---
title: Fusion Category Symmetry
url: https://www.emergentmind.com/topics/fusion-category-symmetry
type: topic
---

# Fusion Category Symmetry

Fusion category symmetry is a generalized symmetry structure in which topological defects are not organized by a group of invertible operators but by a fusion category. In \(1+1\) dimensions, the basic symmetry defects are topological line operators labeled by simple objects of a unitary fusion category \(\mathcal A\); their fusion, duality, and associativity are encoded by fusion multiplicities, dual objects, and \(F\)-symbols. This framework subsumes ordinary finite-group symmetry as a special case, accommodates non-invertible defects with quantum dimensions larger than one, and admits a bulk interpretation through the Drinfeld center and symmetry topological field theory. The modern literature treats fusion category symmetry simultaneously as an algebraic structure, an anomaly-detection problem, a classification principle for gapped phases and SPT phases, and a source of explicit lattice constructions and higher-categorical generalizations [1912.02817], [2304.01262].

## 1. Defect data and categorical structure

In \(1+1\) dimensions, a fusion category symmetry is specified by a unitary fusion category \(\mathcal A\) whose simple objects \(a\in\mathrm{Ob}(\mathcal A)\) label topological line defects. Their fusion is
\[
a\otimes b \;\cong\; \bigoplus_c N_{ab}^c\,c,
\]
with \(N_{ab}^c\in\mathbb N\), a tensor unit \(1\), and duals \(\bar a\) such that \(1\) appears in \(a\otimes\bar a\). Associativity is controlled by \(F\)-symbols \(F^{abc}_d\), which satisfy the pentagon equation. Quantum dimensions \(d_a\) are determined by
\[
d_a d_b=\sum_c N_{ab}^c\,d_c.
\]
Invertible simple objects satisfy \(d_a=1\), while genuinely non-invertible defects have \(d_a>1\). If the category is braided, it also has \(R\)-symbols obeying hexagon equations [2304.01262].

This recovers ordinary finite-group symmetry when the category is group-like, for example \(\mathrm{Vec}_G\), but the categorical formulation is strictly broader. In the defect-line language of \(1+1\)-dimensional QFT, the operators \(L_a\) obey
\[
L_a\times L_b=\sum_c N_{ab}^c\,L_c,
\]
and consistency of different fusion orderings is precisely the pentagon constraint on the \(F\)-symbols. The same structure appears in rational CFT, anyon-chain models, and topological boundary theories [1912.02817].

The \(F\)-symbols are the categorical analogue of \(6j\)-symbols. Under minimal assumptions—\(k\)-linearity, semisimplicity, finitely many simple objects, and simple unit—these \(6j\)-symbols enjoy tetrahedral \(S_4\)-symmetry, and this result does not require pivotality, braiding, or unitarity. The formulation via veined fusion categories isolates the associator data while remaining monoidally equivalent to the original fusion category [2106.16186].

The framework is not restricted to isolated finite examples. In \(1+1\)-dimensional CFT, continuous families of topological defect lines can also occur; in the \(c=1\) analysis, continuous fusion category symmetries are related by a Noether theorem to non-local conserved currents. This places categorical symmetry alongside both discrete and continuous symmetry structures, but with the crucial distinction that the defect algebra is generally non-invertible [2106.12577].

## 2. Anomalies, fiber functors, and the bulk Drinfeld center

A central distinction is between anomaly-free and anomalous fusion category symmetry. Physically, \(\mathcal A\) is non-anomalous if there exists a trivially gapped \(1+1\)-dimensional phase with a unique ground state on the circle whose topological lines realize exactly \(\mathcal A\). Mathematically, this is equivalent to the existence of a fiber functor
\[
F:\mathcal A\to \mathrm{Vec}_{\mathbb C},
\]
or, in Hilbert-space form, to \(\mathrm{Hilb}\). In the group-like case \(\mathcal A=\mathrm{Vec}_G^\omega\), the obstruction is the familiar \([\omega]\in H^3(G,U(1))\); in the non-invertible case the obstruction is categorical rather than purely group-cohomological [2304.01262].

The bulk object governing anomalies is the Drinfeld center \(Z(\mathcal A)\). Its simple objects are pairs \((a,u)\), with \(a\in\mathrm{Ob}(\mathcal A)\) and half-braiding
\[
u_X:a\otimes X\to X\otimes a
\qquad \forall X\in\mathcal A,
\]
satisfying the standard compatibility condition \(u_{X\otimes Y}=(1_X\otimes u_Y)(u_X\otimes1_Y)\). The center is a braided fusion category and describes the \((2+1)\)-dimensional topological order, or symmetry TFT, whose gapped boundaries classify all \(1+1\)-dimensional gapped phases with \(\mathcal A\)-symmetry, including symmetry-breaking phases [2304.01262].

Gapped boundaries of \(Z(\mathcal A)\) are in one-to-one correspondence with Lagrangian algebras \(L\subset Z(\mathcal A)\). Such an algebra is a commutative separable algebra object with
\[
\dim L=\sqrt{\dim Z(\mathcal A)}.
\]
There is always a canonical electric Lagrangian algebra \(L_e\), corresponding to a fully symmetry-broken boundary. The anomaly criterion is the existence of a distinct magnetic Lagrangian algebra \(L_m\) disjoint from \(L_e\), equivalently \(L_e\cap L_m=\{1\}\). The symmetry is anomaly-free exactly when such an \(L_m\) exists [2304.01262].

This criterion refines the boundary classification of \(1+1\)-dimensional gapped phases by module categories. In the Turaev–Viro/Levin–Wen picture, gapped phases with symmetry \(\mathcal A\) are classified by module categories over \(\mathcal A\), and symmetry-preserving SPT phases correspond to the single-simple-object case, equivalently to fiber functors. The boundary/bulk relation is therefore not auxiliary: it is the mechanism by which ’t Hooft anomalies become topological obstructions in one higher dimension [1912.02817].

## 3. Gapped phases, SPT phases, and symmetry breaking

For non-anomalous fusion category symmetry, the classification of \(1+1\)-dimensional gapped phases parallels but extends the finite-group story. Indecomposable semisimple \(\mathcal C\)-module categories classify gapped phases, with symmetry-preserving SPT phases corresponding to simple algebraic data and symmetry-breaking phases corresponding to module categories with multiple simples. In the Hopf-algebraic realization \(\mathcal C\simeq \mathrm{Rep}(H)\), these phases are classified by \(H\)-simple left \(H\)-comodule algebras up to equivariant Morita equivalence [2110.12882].

A recurring point in recent work is that an anomaly-free fusion-category symmetry is not specified by the fusion category alone. The ultraviolet datum is the pair \((\mathcal C,f)\), where \(f:\mathcal C\to\mathrm{Hilb}\) is a unitary fiber functor. This pair determines an on-site kinematical realization, a charge category \(\mathcal C_M^\vee\), and a trivial phase. In this formulation, SPT phases are classified by Q-systems in the charge category whose image under the forgetful functor is a full matrix algebra \(M_n(\mathbb C)\). Fixed-point MPS/TN models built from such Q-systems give explicit microscopic realizations of non-invertible SPT phases [2412.20546].

The example \(\mathcal C=\mathrm{Rep}^\dagger(D_8)\) makes the structure concrete. The paper constructs all three SPT phases for this symmetry category, including the trivial phase, identifies the corresponding Q-systems in \(\mathrm{Hilb}_{D_8}\), and shows that the three fiber functors form an \(S_3\)-torsor under \(\mathrm{Aut}_\otimes(\mathcal C)\simeq S_3\). The associated lattice dualities realize the categorical statement that the three phases are related by \(S_3\)-duality [2412.20546].

Time-reversal enlarges the classification problem from oriented to unoriented TQFT. In the SPT regime, \(1+1\)-dimensional bosonic phases with \( \mathbb Z_2^T\times \mathcal C\) symmetry are classified by equivalence classes of quintuples
\[
(Z,M,i,s,\phi),
\]
where \((Z,M,i)\) is a fiber functor, \(s\in\mathbb Z_2\) is the purely time-reversal sign, and \(\phi_x:V_x\to V_{x^*}\) is compatible with duality and fusion. This reproduces the group-cohomological classification in the special case \(\mathcal C=\mathrm{Vec}_G\), while extending it to non-invertible and self-dual settings such as Tambara–Yamagami symmetry [2103.15588].

## 4. Lattice realizations and operator-algebraic formulations

Fusion category symmetry admits several exact lattice realizations. One approach starts from a finite-dimensional semisimple Hopf algebra \(H\) with \(\mathcal C=\mathrm{Rep}(H)\) and an \(H\)-simple left \(H\)-comodule algebra \(K\). The resulting state-sum TQFT and commuting-projector Hamiltonian realize all \(1+1\)-dimensional gapped phases with non-anomalous symmetry \(\mathcal C\). The boundary-condition category is the \( \mathcal C\)-module category \({}_K\mathcal M\), while the Hamiltonian
\[
H=-\sum_i h_{i,i+1},
\qquad
h_{i,i+1}=\Delta_K\circ m_K,
\]
has the same ground-state projector as the state-sum cylinder [2110.12882].

A complementary route constructs the symmetry operators as matrix-product operators from the Hopf algebra reconstructed by the fiber functor. For a chain with local Hilbert space \(V_i\simeq f(A)\), the virtual bond spaces of the MPOs are \(f(s)\) for each simple \(s\in\mathrm{Irr}(\mathcal C)\), and dualizability forces
\[
\dim f(s)=d_s.
\]
The coend
\[
H \cong \bigoplus_{s\in\mathrm{Irr}(\mathcal C)} f(s)\otimes f(s)^*
\]
produces the charge category and the local MPO tensors \(W_s^\Lambda\); Q-systems in the charge category then furnish fixed-point commuting-projector models [2412.20546].

The operator-algebraic formulation recasts the same structure in the thermodynamic limit. Given a physical boundary subalgebra \(B\subseteq A\) of a quasi-local algebra, one obtains a canonical fusion category \(\mathcal C\) acting by bimodules, together with locality-preserving quantum channels \(\Phi_X:A\to A\) for each simple \(X\in\mathrm{Irr}(\mathcal C)\). The invariant algebra is
\[
B=\{\,a\in A:\Phi_X(a)=a\ \forall X\,\}.
\]
This framework yields a categorical Lieb–Schultz–Mattis theorem: if a pure symmetric state is topological, the corresponding algebra in \(Z(\mathcal C)\) must be Lagrangian; if \(\mathcal C\) has no fiber functor, then any pure symmetric state is necessarily gapless [2507.05185].

These realizability results come with sharp constraints. If a fusion category acts by bounded-spread bimodules on a tensor-product spin chain, then all quantum dimensions must be integers; conversely, every integral fusion category admits such a bounded-spread action. A strictly on-site action exists if and only if the category admits a fiber functor [2507.05185]. Refining the realization by QCAs extends the construction to all weakly integral fusion categories on tensor-product Hilbert spaces, with QCA and symmetry-operator indices fixed by the categorical data up to redefinitions of the symmetry operators [2605.15194]. After stabilization by infinite-dimensional ancilla spaces, any unitary fusion category can be realized on a tensor product of infinite-dimensional Hilbert spaces, and any two anyon chains with the same symmetry category become stably equivalent under a locality-preserving unitary [2605.21327].

## 5. Canonical examples and physical phenomena

Tambara–Yamagami categories are the standard family of examples. For
\[
\mathrm{TY}_G^{\chi,\varepsilon},
\]
the simple objects are \(G\cup\{\sigma\}\), with
\[
g\otimes h=gh,\qquad g\otimes \sigma=\sigma\otimes g=\sigma,\qquad
\sigma\otimes\sigma=\bigoplus_{g\in G} g.
\]
Their anomaly structure is particularly explicit. For \(G\simeq \mathbb Z_N\times \mathbb Z_N\), the magnetic Lagrangian algebra exists exactly in the cases stated by the fiber-functor classification: for \(|G|\) odd, anomaly-freedom requires \(\varepsilon=+1\) and \(\chi\) hyperbolic; concretely for \(G=\mathbb Z_N\times\mathbb Z_N\), \(L_m\) exists exactly when \(\chi\) is either the off-diagonal bicharacter or the diagonal bicharacter with \(-1\) a quadratic residue mod \(N\), and \(\varepsilon\) must be \(+1\) [2304.01262].

The Ising category is the minimal non-invertible example, with simple objects \(1,\psi,\sigma\), dimensions \(d_1=d_\psi=1\), \(d_\sigma=\sqrt2\), and fusion
\[
\psi\otimes\psi=1,\qquad \psi\otimes\sigma=\sigma,\qquad \sigma\otimes\sigma=1+\psi.
\]
Because \(d_\sigma=\sqrt2\), it does not admit a trivially gapped symmetry-preserving phase, and this is reflected both in continuum and lattice analyses. The \(c=1\) CFT study identifies Ising- and TY-type duality lines, continuous fusion symmetries, and RG constraints from anomaly matching [2106.12577]. The \(2026\) lattice study finds three phases in an Ising fusion-category chain: a symmetric critical phase in the usual critical Ising universality class, a categorical ferromagnetic phase with threefold ground-state degeneracy and fully broken Ising fusion symmetry, and a categorical antiferromagnetic phase that breaks translation and part of the symmetry while remaining critical; no gapped symmetric phase appears [2604.20201].

These examples correct a common misconception: non-invertibility does not by itself imply anomaly. Tambara–Yamagami categories can be anomaly-free for specific \((\chi,\varepsilon)\), and \(\mathrm{Rep}^\dagger(D_8)\) admits explicit non-invertible SPT phases [2304.01262], [2412.20546]. The correct criterion is the fiber-functor or magnetic-Lagrangian-algebra condition, not the mere presence of a simple object with \(d_a>1\).

Explicit weak-Hopf realizations further enlarge the example set. A weak Hopf boundary tube algebra for the Ising category yields an exactly solvable cluster-state Hamiltonian whose cocommutative symmetry operators satisfy the Ising fusion relations on both open and closed chains [2508.02424]. A related construction reconstructs the Haagerup weak Hopf algebra \(H_3\), realizes Haagerup fusion category symmetry on a tensor-product chain, and produces a weak-Hopf MPS ground state described as a natural generalization of the cluster state [2412.19657].

## 6. Higher-categorical extensions and related directions

Fusion category symmetry in \(1+1\) dimensions is part of a broader hierarchy of categorical symmetries. One systematic upgrade replaces a braided fusion category \(\mathcal B\) by its condensation completion \(\Sigma\mathcal B\), a fusion \(2\)-category whose objects are separable algebras in \(\mathcal B\), whose \(1\)-morphisms are bimodules, and whose higher associativity data are encoded in \(10j\)-symbols. The worked example \(\Sigma\,\mathrm{sVec}\) makes these data explicit and computable [2312.15947].

In \(2+1\) dimensions, categorical symmetry is described by fusion \(2\)-categories. One construction starts from a fusion \(2\)-category \(\mathcal C\) and produces fusion surface models: \(2+1\)-dimensional commuting-projector systems with topological surface and line operators acting as non-invertible symmetries. This includes group anomalies, one-form symmetries, non-invertible one-form symmetries, and the Kitaev honeycomb model as special cases [2305.05774]. A more general classification result argues that finite spherical fusion \(2\)-categories arise from triples \((\mathcal C,G,\omega)\) encoded by \(\omega\)-twisted \(G\)-crossed extensions of modular categories, thereby organizing categorical symmetries in oriented \(2+1\)-dimensional field theory through anomaly cancellation in \(3+1\)-dimensional Dijkgraaf–Witten theory [2408.13931].

In \(3+1\) dimensions, self-duality defects lead to fusion \(3\)-categories. The generalized Tambara–Yamagami fusion \(3\)-categories studied in the duality-defect literature are classified by graded extensions of \(\mathbf{3Vect}(A[1])\), with Brauer–Picard and Picard \(4\)-groupoids controlling the extension data. Their centers, sylleptic structures, and generalized Witt groups provide the higher-dimensional analogue of the center/Lagrangian-algebra viewpoint familiar from \(1+1\)-dimensional fusion category symmetry [2408.13302].

Related extensions include \(G\)-graded fusion-category lattice models interpolating between anyon chains and SPT-edge models [2301.06416], and fermionization procedures in which a bosonic \(\mathrm{Rep}(H)\) symmetry becomes a superfusion-category symmetry \(\mathrm{SRep}(\mathcal H^u)\) for a weak Hopf superalgebra \(\mathcal H^u\) determined by a non-anomalous \(\mathbb Z_2\) subgroup [2206.13159]. Taken together, these developments indicate that fusion category symmetry is best regarded not as an isolated \(1+1\)-dimensional phenomenon but as the first layer of a higher-categorical hierarchy linking defects, anomalies, topological phases, and lattice realizations across dimensions.

Source: https://www.emergentmind.com/topics/fusion-category-symmetry