Hilbert Space Fragmentation from Generalized Symmetries

✨

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

The Big Picture: A Room Full of Locked Doors

Imagine a giant, chaotic dance floor (this is the Hilbert Space, or the universe of all possible states a quantum system can be in). Usually, if you drop a dancer onto this floor, they will eventually mix with everyone else, moving randomly until the whole room looks like a uniform, well-mixed crowd. This is called thermalization or ergodicity. It's like dropping a drop of ink into a glass of water; eventually, the ink spreads everywhere.

However, physicists have discovered some systems where the ink doesn't spread. The dancer gets stuck in a tiny corner, or the room is filled with invisible walls. This is called Hilbert Space Fragmentation. It means the system gets "stuck" and never fully mixes.

For a long time, scientists thought this only happened because of very specific, rigid rules (like a bouncer checking IDs at the door). But this paper argues: No, the room is actually full of hidden, complex doors that we didn't know about.

The New Discovery: "Generalized Symmetries"

The authors say that the reason the dance floor is fragmented isn't just because of simple rules, but because of Generalized Symmetries.

Think of a standard symmetry like a Global Rule: "Everyone must wear a red hat." This divides the room into two groups: Red Hats and Blue Hats. That's simple.

Generalized Symmetries are like Local, Shape-Shifting Rules:

Higher-Form Symmetries: Imagine a rule that says, "Every row of dancers must have the same number of people." If you have a 10x10 grid, you have 10 rows. If you can change the number of people in each row independently, you suddenly have millions of different ways to arrange the crowd that are all "legal" but completely separate from each other.
Subsystem Symmetries: Imagine a rule that only applies to specific shapes, like "Every 2x2 square must have a specific pattern."
Non-Invertible Symmetries: These are like a magic trick where you can combine two groups of dancers to make a third group, but you can't easily reverse the trick to get the original two back.

The Result: Instead of just a few big rooms (like "Red Hat Room" and "Blue Hat Room"), these generalized rules create exponentially many tiny, isolated rooms. The number of rooms grows so fast that for a large system, it's impossible to jump from one room to another. The system is "fragmented."

The Twist: It's Not Always "Broken"

Usually, when a system gets stuck in a tiny room and can't mix, we say it has "broken" the laws of thermodynamics (it's not ergodic).

The authors make a crucial point: Just because the system is stuck in a tiny room doesn't mean the laws of physics are broken.

The Old View: "The system is stuck, so it's broken."
The New View: "The system is stuck because it's following a very specific, complex rule (a generalized symmetry) that keeps it in that tiny room."

If you understand the rule, the system is actually behaving perfectly normally within its own room. It's not broken; it's just very specific.

Real-World Examples from the Paper

The paper uses two main examples to prove this:

The Quantum Link Model (The 3D Lattice):
Imagine a 3D grid of magnets. There are rules about how the magnets on the edges of a square must balance. The paper shows that there is a hidden "partial" rule (a non-invertible symmetry) that says, "If these two layers of magnets are perfectly aligned, the layer in between is frozen." This creates millions of tiny, isolated pockets where the magnets can't move freely.
The PXP Model (The Rydberg Atoms):
This is a famous model used to study atoms that are excited to high energy states. There is a rule: "You can't have two excited atoms next to each other."
- The Old Way: Scientists thought this rule just created a few big sections.
- The New Way: The authors show that this simple "no neighbors" rule actually creates a massive number of tiny, isolated pockets based on exactly where the excited atoms are. The system is fragmented because of this local rule, which acts like a generalized symmetry.

Disorder-Free Localization: The "Frozen" Crowd

One of the most exciting parts of the paper is explaining Disorder-Free Localization.

Usually, for a system to get stuck (localize), you need disorder—like throwing random rocks on the dance floor to trip people up. This is called "Anderson Localization."

But this paper shows you can get the same "stuck" effect without any rocks.

How? If the "rooms" (Krylov sectors) created by the generalized symmetries are themselves uneven or lopsided, the system gets stuck in that unevenness.
The Analogy: Imagine a dance floor where the floor tiles are slightly different heights in a specific pattern. Even if the dancers are perfectly coordinated and there are no obstacles, they can't move smoothly across the floor because the structure of the room itself prevents it. The "defect" isn't an obstacle; it's the shape of the room.

The Conclusion: A New Map for Physics

The authors conclude that we need to redraw our map of quantum physics.

Before: We thought "Fragmentation" meant "Something is broken."
Now: We realize "Fragmentation" often just means "There are more complex rules (Generalized Symmetries) than we realized."

By recognizing these hidden, complex rules, we can explain why some quantum systems behave strangely without needing to say the laws of physics are broken. It unifies many different weird phenomena (like "Quantum Scars" and "Disorder-Free Localization") under one single umbrella: Generalized Symmetries.

In short: The universe isn't broken; it's just much more compartmentalized than we thought, with millions of tiny, invisible rooms created by complex, shape-shifting rules.

1. Problem Statement

Isolated, non-integrable quantum many-body systems are generally expected to be ergodic, meaning they thermalize according to the Eigenstate Thermalization Hypothesis (ETH). However, certain models exhibit Hilbert space fragmentation, where the Hilbert space splits into an exponentially large number of dynamically disconnected sectors (Krylov sectors) as the system size increases.

The Conflict: Traditionally, fragmentation has been interpreted as a breakdown of ergodicity caused by the absence of sufficient conserved quantities. Conventional global symmetries (group structures) generate at most a polynomial number of sectors; thus, an exponential number of sectors was thought to imply "ergodicity breaking" beyond standard symmetry constraints.
The Gap: Recent discoveries of generalized symmetries (higher-form, subsystem, gauge, and non-invertible symmetries) have not been fully integrated into the understanding of real-time dynamics. It remains unclear whether the exponential fragmentation observed in various models is a genuine breakdown of ergodicity or simply a consequence of these broader symmetry classes.

2. Methodology

The authors employ a combination of rigorous mathematical proofs, operator algebra analysis, and specific model examinations to re-evaluate the origins of Hilbert space fragmentation.

Propositional Framework: The authors formulate two key propositions to establish sufficient conditions for fragmentation:
- Proposition I: Establishes that if a local, translation-invariant Hamiltonian possesses a conserved operator $U$ that is diagonal in a product basis and admits a number of disjoint translations $t(L) \sim L^f$ (where $f \geq 1$ ), the system exhibits at least $\exp(cL^f)$ Krylov sectors.
- Proposition II: Extends this to non-invertible symmetries (partial isometries). It proves that even if an operator does not commute with the Hamiltonian globally, it can induce fragmentation within a specific symmetry sector if it acts as a partial isometry ($D = UP$) that commutes with the Hamiltonian within that projected subspace.
Model Analysis: The authors apply these propositions to three specific systems:
1. 3D U(1) Quantum Link Model (QLM): Analyzing the "maximal-winding" sector previously thought to exhibit ergodicity breaking.
2. PXP Model: Investigating the Rydberg blockade model known for quantum many-body scars and fragmentation.
3. Gauge Theories: Examining the mechanism of disorder-free localization.
Krylov-Restricted Thermalization: The authors analyze thermalization within individual Krylov sectors, demonstrating that these sectors can thermalize to non-translation-invariant ensembles without violating ergodicity within the sector.

3. Key Contributions

A. Unification of Fragmentation and Generalized Symmetries

The paper demonstrates that generalized symmetries are the primary origin of exponential Hilbert space fragmentation.

Higher-Form and Subsystem Symmetries: The authors prove that $p$ -form symmetries (supported on $(d-p)$ -dimensional submanifolds) and subsystem symmetries naturally generate exponentially many disjoint translations. Since these operators are diagonal in the product basis, they fragment the Hilbert space into $\exp(cL^p)$ sectors.
Gauge Symmetries: Local gauge symmetries (e.g., Gauss' law constraints) also generate exponentially many superselection sectors. The fragmentation is often "strong" because the dimension of the largest physical sector grows exponentially slower than the full product Hilbert space.

B. Non-Invertible Symmetries as Fragmentation Drivers

A novel contribution is the identification of non-invertible symmetries (realized as partial isometries) as a mechanism for fragmentation within existing symmetry sectors.

The authors show that operators like the 2-form partial isometry $D_{xy}$ in the 3D QLM commute with the Hamiltonian only within specific subspaces (e.g., maximal-winding sectors).
This explains why certain sectors appear to fragment further: it is not a lack of symmetry, but the presence of a non-invertible symmetry that restricts dynamics further.

C. Reinterpretation of Ergodicity Breaking

The paper argues that the term "Hilbert space fragmentation" should be reserved for cases where the Krylov structure cannot be explained by generalized symmetries.

If an exponential number of sectors arises from generalized symmetries (higher-form, gauge, non-invertible), the system is not breaking ergodicity in the fundamental sense; it is simply evolving within a symmetry-defined subspace.
This provides a unified framework for phenomena previously categorized as distinct: quantum many-body scars, gauge theory constraints, and fragmented models.

D. Disorder-Free Localization without Ergodicity Breaking

The authors resolve the mechanism behind disorder-free localization (localization in clean systems).

They demonstrate that disorder-free localization arises from Krylov-restricted thermalization.
If the initial state lies in a Krylov sector that lacks translation invariance (e.g., due to spatially inhomogeneous Gauss' law eigenvalues), the system thermalizes within that sector to a non-translation-invariant ensemble.
This leads to the appearance of localized defects and suppressed transport, without requiring the breaking of ergodicity or the presence of gauge symmetry in the traditional sense.

4. Key Results

Exponential Scaling: The number of Krylov sectors in models with higher-form or subsystem symmetries scales as $\exp(cL^p)$ , where $p$ is the dimension of the symmetry support, proving that these symmetries are sufficient to cause strong fragmentation.
Partial Isometry Fragmentation: In the 3D U(1) QLM, the "ergodicity breaking" observed in the maximal-winding sector is shown to be caused by a non-invertible 2-form symmetry ( $D_{xy}$ ), which acts as a partial isometry. This operator commutes with $H$ only within the projected sector, generating the extra Krylov sectors.
PXP Model: The fragmentation in the PXP model is attributed to a local conserved 2-local projector ( $G_n = |\uparrow\uparrow\rangle\langle\uparrow\uparrow|$ ), which fits Proposition I. This reinterprets the PXP model as an unconventional $Z_2$ gauge theory.
Thermalization in Sectors: Numerical simulations of the 1D U(1) quantum link model show that while local observables relax (thermalize), the long-time state retains the spatial inhomogeneity of the initial Gauss' law configuration. This confirms that thermalization can occur within a non-translation-invariant Krylov sector.

5. Significance

Theoretical Unification: The paper bridges the gap between high-energy physics (gauge theories, higher-form symmetries) and condensed matter physics (fragmentation, scars). It suggests that generalized symmetries are the fundamental organizing principle for non-ergodic dynamics.
Refining Definitions: It challenges the community to refine the definition of "ergodicity breaking." If a system has exponentially many sectors due to generalized symmetries, it is not "breaking" ergodicity but rather respecting a broader symmetry structure.
Quantum Simulation Implications: For quantum simulations of gauge theories, the paper highlights that noise may couple to the exponentially many "unphysical" symmetry sectors. Understanding the full fragmented structure is crucial for error mitigation and interpreting simulation results.
New Mechanisms for Localization: By linking disorder-free localization to Krylov-restricted thermalization rather than disorder or gauge symmetry, the authors provide a new pathway to engineer localized phases in clean quantum systems.

In conclusion, the authors establish that generalized symmetries (higher-form, subsystem, gauge, and non-invertible) are the root cause of the exponential Hilbert space fragmentation observed in diverse quantum models. This insight reclassifies many "ergodicity-breaking" phenomena as symmetry-constrained dynamics, offering a unified theoretical framework for understanding non-equilibrium quantum matter.