Realizing Braided Fusion 2-Categories as (3+1)D Mixed-State Topological Orders

AI summary

Overview

Research area: Condensed matter theory and quantum information — specifically topological phases of matter, quantum error correction under decoherence, and the higher-category mathematics of topological order in three spatial dimensions.

Technical level: Advanced. The paper assumes familiarity with topological order, anyon theory, higher-form symmetries, 't Hooft anomalies, and fusion 2-categories. The lattice models are built from toric-code stabilizer Hamiltonians, which makes the constructions concrete, but the categorical framing (braided fusion 2-categories, crossed extensions, group cohomology) is graduate-level and beyond.

Scope (one sentence): The paper shows that every braided fusion 2-category in two stated families arises as the algebra of topological operators of a (3+1)D mixed state produced by incoherently proliferating (possibly anomalous) excitations of an Abelian gauge theory and then gauging 0-form symmetries strongly and weakly.

What This Paper Is About

Topological order in pure quantum states is well studied, but real devices are noisy, and the mixed states left behind by decoherence can host forms of order that no pure ground state can realize. This paper asks what that looks like in three spatial dimensions, where topological excitations are both point-like and loop-like and are therefore described by a braided fusion 2-category rather than a braided fusion category.

The goal is to build explicit (3+1)D mixed-state topological orders — including intrinsically mixed ones that cannot be connected to any gapped pure ground state by finite-depth local channels in both directions — and to show that the categorical data of braided fusion 2-categories is exactly what strong higher-form symmetries and their anomalies encode physically.

Key Contributions

1. A dictionary between categorical data and physical symmetry data. The authors relate the parameters of a braided fusion 2-category to strong higher-form symmetries and their anomalies, and formalize this as a theorem: the data is a finite abelian group E, a class β ∈ H⁵(B²E; ℂ^×) specifying the proliferated loops, a faithfully graded crossed extension encoding the strong 0-form symmetry N, and a group H = G/N acting by braided autoequivalences encoding the weak symmetry.

2. A single universal construction. Every braided fusion 2-category in the two families considered is realized by one procedure: incoherently proliferate possibly anomalous excitations of an Abelian gauge theory, then perform strong and weak 0-form gauging. The authors implement this explicitly on the lattice for representative examples.

3. A new mechanism for intrinsic mixedness with no (2+1)D analogue. They construct an imTO whose topological operators braid non-degenerately — so remote detectability holds — but which carries a gravitational anomaly, namely the (3+1)D ℤ₂ gauge theory with a global gravitational anomaly, called the "FcFl theory." This is intrinsically mixed not because of transparent operators but because of the anomaly.

4. Lattice models realizing each case. Concrete stabilizer models are given for an imTO with a single transparent fermion (from the (3+1)D fermionic ℤ₂ toric code), one with a single fermionic loop (from the bosonic ℤ₂ toric code, reviewing prior work), one with a non-Abelian fermionic loop (via gauging a weak ℤ₂ layer-swap symmetry of two copies), and the FcFl theory (from a stack of bosonic and fermionic ℤ₂ toric codes).

Main Findings

• The central ansatz: The topological operators of a (3+1)D mixed-state topological order are assumed to form a braided fusion 2-category that need not be non-degenerate. Transparent operators — trivial mutual braiding with everything — can still carry nontrivial self-statistics (a fermionic particle or fermionic loop), and such anomalous transparent operators violate remote detectability and enforce long-range entanglement.

• Two distinct routes to intrinsic mixedness: (i) a degenerate braided fusion 2-category whose transparent operators generate anomalous strong symmetries, generalizing the (2+1)D construction; (ii) a non-degenerate category carrying a gravitational anomaly that no strictly (3+1)D ground state can realize. Degeneracy alone is insufficient: transparent operators that are anomaly-free bosons give classical ensembles two-way connected to trivial states.

• Fermionic point particle imTO: Decohering the (3+1)D fermionic ℤ₂ toric code with a channel built from the framed string operators L̃ₑ preserves an anomalous ℤ₂ 2-form symmetry generated by fermion loops L_f(γ). The anomaly is the (4+1)D response π∫Sq²B = π∫w₂ ∪ B, reflecting fermionic statistics of the line. At p = 1/2 the state is a zero-correlation-length state built from dressed plaquette stabilizers B̂ₚ = B̃ₚA_{v(p)}. The surviving strong operators form 2SVect, with the fermion transparent — the direct (3+1)D analogue of the (2+1)D {1, ψ} ≃ SVect imTO.

• Choosing the framing matters: Using Lₑ rather than L̃ₑ as the Kraus operator would fail, because Lₑ do not mutually commute (they transport fermions, which have exchange statistics), so the channel would not preserve the fermion loop as a strong symmetry.

• The bosonic counterpart is trivial: Maximally decohering the bosonic charges of the (3+1)D ℤ₂ toric code yields ρ ∝ ∏ₚ(1 + Bₚ), a classical ensemble of closed ℤ₂ strings preparable from |0⟩^⊗N by applying each A_v independently with probability 1/2 — two-way connected to a trivial state. Nontriviality in the fermionic case hinges entirely on the fermionic statistics of the proliferated particle.

• Fermionic loop imTO (review): Starting from the (3+1)D bosonic ℤ₂ toric code (presented as a ℤ₄ toric code with m² fluxes condensed), decohering with S̃ₑ proliferates fermionic loops. The strong symmetry S_f(Σ) has a fermionic loop anomaly with (4+1)D response π∫B ∪ Sq¹B = π∫w₃ ∪ B, and the strong operators form 2Vect^β_{ℤ₂}. Notably, in (3+1)D a purely bosonic parent theory already suffices to produce a nontrivial imTO — unlike (2+1)D.

• Non-Abelian fermionic loop: Two copies of the decohered ℤ₂ toric code plus a product state, followed by gauging the weak ℤ₂ layer-swap symmetry with a vertex-local channel built from the Gauss operator G_v, yields an imTO whose strong symmetry is a gauge-invariant non-invertible operator with fermionic loop statistics. This corresponds to E = ℤ₂ × ℤ₂ with fermionic-loop anomalies in both factors, trivial N, and H = ℤ₂ acting by the swap. The non-Abelian fermionic loop is found in 𝔻₄ gauge theory (dihedral group of order 8).

• FcFl theory — the flagship new example: Stacking a bosonic and a fermionic ℤ₂ toric code and applying the channel built from L̃^A_e and X^A_e S̃^B_e produces an anomalous ℤ₂ 2-form symmetry generated by L^{AB}_f(γ) = ∏ L^A_e (Z^B_e)² and an anomalous ℤ₂ 1-form symmetry generated by S^B_f(Σ). These have nontrivial mutual braiding. The strong operators form a non-degenerate braided fusion 2-category 𝒯, so remote detectability is not violated; the state is nevertheless intrinsically mixed because 𝒯 carries the gravitational anomaly π∫w₂w₃ and is not expected to be realizable in the ground state of any gapped local Hamiltonian in strictly three spatial dimensions.

• A dimensional contrast: In (3+1)D, line operators cannot braid with one another, so they always form a symmetric fusion category that may contain an emergent fermion; non-degeneracy instead concerns how lines are detected by surfaces. Because fusion 2-categories with no nontrivial line operators are necessarily grouplike, every theory encountered can be obtained by gauging symmetries of a parent gauge theory and labelled by group-cohomological data — in contrast to (2+1)D, where modular tensor categories include non-group-theoretic examples such as the Fibonacci anyon theory.

• Stated limitations: The results rest on an ansatz that is motivated but not proven; the analysis is restricted to bosonic theories (though emergent fermions are allowed); the authors do not establish that distinct categorical data necessarily label states inequivalent under two-way channels; and chiral mixed-state phases and multi-loop braidings are not captured. The work is explicitly not a classification.

Methodology in Plain English

The strategy is to build examples first, then abstract.

First, take a known exactly solvable model — a toric code on a cubic lattice, with qubits or ℤ₄ qudits on edges — whose ground state has a known topological order. Second, apply a local noise channel of the form "with probability p do nothing, with probability 1−p apply this local operator," where the operator is chosen to create the excitation you want to proliferate: a fermion, a fermionic loop, or a combination across two layers. The choice of operator is delicate; it must commute with the symmetry you want to keep strong, which is why "framed" versions of the string operators are used.

Third, analyze the resulting density matrix. At the maximally decohered point p = 1/2, because the channel is a Pauli channel and the initial state is a stabilizer state, the calculation reduces to twirling: stabilizers that anticommute with some Kraus operator are projected out, and those that commute survive. This gives an explicit, zero-correlation-length characterization of the mixed state, plus a clean read-off of which symmetries remain strong (invariant under the channel) and which degrade to weak (invariant only up to the channel acting on both sides).

Fourth, to obtain more elaborate theories, gauge a 0-form symmetry of the mixed state — either strongly, or weakly (so-called "classical gauging," which amounts to averaging over gauge-field configurations as a classical ensemble). Gauging the weak ℤ₂ swap of two decohered layers is what produces the non-Abelian fermionic loop.

Finally, the authors abstract the pattern: start with an Abelian gauge theory with group E, proliferate loops carrying an anomaly β, then gauge a 0-form symmetry G with a normal subgroup N acting strongly and H = G/N acting weakly. They then show this procedure reproduces exactly the data parametrizing braided fusion 2-categories. Pure-state topological orders correspond to the special case of faithful grading, where equivariantization restores the line operators and preserves non-degeneracy; mixed states unlock non-faithful gradings.

Why This Matters

Impact on research. The paper extends the program of classifying mixed-state topological order from (2+1)D to (3+1)D, where the relevant mathematics shifts from braided fusion categories to braided fusion 2-categories. It identifies a genuinely new mechanism for intrinsic mixedness — non-degenerate braiding plus a gravitational anomaly — that has no lower-dimensional analogue, which changes what one should look for when asking whether a mixed state is fundamentally non-pure-state-realizable. It also supplies a constructive bridge: rather than classifying abstractly, it gives a recipe (proliferate, then gauge strongly and weakly) that produces the categorical data physically, with lattice models for each step.

Real-world applications (near- to long-term):

• Understanding how logical information in topological quantum memories degrades under noise, and where decoding thresholds lie — the paper notes that at weak decoherence the fermions should be recoverable by a quasi-local decoder, with a decoding transition at some finite threshold separating mTO from imTO.
• Designing decoders that exploit the strong symmetries and anomalies surviving decoherence, since anomalous strong symmetries protect robust classical information (the paper notes fermion-loop logicals survive as classical information while conjugate membrane logicals are completely randomized).
• Benchmarking and characterizing noisy intermediate-scale quantum simulators on which topological states have already been prepared, by giving invariant-based diagnostics (generalized statistics invariants evaluated directly on the decohered state).
• Informing error-correction architecture choices for (3+1)D topological codes, where loop-like excitations and their self-statistics introduce failure modes absent in 2D.

Industry relevance. Any hardware roadmap targeting topological protection — superconducting qubit arrays, neutral-atom simulators, trapped ions — must contend with the fact that decoherence can produce phases that are not merely "worse versions" of the target state but qualitatively different phases with their own protected data and their own obstructions. The paper's message that a purely bosonic (3+1)D code can decohere into a nontrivial imTO is directly relevant to architecture selection, since it means bosonic-only designs do not insulate a device from anomalous mixed-state behavior.

Future Directions

1. Proving or refuting the central ansatz. The claim that topological operators of a (3+1)D mixed-state topological order form a braided fusion 2-category that need not be non-degenerate is motivated but not proven. A proof, or a counterexample, would settle the foundations of the framework.

2. Moving beyond bosonic theories and toward a classification. The authors restrict to bosonic theories (allowing emergent fermions) and explicitly disclaim a classification. Extending to fermionic theories, capturing chiral mixed-state phases, and incorporating multi-loop braidings are all named as open.

3. Establishing inequivalence of categorical data. Distinct categorical data are not shown to label states inequivalent under two-way channel connectivity, nor under the finer finite-Markov-length condition. Sharpening this would turn the construction into a genuine classification tool.

4. Locating the decoding transition quantitatively. The paper argues heuristically that for p sufficiently close to 1 the fermionic toric code state remains two-way connected to the parent pure state, with an imTO phase separated by a decoding transition at finite threshold. Determining that threshold for the concrete lattice models, and understanding what order parameter detects it, is a natural next step.

5. Realizing the FcFl theory and the non-Abelian fermionic loop on quantum simulators. Since the models are stabilizer-based and the channels are Pauli, these constructions are plausibly accessible to existing programmable platforms — a natural experimental follow-up.

Target Audience

Condensed matter theorists and mathematical physicists working on topological phases, higher-form symmetries, and topological quantum field theory will get the most from this paper, particularly those already fluent in fusion categories and anomaly inflow. Quantum information researchers focused on error correction and decoherence thresholds in topological codes will find the lattice models and the mTO/imTO distinction operationally useful, though the categorical sections will require background reading. Mathematicians interested in fusion 2-categories will find a physical realization theorem for the structures they study. Experimentalists building topological-order simulators will benefit from the concrete stabilizer constructions but will need to work through substantial formalism to extract protocol-level prescriptions.

Authors’ abstract

We construct large families of mixed-state topological orders in (3+1)D, including intrinsically mixed-state topological orders. We realize these theories by relating the data parameterizing braided fusion 2-categories to the strong higher-form symmetries and their anomalies that characterize mixed-state topological order. We focus here on bosonic theories, which include theories which admit an emergent fermion in the spectrum of line operators. Our general framework is exemplified on the lattice, where we present concrete models and identify how the data of braided fusion 2-categories are physically encoded. We then formalize this data and show that every braided fusion 2-category arises from a general construction (decoherence followed by strong and weak gauging) as the category of topological operators of a (3+1)D mixed state, with intrinsically mixed-state topological orders distinguished by the presence of strong higher-form symmetries whose anomalies cannot be realized in any (3+1)D gapped ground state. Notably, we show that (3+1)D admits a new mechanism with no (2+1)D counterpart: intrinsically mixed-state order whose topological operators braid non-degenerately, but which carries a gravitational anomaly.