential gradient.
  3. A weak optical lattice ($V_0 = 0.375 E_r$) was superimposed along the axial direction.
  4. Wannier-Stark Localization: The gradient force caused the BEC to undergo Bloch oscillations with small amplitude, localizing it near the initial position.
  5. Landau-Zener Tunneling: The lattice induced interband transitions, causing the BEC to emit atoms into the first band. These atoms accelerated toward the trap center, acquiring high kinetic energy ($E_0 \approx h \times 2715\,\text{Hz}$).
  6. Microcanonical Ensemble Formation: Due to the tight initial confinement ($\Delta x \ll x_0$), the emitted atoms possessed a narrow energy spread ($\Delta E \ll E_0$), effectively forming a high-energy microcanonical ensemble.

Data Analysis & Reconstruction

• Machine Learning (Denoising Autoencoder): To analyze the system, the team needed to reconstruct the Wigner function $f(x, \tilde{p})$ from the measured real-space density profiles $n(x)$. This is an ill-posed inverse problem exacerbated by experimental noise.
  • They trained a Deep Learning model (Denoising Autoencoder) on a synthetic dataset generated from an ansatz Wigner function: $f(x, \tilde{p}) = A \exp[-(\rho - \rho_0)^2 / 2\sigma_\rho^2]$, where $\rho = \sqrt{x^2 + \tilde{p}^2}$.
  • The model successfully denoised experimental data and extracted the parameters $\rho_0$ (related to mean energy) and $\sigma_\rho$ (related to energy spread).

Theoretical Modeling

• Modified Boltzmann Equation: Standard 1D Boltzmann equations predict zero collision integrals for elastic collisions. To explain the observed slow relaxation, the authors introduced a weak integrability-breaking term.
  • They relaxed strict energy conservation, allowing for a small energy mismatch $\epsilon$ transferred to transverse modes during collisions: $p_1^2/2m + p_2^2/2m + \epsilon = p_1'^2/2m + p_2'^2/2m$.
  • A phenomenological collision integral was derived involving a probability distribution $g(\epsilon) = e^{-\beta\epsilon/2}h(\epsilon)$, where $h(\epsilon)$ is an even function characterizing the energy uncertainty.

3. Key Results

• Observation of Slow Relaxation: The experiment observed that the atomic density profile evolved from a "plateau-like" shape (characteristic of a microcanonical ensemble) to a Gaussian distribution (canonical ensemble) over a timescale of several seconds. This is orders of magnitude slower than typical thermalization times in ultracold atomic gases.
• Phase-Space Evolution: The reconstructed Wigner functions showed a transition from a narrow ring (constant energy shell) to a broad Gaussian distribution. The ratio of the ring radius to its width ($\rho_0/\sigma_\rho$) decreased linearly over time, indicating a smooth crossover from microcanonical to canonical behavior.
• Dephasing Mechanism: The initial rapid dephasing (formation of the plateau) was attributed to the shallow optical lattice. The lattice induced Landau-Zener tunneling and band-mediated reflections, causing atoms to lose phase coherence much faster than trap anharmonicity alone would allow.
• Quantitative Agreement:
  • The relaxation rate $\kappa$ was found to increase monotonically with the scattering length $a_s$ (interaction strength).
  • Numerical simulations of the modified Boltzmann equation, using different ansatz forms for $h(\epsilon)$ (Gaussian, Lorentzian-Gaussian, Step-function), yielded relaxation rates consistent with experimental data.
  • The dimensionless parameter $\eta$ (relating energy uncertainty $\sigma$ to scattering length) extracted from fitting the theory to experiment ($\eta \approx 0.25$) matched the order of magnitude predicted by microscopic two-body scattering theory ($\eta \approx 0.57$).

4. Key Contributions

1. First Experimental Observation of High-Energy Microcanonical Relaxation: The study demonstrates a "high-energy Newton's Cradle" effect, where a system remains in a microcanonical state for seconds despite ongoing collisions, before slowly relaxing due to weak integrability breaking.
2. Novel State Preparation Protocol: The combination of Wannier-Stark localization and Landau-Zener tunneling provides a robust method to prepare high-energy, narrow-bandwidth atomic ensembles in optical traps.
3. Machine Learning for Phase-Space Reconstruction: The successful application of deep learning to reconstruct Wigner functions from noisy density data offers a powerful tool for analyzing complex quantum many-body dynamics.
4. Modified Boltzmann Framework: The paper proposes and validates a modified Boltzmann equation that incorporates stochastic energy mismatch to transverse modes, providing a microscopic description of slow thermalization in quasi-1D systems.

5. Significance

This work significantly extends the understanding of integrability and thermalization in low-dimensional quantum systems. By moving beyond the low-energy regime, it reveals that even in highly excited states, the constraints of 1D kinematics dominate dynamics for remarkably long times. The findings bridge the gap between ideal integrable models and real-world quasi-1D systems, highlighting the role of transverse degrees of freedom as the primary mechanism for integrability breaking. The developed theoretical framework and experimental techniques are crucial for future studies on non-equilibrium quantum dynamics, prethermalization, and the limits of statistical mechanics in isolated quantum systems.