Geometric fragmentation and anomalous thermalization in cubic dimer model

This paper demonstrates that 3D $U(1)$ quantum dimer models with staggered matter exhibit geometric fragmentation and anomalous thermalization under external electric fields, where flux polarization and Gauss Law constraints trap excitations in 2D planes to create exponentially many athermal fragments, including sectors hosting immobile fractonic excitations.

The Gist

The Big Picture: Why Things Usually "Mix" (and Why This One Doesn't)

Imagine you have a cup of hot coffee and you drop a spoonful of cold milk into it. In the real world, the milk swirls around, mixes with the coffee, and eventually, you get a uniform, lukewarm drink. In physics, this process is called thermalization. It's how systems forget their starting point and settle into a comfortable, random equilibrium.

Usually, if you start with a specific arrangement of particles, they will eventually "mix" and forget where they started. This is the rule for almost everything in nature.

However, this paper discovers a special case where the milk refuses to mix.

The researchers found a way to trap energy in a 3D grid (like a giant Rubik's cube made of quantum particles) so that it gets stuck in specific patterns. Instead of mixing into a uniform soup, the system breaks apart into isolated islands that never talk to each other. They call this Geometric Fragmentation.

---

The Setup: The "Quantum Dimer" Game

To understand how this happens, imagine a giant 3D grid of street corners.

• The Players: On the streets connecting these corners, there are "flux arrows" (like one-way signs).
• The Rules (Gauss Law): At every corner, the number of arrows pointing in must equal the number pointing out (or a specific fixed number). This is a strict traffic rule that cannot be broken.
• The Goal: The system wants to flip these arrows around to create new patterns, but it can only do so if it doesn't break the traffic rules.

Normally, the arrows can flip all over the place, allowing the system to explore every possible pattern and eventually "thermalize" (mix).

The Twist: The "Traffic Jam" in 3D

The researchers applied a strong "wind" (an external electric field) blowing in one direction (let's say, straight up, the Z-axis).

The Analogy: The Frozen Elevator Shaft
Imagine the 3D grid is a skyscraper.

1. The Wind: The wind is so strong that all the elevators (the vertical links) are locked in place. They can't move up or down.
2. The Result: Because the elevators are frozen, the floors (the horizontal XY planes) become isolated from each other. You can't move from the 1st floor to the 2nd floor because the stairs are gone.
3. The Fragmentation: The 3D skyscraper effectively shatters into a stack of independent 2D apartment buildings. The system is now fragmented.

But it gets weirder. Even within a single 2D floor, the system doesn't mix perfectly. It breaks into even smaller, isolated rooms.

The Two Types of "Stuck" Systems

The paper identifies two ways the system gets stuck, depending on the specific pattern of the arrows:

1. The "Fracton" Dance (The Inchworm)

In some specific rooms, the particles are so restricted they can barely move.

• The Analogy: Imagine a game of "Red Light, Green Light" played on a grid. Most pieces are frozen solid. However, there are two special pieces (called Fractons) that are stuck together.
• The Movement: These two pieces can only move if they do a very specific, synchronized "inchworm" dance. They move diagonally across the room, step-by-step, in a perfect loop.
• The Result: They never stop dancing, but they never actually explore the whole room. They just cycle through the same few steps forever. Because they are stuck in this loop, the system never thermalizes. It remembers exactly where it started.

2. The "Non-Fracton" Shuffle

In other rooms, the particles have a little more freedom, but they are still trapped by the geometry of the room.

• The Analogy: Imagine a dance floor where the music is playing, but the dancers are tied to specific spots by invisible elastic bands. They can wiggle and sway (fluctuate), but they can't leave their zone.
• The Result: They don't freeze in a perfect loop like the Fractons, but they still can't mix with the rest of the building. They oscillate back and forth forever, never settling down.

Why This Matters: "Weak" vs. "Strong" Fragmentation

The researchers had to figure out how broken the system is.

• Strong Fragmentation: Imagine a library where 99% of the books are locked in a single, massive vault, and only a few are free. The system is dominated by one big stuck piece.
• Weak Fragmentation (What they found): Imagine a library where the books are scattered into millions of tiny, separate piles. No single pile is huge, but there are so many of them that the system is effectively broken into pieces.

They proved that in their model, the system is Weakly Fragmented.

• The Math Metaphor: If you double the size of the building, the number of these isolated "rooms" doesn't just double; it explodes exponentially.
• The Consequence: Even though the system isn't "frozen" in one giant block, the sheer number of isolated islands means the system as a whole never mixes. It's like trying to mix a million cups of coffee that are all in separate, sealed rooms.

The Takeaway

This paper shows that you don't need disorder (messiness) or perfect symmetry (rigid rules) to stop a system from mixing. You just need geometry.

By arranging a 3D quantum system in a specific way and applying a field, the laws of physics force the system to break into a million tiny, isolated pockets. In these pockets, particles get stuck in loops or dance in place, refusing to forget their past.

In simple terms: The researchers found a way to build a quantum "maze" where the walls are made of math itself, trapping energy in tiny rooms so it can never escape to mix with the rest of the universe. This could be crucial for building quantum computers that don't lose their information (thermalize) too quickly.

Technical Summary

1. Problem Statement

The paper addresses the breakdown of the Eigenstate Thermalization Hypothesis (ETH) in interacting quantum many-body systems. While ETH predicts that generic isolated quantum systems thermalize, certain systems exhibit "anomalous thermalization" or remain in non-thermal states.

• Context: Traditional examples of non-thermalizing systems include integrable models or disordered systems (Many-Body Localization). Recently, translationally invariant systems with local constraints or higher-form symmetries have been found to violate ETH.
• Specific Gap: The authors investigate 3D U(1) Quantum Link Models (QLMs), specifically cubic dimer models with staggered static charges (doped bosonic QLMs). They aim to understand how geometric fragmentation arises in these models under external electric fields, leading to athermal behavior and the emergence of fractons (excitations with restricted mobility).

2. Methodology

The authors employ a combination of analytical derivations and exact numerical diagonalization to study the model.

• Model Definition:

  • System: A 3D cubic lattice ($L_x \times L_y \times L_z$) with periodic boundary conditions.
  • Degrees of Freedom: Defined on links using spin-1/2 operators ($S=1/2$), representing a finite-dimensional Hilbert space (Quantum Link Model).
  • Hamiltonian: Includes electric field energy, magnetic field energy (plaquette terms), and a Rokhsar-Kivelson (RK) term.
  • Constraints: The system is subject to a local Gauss Law with staggered static charges ($G_x = Q(-1)^x$), where $Q=2$ is a specific case studied.
  • External Field: The study focuses on sectors with maximal winding numbers in one direction (e.g., $W_z^{max}$), effectively simulating a strong external electric field that aligns fluxes in that direction.

• Analytical Approach:

  • Dimensional Reduction: They demonstrate that in the maximal winding sector, the 3D dynamics effectively decouple into a stack of independent 2D Quantum Dimer Models (QDMs).
  • Conserved Quantities: They identify new emergent conserved quantities (2D winding numbers of the individual layers) that prevent the Hamiltonian from connecting different 2D configurations, leading to fragmentation.
  • Scaling Analysis: They derive upper and lower bounds for the ratio of the largest fragment size to the total Hilbert space size ($N_{largest}/N_{total}$) to classify the fragmentation as "weak" or "strong."
  • Fracton Dynamics: They analytically solve the eigenstates for specific 2D winding sectors containing "fractons" (constrained excitations), mapping the Hamiltonian to a cycle graph adjacency matrix.

• Numerical Approach:

  • Exact Diagonalization: Performed on small lattices (e.g., $2\times2\times4$, $2\times4\times2$) to explore real-time dynamics.
  • Observables: Tracked time evolution of:
    • Fidelity ($F(t)$): Overlap with the initial state.
    • Shannon Entropy ($S(t)$): Spread of the wavefunction in the Fock basis.
    • Kinetic and Potential Energy: To distinguish between frozen and dynamic sectors.

3. Key Contributions and Results

A. Geometric Fragmentation

The central finding is that in the maximal winding sector (e.g., $W_z^{max}$), the 3D Hilbert space fragments into exponentially many disconnected sectors.

• Mechanism: The external field freezes the flux in the $z$-direction. The Hamiltonian can only flip plaquettes in the $xy$-planes. However, the global 3D winding numbers impose constraints on the sum of the 2D winding numbers of the stacked layers ($W_{3D} = \sum W_{2D}$).
• Result: Different combinations of 2D winding sectors can satisfy the same global 3D constraint. Since the Hamiltonian cannot change the 2D winding number of a specific layer (due to frozen inter-layer fluxes), these combinations form disconnected subspaces.
• Classification (Weak vs. Strong):
  • The authors prove this is Weak Fragmentation.
  • Reasoning: The ratio $N_{largest}/N_{total}$ scales as $e^{-L}$ (where $L$ is linear size) rather than $e^{-L^3}$. The entropy density of the largest fragment approaches the thermodynamic entropy density of the full system as $L \to \infty$ (correction scales as $O(1/L^2)$). This distinguishes it from "Strong Fragmentation" where the largest fragment is negligible.

B. Emergence of Fractons and Anomalous Dynamics

Within the fragmented sectors, the authors identify two distinct dynamical classes:

1. Fractonic Fragmentation Class:

   • Structure: Contains layers where the 2D winding numbers correspond to sectors hosting fractons (e.g., $|W_y| = L-1, |W_x| = n$).
   • Dynamics: These excitations have severely restricted mobility, moving only along specific diagonal lines ("inchworm motion").
   • Observables:
     • Potential energy ($E_{pot}$) and Kinetic energy ($E_{kin}$) remain constant in time.
     • Fidelity and Entropy exhibit persistent, non-decaying oscillations.
     • The system never thermalizes; it remains trapped in a subspace of the Hilbert space.
   • Analytical Solution: The authors provide exact eigenstates and eigenenergies for these sectors, showing the spectrum is determined by the discrete Fourier transform of a cycle graph, leading to quasi-periodic dynamics.

2. Non-Fractonic Fragmentation Class:

   • Structure: Fragmented sectors where at least one layer does not host fractons (i.e., has a variable number of flippable plaquettes).
   • Dynamics: Energy fluctuates in time, and the system explores a larger portion of the subspace.
   • Observables: While they do not thermalize to the ETH prediction (fidelity does not go to zero, entropy does not saturate at the maximum), they show more complex fluctuations than the fractonic class. They still exhibit athermal behavior due to the geometric constraints.

C. Non-Fragmented Subspaces

For comparison, the authors study sectors without maximal flux (e.g., $W=(2,2,1)$). In these cases, the Hilbert space is connected, and the system thermalizes rapidly: fidelity decays to zero, and entropy saturates at the maximal value, consistent with ETH.

4. Significance

• New Mechanism for ETH Violation: The paper establishes geometric fragmentation driven by external fields and local constraints in 3D translationally invariant systems as a robust mechanism for preventing thermalization, distinct from disorder or integrability.
• Fractons in 3D Gauge Theories: It provides a concrete realization of fractons (subdimensional particles) in a 3D quantum dimer model, linking lattice gauge theory concepts with condensed matter phenomena like spin ices and high-$T_c$ superconductors.
• Thermodynamic Limit Validity: Unlike many previous studies limited to small systems, this work provides analytical proofs that the fragmentation persists and is "weak" in the thermodynamic limit, offering a rigorous classification of the non-ergodic phase.
• Quantum Simulation Relevance: The findings are highly relevant for near-term quantum simulators (e.g., Rydberg atoms, cold atoms) which can realize these lattice gauge theories. The distinct signatures (oscillating fidelity, constant energy in fractonic sectors) provide clear experimental targets to observe athermal dynamics and geometric fragmentation.

In summary, the paper demonstrates that applying an external electric field to a 3D doped quantum dimer model induces a geometric fragmentation of the Hilbert space. This leads to the emergence of fractonic excitations and athermal dynamics, characterized by persistent oscillations and a lack of thermalization, even in the thermodynamic limit.