Operator-Space Fragmentation

• Operator-space fragmentation is the decomposition of the full operator algebra into exponentially many invariant subspaces due to local constraints and emergent symmetries.
• It creates a hierarchy of non-mixing sectors that disrupt traditional ergodic behavior, influencing operator spreading, localization, and entanglement growth.
• Key examples include Floquet-Clifford circuits with emergent walls and Lindbladian models using local projectors to enforce block-diagonal dynamics.

Operator-space fragmentation refers to the decomposition of the space of operators—typically the full algebra of observables or superoperators in a quantum many-body system—into a direct sum of dynamically invariant subspaces due to model-specific constraints, symmetries, or dynamical blockades. This yields a proliferation of superselection sectors beyond those associated with global symmetries, often with exponential scaling in the system size, resulting in a hierarchy of disconnected operator subspaces that do not mix under the relevant dynamics. Operator-space fragmentation fundamentally modifies ergodicity, operator spreading, entanglement growth, and spectral statistics, with implications for both closed and open quantum systems.

1. Conceptual Foundations of Operator-Space Fragmentation

Operator-space fragmentation is defined as a block-diagonalization of the algebra of operators (e.g., $B(\mathcal{H})$ for quantum state Hilbert space $\mathcal{H}$) into exponentially many invariant subspaces under the system’s dynamical generator, which can be a Floquet unitary, a Lindblad superoperator, or an interacting Hamiltonian (Kovács et al., 2024, Paszko et al., 19 Jun 2025, Moudgalya et al., 2021, Essler et al., 2020). The fragmentation scenario is distinguished from standard global symmetries, which induce only polynomially many sectors. Rather, fragmentation occurs when the bond or commutant algebra of the system contains an exponentially large Abelian or non-Abelian structure, implying the existence of a large set of strong or emergent local (or quasi-local) integrals of motion (IoMs).

A precise algebraic criterion: operator-space fragmentation occurs if the commutant algebra $$\mathcal{C} = \{O \in L(\mathcal{H})\,|\, [O, h_j] = 0\,\forall j\}$$ has dimension $\mathrm{dim}\,\mathcal{C} \sim e^{\alpha N}$ for some $\alpha > 0$ in the thermodynamic limit ($N$ number of sites) (Moudgalya et al., 2021). The fragmented sectors correspond to non-mixing Krylov subspaces, labeled by local or nonlocal occupancy, blockade, or configuration constraints.

2. Microscopic Mechanisms: Wall Configurations and Local Blockades

In Floquet quantum circuits, particularly those constructed from nearest-neighbor Clifford gates, operator-space fragmentation is induced by emergent "wall" configurations. A wall is a spatial arrangement of gates (e.g., involving 2 or more contiguous qubits) that completely isolates the Pauli support of any operator attempted to be transported across its boundaries under Heisenberg evolution. These walls arise with finite probability in random Clifford circuits, partitioning the global Pauli algebra into direct sums of non-communicating fragments (Kovács et al., 2024).

For a wall (of width $k$), its existence is rigorously formalized: Let $L$, $C$, $R$ denote contiguous left, center, and right blocks of the chain. A $\mathcal{H}$0-wall circuit $\mathcal{H}$1 endows $\mathcal{H}$2 with an invariant Pauli operator $\mathcal{H}$3 such that the full algebra decomposes as

$\mathcal{H}$4

$\mathcal{H}$5

and this fragmentation cannot be undone under the circuit evolution; the fragments never mix for any $\mathcal{H}$6.

Walls are robust to perturbations: In perturbed Floquet-Clifford circuits with disordered non-Clifford gates applied with probability $\mathcal{H}$7 per site, walls survive with probability determined by the destruction of the conserved subspace on each perturbed site (probability $\mathcal{H}$8 per gate), rendering the mean operator-spreading length tunable by $\mathcal{H}$9 (Kovács et al., 2024).

3. Quantitative Signatures: Localization Length, Entanglement Bottlenecks, and Spectral Statistics

The presence of persistent walls imposes a typical operator localization length $$\mathcal{C} = \{O \in L(\mathcal{H})\,|\, [O, h_j] = 0\,\forall j\}$$0, setting the mean size of dynamically accessible operator fragments. In the Floquet-Clifford context, for random circuits,

$$\mathcal{C} = \{O \in L(\mathcal{H})\,|\, [O, h_j] = 0\,\forall j\}$$1

and the stopping probability $$\mathcal{C} = \{O \in L(\mathcal{H})\,|\, [O, h_j] = 0\,\forall j\}$$2 for a fragment is

$$\mathcal{C} = \{O \in L(\mathcal{H})\,|\, [O, h_j] = 0\,\forall j\}$$3

with $$\mathcal{C} = \{O \in L(\mathcal{H})\,|\, [O, h_j] = 0\,\forall j\}$$4, diverging only as $$\mathcal{C} = \{O \in L(\mathcal{H})\,|\, [O, h_j] = 0\,\forall j\}$$5 (fully non-Clifford, delocalized limit).

Entanglement dynamics are sharply constrained: unbroken walls act as entanglement bottlenecks. In the stabilizer basis, the von Neumann entropy across a wall is strictly bounded, $$\mathcal{C} = \{O \in L(\mathcal{H})\,|\, [O, h_j] = 0\,\forall j\}$$6, and ensemble statistics yield $$\mathcal{C} = \{O \in L(\mathcal{H})\,|\, [O, h_j] = 0\,\forall j\}$$7 in the localized ($$\mathcal{C} = \{O \in L(\mathcal{H})\,|\, [O, h_j] = 0\,\forall j\}$$8) phase, compared to volume-law scaling within fragments (Kovács et al., 2024).

Spectral form factors (SFF) probe ergodicity breaking. In fragmented circuits, SFF displays a characteristic quadratic ramp $$\mathcal{C} = \{O \in L(\mathcal{H})\,|\, [O, h_j] = 0\,\forall j\}$$9 (for $\mathrm{dim}\,\mathcal{C} \sim e^{\alpha N}$0), and an early plateau at $\mathrm{dim}\,\mathcal{C} \sim e^{\alpha N}$1, consistent with a direct sum of independent blocks—a clear signature of non-ergodicity. For $\mathrm{dim}\,\mathcal{C} \sim e^{\alpha N}$2, the SFF restores the linear CUE ramp expected from random matrix theory (Kovács et al., 2024).

4. Operator-Space Fragmentation in Open Quantum Systems

Lindbladian dynamics can also exhibit operator-space fragmentation. For quantum chains governed by frustration-free Hamiltonians with Pauli-string jump operators (as in many stabilizer code models), the Lindblad superoperator $\mathrm{dim}\,\mathcal{C} \sim e^{\alpha N}$3 can be block-diagonalized into exponentially many invariant fragments. This is codified via representations built from bond and commutant algebras of superoperators (Paszko et al., 19 Jun 2025, Essler et al., 2020).

Fragments arise naturally via local projectors (e.g., $\mathrm{dim}\,\mathcal{C} \sim e^{\alpha N}$4 on site $\mathrm{dim}\,\mathcal{C} \sim e^{\alpha N}$5) that commute with the Lindbladian. The joint eigenspaces $\mathrm{dim}\,\mathcal{C} \sim e^{\alpha N}$6—labeled by local projectors—yield a direct-sum decomposition $\mathrm{dim}\,\mathcal{C} \sim e^{\alpha N}$7, with each fragment hosting independent dynamics, often corresponding to integrable or free-fermion effective Hamiltonians.

A paradigmatic example is the quantum ASEP, where operator-space fragmentation splits the Lindbladian into sectors governed by integrable XXZ chain Hamiltonians (twisted or open boundary conditions) (Essler et al., 2020).

5. Algebraic and Graph-Theoretic Structure

The organizing principle for operator-space fragmentation is the algebraic structure of the model: the bond algebra (generated by local Hamiltonian or gate terms) and its commutant. If the commutant algebra $\mathrm{dim}\,\mathcal{C} \sim e^{\alpha N}$8 is Abelian and simultaneously diagonalizable in a product-state basis, fragmentation is termed "classical," with Krylov subspaces labeled by configuration patterns (e.g., spin or occupation). If $\mathrm{dim}\,\mathcal{C} \sim e^{\alpha N}$9 is non-Abelian, fragmentation is "quantum," and subspaces may require entangled bases (e.g., dimer or dot-and-link basis) (Moudgalya et al., 2021).

Graph-theoretic tools, especially frustration graphs, map the structure of superoperators: vertices represent dynamical terms, and edges their (anti)commutation. Claw-free graphs are linked to free-fermion solvability in the fragments; line graphs of bipartite graphs admit Jordan-Wigner mappings to quadratic fermion Hamiltonians (Paszko et al., 19 Jun 2025).

6. Physical Implications: Localization, Quantum Memory, and Nonergodic Phases

Operator-space fragmentation results in a spectrum of physical phenomena distinct from conventional thermalization:

• Localization: Operator support remains confined within fragments, suppressing ballistic spreading even in the absence of disorder. Emergent local IoMs (e.g., from wall intersections) act as dynamically generated conservation laws, confining operator evolution (Kovács et al., 2024).
• Entanglement Protection: Bottlenecking at fragment boundaries strictly limits entanglement growth, suggesting applications to engineered decoherence protection and stabilized subspaces—a potential mechanism for quantum error correction in open systems (Paszko et al., 19 Jun 2025).
• Slow Relaxation and Rich Spectral Structure: The full dissipative spectrum can be obtained as a sum over independent integrable Hamiltonians. Relaxation is governed by the fragment with slowest spectral gap; quantum operators decay more quickly than the diagonal (classical) block (Essler et al., 2020).
• Phase Structure and Exceptional Points: Tuning Lindbladian parameters (e.g., dissipation strength $\alpha > 0$0 or perturbation probability $\alpha > 0$1) can induce phase transitions in the operator dynamics, signaled by coalescence of eigenvalues at exceptional points and sudden changes in spectral or correlational properties (Paszko et al., 19 Jun 2025).

7. Exemplar Models and Summary Table

The most prominent systems exhibiting operator-space fragmentation include:

Model Type	Fragmentation Mechanism	Algebraic Signature
Floquet-Clifford Circuits (Kovács et al., 2024)	Emergent multi-qubit walls	Abelian wall algebra
Pauli-Lindblad Models (Paszko et al., 19 Jun 2025, Essler et al., 2020)	Local projectors from jump operators	Commutant algebra structure
$\alpha > 0$2–$\alpha > 0$3 and Dipole Models (Moudgalya et al., 2021)	Kinematical constraints or blockades	Abelian (classical fragmentation)
Temperley-Lieb Chains (Moudgalya et al., 2021)	Quantum group algebraic symmetries	Non-Abelian (quantum fragmentation)

Each model demonstrates distinct manifestations of operator-space fragmentation: localization of operator support in Clifford circuits, integrable block structure in Lindbladian quantum chains, and dimensionally large commutant algebra in kinetically constrained or quantum group symmetric models.

---

Operator-space fragmentation is thus established as a universal organizational principle of constrained many-body quantum dynamics, with profound implications for ergodicity breaking, integrability, entanglement structure, and quantum information protection in both closed and open quantum systems (Kovács et al., 2024, Paszko et al., 19 Jun 2025, Moudgalya et al., 2021, Essler et al., 2020).