🔬 materials science

Exploring the Relaxation Landscape of a 2D Quantum Magnet on a 256-Qubit Processor

Using a 256-qubit Rydberg atom array to simulate the 2D transverse-field Ising model, researchers discovered unexpected relaxation regimes, including a prethermal phase and a slowdown crossover where classical methods fail, thereby demonstrating the platform's potential for scientific discovery in nonequilibrium quantum dynamics.

Original authors: Tiago Mendes-Santos, Joseph Vovrosh, Sergi Julià-Farré, Dorian Claveau, Guillaume Villaret, Lucas Béguin, Lucas Leclerc, Laurin Brunner, Wladislaw Krinitsin, Matthias Hecker, Fergus Hayes, Boris Albre

Published 2026-08-10

📖 3 min read☕ Coffee break read

CC BY 4.0

Original authors: Tiago Mendes-Santos, Joseph Vovrosh, Sergi Julià-Farré, Dorian Claveau, Guillaume Villaret, Lucas Béguin, Lucas Leclerc, Laurin Brunner, Wladislaw Krinitsin, Matthias Hecker, Fergus Hayes, Boris Albrecht, Lilian Bourachot, Clémence Briosne-Frejaville, Antoine Cornillot, Julius de Hond, Djibril Diallo, Clément Dupays, Robin Dupont, Thomas Eritzpokhoff, Loïc Henriet, Lucas Lassablière, Arvid Lindberg, Yohann Machu, Hadriel Mamann, Thomas Pansiot, Julien Ripoll, Bruno Ximenez, Henrique Silvério, Joseph Tindall, Markus Schmitt, Markus Heyl, Adrien Signoles, Constantin Dalyac, Antoine Browaeys, Alexandre Dauphin

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 crowded dance floor where everyone is trying to move to the music. In the world of physics, this is similar to how particles in a material behave when they are jostled out of their comfortable, resting state. Usually, physicists expect these particles to eventually "relax" or settle down into a chaotic, lukewarm mess called thermal equilibrium, where all memory of how they started is lost. This process is called thermalization. However, sometimes the music is weird, or the dancers are too stubborn, and the system gets stuck in a long-lived, weird state that refuses to settle down. This is a huge mystery in the field of quantum physics, especially when you have a lot of particles interacting in two dimensions (like a flat sheet), because it is incredibly hard for our best supercomputers to predict exactly what will happen. Understanding these "stuck" states is crucial because it could help us build better quantum computers and understand how the universe works at its most fundamental level.

Now, picture a team of scientists using a giant, high-tech playground made of 256 tiny atoms to solve this mystery. They call this playground a "quantum simulator." Instead of trying to calculate the dance moves on a computer, they actually build the dance floor and watch the atoms move in real-time. They set up a specific scenario called the "transverse-field Ising model," which is like a grid of magnets that can point up or down. They start by freezing all the magnets pointing down, then suddenly hit them with a strong magnetic "wind" (the transverse field) to see how they relax.

What they found was a surprise party of three different behaviors. First, when the wind was just right, the magnets relaxed quickly, just like everyone expected. Second, when the wind was super strong, the magnets got stuck in a "prethermal" trance. It's as if they found a secret rhythm that kept them dancing in a specific pattern for a very long time before they finally gave up and settled down. But the most exciting discovery was a third, middle-ground zone. Here, the relaxation didn't just slow down; it crawled. The magnets seemed to get stuck in a limbo state, moving incredibly slowly toward equilibrium.

This slow-motion zone is a big deal because it's exactly where the world's most powerful classical supercomputers start to fail. When the scientists tried to simulate this slow zone using traditional math, their computers got confused and gave up, losing control of the prediction. But the quantum simulator? It kept working perfectly, showing that the atoms were indeed stuck in this slow dance. This proves that these quantum machines aren't just good at copying what we already know; they are powerful enough to discover new physics that our current math tools can't handle. The researchers suggest that this slow relaxation might be caused by a tug-of-war between different forces in the system, but they admit they don't have the full explanation yet. What they do know is that they've found a new territory in the quantum world, and their 256-atom playground is the first map we have of it.

Technical Summary: Exploring the Relaxation Landscape of a 2D Quantum Magnet on a 256-Qubit Processor

Problem Statement
The relaxation of quantum matter far from equilibrium remains a central open problem in many-body physics. While analog quantum simulators have successfully reproduced theoretical predictions regarding universal dynamical phenomena (such as dynamical phase transitions and emergent hydrodynamics), their potential to act as genuine discovery platforms for physics unforeseen by theory is a frontier only beginning to be explored. Specifically, the relaxation pathways of two-dimensional (2D) quantum systems are less understood than their 1D counterparts. The authors investigate the post-quench dynamics of the 2D transverse-field Ising model (TFIM), a demanding testbed featuring both thermal and quantum phase transitions, to map its relaxation landscape and identify regimes where classical simulation methods fail.

Methodology
The study utilizes a programmable analog quantum processing unit (QPU) based on a 256-qubit ( $16 \times 16$ ) array of trapped $^{87}$ Rb atoms. Local qubits are encoded in the ground state ( $|\downarrow\rangle$ ) and a highly excited Rydberg state ( $|\uparrow\rangle$ ).

Hamiltonian Simulation: The system simulates the 2D TFIM Hamiltonian $\hat{H}_{TFIM} = -\sum_{i<j} J_{ij}\hat{\sigma}^z_i \hat{\sigma}^z_j - h_x \sum_i \hat{\sigma}^x_i$ . The interaction $J_{ij}$ arises from short-range van der Waals forces ( $U \propto 1/r^6$ ), while the transverse field $h_x$ is controlled via the global Rabi frequency $\Omega$ .
Protocol: The system is initialized in a fully polarized product state $|\downarrow \dots \downarrow\rangle$ and quenched to constant $h_x$ and $J$ . The ratio $h_x/J$ is tuned to scan across the phase diagram, ranging from the ferromagnetic (FM) phase to the paramagnetic (PM) phase.
Measurement and Mitigation: Dynamics are probed via projective measurements of $\hat{\sigma}^z$ using 300 experimental shots per time step. Data is corrected for state preparation and measurement (SPAM) errors ( $\epsilon \approx 1\%$ , $\epsilon' \approx 3\%$ ) and averaged over equivalent qubits.
Benchmarking: The QPU results are compared against state-of-the-art classical tensor-network methods, including 2D Tensor Networks (2DTN), Matrix Product States (MPS), and Tree Tensor Networks (TTN), as well as Quantum Monte Carlo (QMC) for thermal equilibrium properties.

Key Contributions and Results
The authors identify three distinct regimes in the relaxation landscape of the 2D TFIM, characterized by the behavior of local classical energy and connected correlations:

Rapid Thermalization (Near the Dynamical Phase Transition):
For quenches near the critical point ( $h_x/J \approx 2.5$ ), the system exhibits rapid thermalization. Correlations relax to their thermal expectation values on timescales of $t \approx 1/J$ . The bulk magnetization retains the sign of the initial state in the FM region, consistent with symmetry breaking, while oscillating around zero in the PM region, consistent with symmetry restoration.
Prethermal Regime (Large Transverse Field):
At large $h_x/J$ (e.g., $h_x/J = 6.0$ ), the system enters a long-lived prethermal plateau. The dynamics are governed by an effective Floquet Hamiltonian derived in the rotating frame, which takes the form of a $U(1)$ -symmetric $ZY$ model: $\hat{H}_{ZY} = \frac{1}{2}\sum_{i<j} J_{ij}(\hat{\sigma}^z_i \hat{\sigma}^z_j + \hat{\sigma}^y_i \hat{\sigma}^y_j)$ .
- Observation: Instead of relaxing to the thermal value of the original TFIM, correlations are sustained at a higher prethermal value predicted by the thermal properties of $\hat{H}_{ZY}$ .
- Correlations: This regime features a strong dynamical buildup of connected correlations extending over large distances, consistent with a quasi-long-range ordered state at low effective temperature.
Slow Relaxation Crossover Regime:
In the intermediate regime (e.g., $h_x/J = 3.5$ ), between the thermal and prethermal limits, the authors discover an unexpected slowdown in thermalization.
- Observation: Correlations decay slowly toward their thermal values, remaining far from equilibrium well beyond the characteristic relaxation time ( $t \gg 1/J$ ).
- Classical Failure: Crucially, this is the regime where state-of-the-art classical tensor-network methods (2DTN, MPS, TTN) lose control at late times. The classical simulations diverge from each other and fail to provide a controlled description due to the rapid growth of entanglement in the 2D system. In contrast, the QPU data remains consistent across system sizes ( $L=8$ and $L=16$ ) and continues to exhibit the slow decay.

Significance and Claims
The paper establishes Rydberg atom arrays as a robust platform for scientific discovery in nonequilibrium quantum many-body dynamics. The primary significance lies in the demonstration that analog quantum simulators can access dynamical regimes that are numerically intractable for classical computers.

Discovery vs. Confirmation: Unlike previous studies that largely reproduced anticipated theoretical effects, this work identifies a "slowdown" regime not fully explained by current theory, highlighting the potential of quantum simulators to reveal new physics.
Benchmarking Limitations: The study explicitly demonstrates the limitations of current tensor-network approaches in 2D non-integrable systems at late times, where entanglement growth renders finite-bond-dimension approximations unreliable.
Robustness: The authors note that while experimental noise affects the quantitative relaxation rates, the qualitative dynamical behavior (including the existence of the slow-relaxation regime) remains robust and does not degrade with system size in the same manner as errors in classical simulations (e.g., MPS).

The work suggests that the slow thermalization may result from the interplay between the breakdown of the leading-order Floquet-Magnus expansion and critical slowing down near the dynamical phase transition, though a definitive theoretical explanation for this specific crossover regime remains an open question for future investigation.

More like this