Generalized Symmetries in

Condensed Matter
John McGreevy
Department of Physics, University of California San Diego, La Jolla, California
92093, USA
arXiv:2204.03045v3 [cond-mat.str-el] 4 Jan 2025

Submitted to Annual Review of Keywords

Condensed Matter Physics 2022. 00:1–38
Copyright © 2022 by Annual Reviews. symmetry, quantum, spontaneous symmetry breaking, low-energy
All rights reserved effective field theory, quantum phases of matter

Abstract

Recent advances in our understanding of symmetry in quantum many-

body systems offer the possibility of a generalized Landau paradigm
that encompasses all equilibrium phases of matter. This is a brief and
elementary review of some of these developments.

1
Contents
1. Extending the Landau Paradigm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Higher-form symmetries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1. Physics examples of higher-form symmetries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2. Spontaneous symmetry breaking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3. Topological order as SSB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4. Photon as Goldstone boson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.5. Effects of IR fluctuations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.6. Robustness of higher-form symmetries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.7. Mean field theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3. Anomalies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.1. SPT phases and anomalies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2. Anomalies of higher-form symmetries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.3. SPT phases of higher-form symmetries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4. Subsystem symmetries and fracton phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5. Categorical symmetries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
6. Gapless states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
6.1. Critical points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
6.2. Gapless phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
7. Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
A. Spontaneous symmetry breaking and long-range order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
B. SSB of higher-form symmetry without topological order, a confession . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

1. Extending the Landau Paradigm

If you have been to a condensed-matter talk in the past few decades, you have seen the
beating that Landau has been taking. The speaker begins by saying that Landau told us
that states of matter are classified by the symmetries they break. After showing a picture
of a donut, the speaker explains that in this talk, in contrast, they will discuss a state of
matter that goes beyond Landau’s limited conception of the world.
Having given such talks myself, I think it is extremely interesting that, in fact, with
modern generalizations of our understanding of symmetry, it may be possible to incorporate
all known equilibrium phases of matter into a suitably extended version of the Landau
Paradigm. Let me attempt to paraphrase the Landau Paradigm:

1. Phases of matter should be labelled by how they represent their symmetries, in par-
ticular whether they are spontaneously broken or not.
2. The degrees of freedom at a critical point are the fluctuations of the order parameter.

A significant corollary of assertion 1 is that gapless degrees of freedom, or groundstate

degeneracy, in a phase, should be swept out by a symmetry. That is, they should arise as
Goldstone modes for some spontaneously broken symmetry.
Beyond its conceptual utility, this perspective has a weaponization, in the form of
Landau-Ginzburg theory, in terms of which we may find representative states, understand
gross phase structure, and, when suitably augmented by the renormalization group (RG),
even quantitatively describe phase transitions.
Indeed there are many apparent exceptions to the Landau Paradigm. Let us focus first

2 McGreevy
on apparent exceptions to item 1. As a preview, exceptions that are only apparent include:

• Topologically-ordered states. These are phases of matter distinguished from the

trivial phase by something other than a local order parameter (1, 2). Symptoms
include a groundstate degeneracy that depends on the topology of space, and anyons,
excitations that cannot be created by any local operator. Real examples found so far
include fractional quantum Hall states, as well as gapped spin liquids.
• Other deconfined states of gauge theory. This category includes gapless spin
liquids such as spinon Fermi surface or Dirac spin liquids (most candidate spin liquid
materials are gapless). Another very visible manifestation of such a state is the photon
phase of quantum electrodynamics in which our vacuum lives.
• Fracton phases. Gapped fracton phases are a special case of topological order,
where there are excitations that not only cannot be created by any local operator,
but cannot be moved by any local operator.
• Topological insulators. Here we can include both free-fermion states with topologi-
cally non-trivial bandstructure, as well as interacting symmetry-protected topological
(SPT) phases.
• Landau Fermi liquid.

Conventions. L is the linear system size. D = d + 1 is the number of spacetime

dimensions. I’ll denote the dimension of a manifold or the degree of a form by a subscript
or superscript. I will use fancy upper-case letters (like Aµ ) for background gauge fields and
lower-case letters (like aµ ) for dynamical gauge fields. I will sometimes use G(p) to denote
a p-form symmetry with group G.
Brief non-symmetry-based accounting of gapped phases. A useful definition of a
gapped phase of matter is an equivalence class of gapped groundstates of local Hamiltonians,
in the thermodynamic limit1 . Two groundstates are considered equivalent if they are related
by adiabatic evolution (for a time of order L0 ) combined with inclusion or removal of product
states. That is, there is a path between the two Hamiltonians along which the gap does not
close (see Fig. 1, left).
This definition poses a difficulty for checking that two Hamiltonians represent distinct
phases: we cannot check all possible paths between them. A crucial role is therefore played
by universal properties of a phase – quantities, such as integers, that cannot change smoothly
within a phase, and therefore can only vary across phase boundaries. A good example of such
a topological invariant is the groundstate degeneracy, which is certainly an integer. A phase
of matter that spontaneously breaks a discrete symmetry G has a groundstate degeneracy
|G|, the order of the group (see Fig. 1, right). This is a topological distinction from the
trivial paramagnetic phase, which has a unique groundstate and a representative that is
a product state with no entanglement at all. In this sense, even spontaneous symmetry
breaking (SSB) is a topological phenomenon.
Non-trivial phases can be divided into two classes: those with topological order and
those without. One way to define topological order (1) is a phase with localized excitations

1 We should pause to comment on the meaning of ‘gapped’. We allow for a stable ground-

state subspace, which becomes degenerate in the thermodynamic limit. ‘Stable’ means that the
degeneracy persists under arbitrary small perturbations of the Hamiltonian, and requires that the
groundstates are not related by the action of local operators. In d spatial dimensions, the logarithm
of the number of such states can grow as quickly as Ld−1 (3) in fracton models.

www.annualreviews.org • Generalized Symmetries 3

 Figure 1
Left: A schematic illustration of the definition of gapped phases of matter. Two distinct phases
are separated in the space of local Hamiltonians by a wall of gaplessness, the codimension-one
locus where the gap closes. Here HA ≃ HA′ . Right: The groundstate degeneracy, Ngs , for
example as swept out by a spontaneously broken global symmetry, is an example of a topological
invariant that can label a phase.

that cannot be created by any local operator. In 2+1 dimensions, such particle excitations
are called anyons; they can be created in pairs by an open-string operator. On a space with
a non-contractible curve C, new groundstates can be made by acting with the operator
that transports an anyon around C. These groundstates are locally indistinguishable, in the
following sense. If and are two such groundstates, then

Ox = 0, 1.

for all local operators Ox . (The picture in the kets is a cartoon of two of the ground-
states on the 2-torus.) A final symptom is the existence of long-range entanglement in the
groundstate; a review focussing on this aspect is (4).
An interesting special case of topologically ordered states is fracton phases (5, 6). A
fracton phase has excitations that cannot be moved by any local operator (perhaps only in
some directions of space). This is a strictly stronger condition than topological order, since
an excitation can effectively be moved by annihilating it and creating it again elsewhere.
Such phases (with a gap) exist in 3+1 dimensions (and higher). A consequence of the
defining property is a groundstate degeneracy whose logarithm grows linearly with system
size, and a subleading linear term in the scaling of the entanglement entropy of a region
with the size of the region.
Even without topological order, there can be phases distinct from the trivial phase. One
way in which they can be distinguished is by what happens if we put them on a space with
boundary, so that there is a spatial interface with the trivial phase. A very rough (and
not entirely correct) idea is that if the gap must close along the path to the trivial phase,
then the coupling must pass through the wall of gap-closing at the edge of the sample.
Phases that are distinguished in this way include integer quantum Hall states, topological
insulators, and, more generally, symmetry-protected topological (SPT) phases such as the
Haldane phase of the spin-one chain, or polyacetylene.
It seems that all of these examples transcend the Landau Paradigm. My goal here is not
to use the Landau Paradigm as a straw man, but rather to pursue it in earnest. The idea
is that by suitably refining and generalizing our notions of symmetry, we can incorporate

4 McGreevy
all of these ‘beyond-Landau’ examples into a Generalized Landau Paradigm. There are two
crucial ingredients, which work in concert: anomalies and generalized symmetries.
In this article, I am speaking of actual symmetries of physical systems, sometimes
called ‘global symmetries’. They act on the Hilbert space and take one state to another.
In contrast, there is no such thing as ‘gauge symmetry’. In a gauge theory, the gauge
invariance is a redundancy of a particular description of the system, and is not preserved by
relabelling degrees of freedom. For example, dualities (equivalences of physical observables
at low energies) often relate a gauge theory with one gauge group to a gauge theory with
a distinct gauge group. A familiar example in condensed matter physics is the duality
between the XY model and the abelian Higgs model in 2+1 dimensions (7, 8), but there
are many others, e.g. (9). This complaint about terminology hides an abyss of human
ignorance: If someone hands you a piece of rock and asks whether its low-energy physics
is described by some phase of a gauge theory, how will you tell? It is certainly true that
phases realizable by gauge theory go beyond other constructions with only short-ranged
entanglement; this begs for a characterization of these phases that transcends a description
in terms of redundancies. Higher-form symmetries offer such a characterization for some
such phases.
I want to highlight early attempts to understand topological order (10, 11), and the
gaplessness of the photon (12) as a consequences of generalized symmetry, as well as early
appearances of generalized symmetries in the string theory literature (13, 14, 15). Other
papers that have explicitly advocated for the utility of a generalized Landau Paradigm
include (16, 17, 18, 19, 20).

2. Higher-form symmetries
The concept of higher-form symmetry that we review here was explained in (21, 16). It is
easiest to introduce using a relativistic notation. Indices µ, ν run over space and time.
Let’s begin by considering the familiar case of a continuous 0-form symmetry. Noether’s
theorem guarantees a conserved current Jµ satisfying ∂ µ Jµ = 0. In the useful language
of differential forms, this is d ⋆ J = 0, where ⋆ is the Hodge duality operation2 . This
R
continuity equation has the consequence that the charge QΣ = Σ ⋆J is independent of
D−1
the choice of time-slice Σ. (Σ here is a closed d-dimensional surface, of codimension one in
spacetime.) Notice that this is a topological condition. QΣ commutes with the Hamiltonian,
the generator of time translations, and therefore so does the unitary operator Uα = eiαQ ,
which we call the symmetry operator3 .
If the charge is carried by particles, QΣ counts the number of particle worldlines piercing
the surface Σ (as in Fig. 2, left), and the conservation law Q̇ = 0 says that charged particle
worldlines cannot end except on charged operators. If instead of a U(1) symmetry, we only
had a discrete Zp symmetry we could simply restrict α ∈ {0, 2π/p, 4π/p...(p − 1)2π/p} in
the symmetry operator Uα . In that case, particles can disappear in groups of p.
Objects charged under a 0-form symmetry are created by local operators. Local opera-
tors transform under the symmetry by O(x) → Uα O(x)Uα† = eiqα O(x), dα = 0, where q is

2 The Hodge dual of a p-form ω on a D-dimensional space with metric g

p µν has components
√ µ ···µ
(⋆ωp )µ1 ···µD−p = detgϵµ1 ···µD ωp D−p+1 D , where indices are raised with the inverse metric
g µν and ϵµ1 ···µD is the antisymmetric Levi-Civita symbol.
3 Throughout I will assume that the normalization is such that Q ∈ Z, so that α ≡ α + 2π.

www.annualreviews.org • Generalized Symmetries 5

 the charge of the operator. The infinitesimal version is: δO(x) = i[Q, O(x)] = iqO(x).

Figure 2
Left: In the case of an ordinary 0-form symmetry, the charge is integrated over a codimension-one
slice of spacetime ΣD−1 , often a slice of constant time. All the particle worldlines (blue curves)
must pass through this hypersurface. Right: The charge of a 1-form symmetry is integrated over a
codimension-two locus of spacetime ΣD−2 (a string in the case of D = 2 + 1). This surface
intersects the worldsheets of strings (blue sheet).

Now let us consider a continuous 1-form symmetry. This means that there is a conserved
current which has two indices, and is completely antisymmetric:

Jµν = −Jνµ with ∂ µ Jµν = 0. 2.

We can regard J as a 2-form and write the conservation law Eq. 2 as d ⋆ J = 0. As a

consequence, for any closed codimension-two locus in spacetime ΣD−2 , the quantity QΣ =
R
ΣD−2
⋆J depends only on the topological class of Σ. The analog of the symmetry operator
is the unitary operator
Uα (Σ) = eiαQΣ . 3.
Notice that reversing the orientation of Σ produces the adjoint of U : Uα (−Σ) = Uα† (Σ).
The charge QΣ in the 1-form case counts the number of charged string worldsheets
intersecting the surface Σ (as in Fig. 2, right). The conservation law Eq. 2 then says that
charged string worldsheets cannot end except on charged operators. The objects charged
under a 1-form symmetry are loop operators, W (C). Fixing a constant-time slice MD−1 ,
such a loop operator transforms as
H
W (C) → Uα (Σ)W (C)Uα† (Σ) = eiα C ΓΣ
W (C), dΓΣ = 0. 4.

Here ΣD−2 ⊂ MD−1 is any closed (D − 2)-manifold, and ΓΣ is its Poincaré dual in MD−1 ,
η (D−2) ∧ ΓΣ = Σ η (D−2) for all η; dΓΣ = 0 because Σ has no
R R
in the sense that M
D−1 D−2
boundary. The infinitesimal version of this transformation law is

δW (C) = i[QΣ , W (C)] = iq# (Σ, C) W (C), 5.

where #(Σ, C) is the intersection number in M 4 .

4 Above I have written the expression for the transformation as U (Σ)W (C)U † (Σ). This operator

6 McGreevy
 In the case of a discrete 1-form symmetry, there is no current, but the symmetry operator
Uα (Σ) is still topological. If the 1-form symmetry group is Zp , strings can disappear or end
only in groups of p.
For general integer p ≥ −1, a p-form symmetry means the existence of topological
operators Uα (ΣD−p−1 ) labelled by a group element α and a closed codimension-(p + 1)
submanifold of spacetime5 . For coincident submanifolds, these operators satisfy the “fusion
rule” Uα (Σ)Uβ (Σ) = Uα+β (Σ). The operators charged under a p-form symmetry are sup-
ported on p-dimensional loci, and create p-brane excitations. The conservation law asserts
that the (p + 1)-dimensional worldvolume of these excitations will not have boundaries.
For p ≥ 1, the symmetry operators commute with each other – higher-form symmetries
are abelian (16). To see this, consider a path integral representation of an expectation value
with two symmetry operators U (Σ1 )U (Σ2 ) inserted on the same time slice t. The ordering
of the operators can be specified in the path integral by shifting the left one to a slightly
later time t + ϵ. If p ≥ 1, then Σ1,2 have codimension larger than one, and their locations
can be continuously deformed to reverse their order.

2.1. Physics examples of higher-form symmetries

• Maxwell theory in D = 3 + 1 with electric charges but no magnetic charges has
µν 1 µνρσ
a continuous 1-form symmetry with current J(m) = 4π ϵ Fρσ ≡ 2π 1
(dÃ)µν . The
µν
statement that this current is conserved ∇µ J(m) = 0 is the Bianchi identity expressing
(m) iα R
the absence of magnetic charge. The symmetry operator is Uα (Σ) = e 2π Σ F . The
R
fact that the charge operator Σ F depends only on the topological class of Σ is
the magnetic gauss law – when Σ is contractible, it counts the number of magnetic
monopoles inside. This symmetry shifts the dual gauge fieldH Ã by a flat connection;
the charged line operator is the ’t Hooft line, W (m) [C] = ei C Ã .
In free Maxwell theory without electric charges, there is a second 1-formH current,
J(e) = F whose charged operator is the Wegner-Wilson line W (e) [C] = ei C A . The
i 2α
R
(e) ⋆F
symmetry operator for this ‘electric’ 1-form symmetry is Uα (Σ2 ) = e g2 Σ2 , which
(by canonical commutators) shifts the gauge field A by a flat connection.
• Pure SU(N ) gauge theory or ZN gauge theory or U(1) gauge theory with charge-N
matter has a ZN 1-form symmetry, called the ‘center symmetry’. TheHcharged line
operator is the Wegner-Wilson line in the minimal irrep, W [C] = trP ei C A .
• When we spontaneously break a 0-form U(1) symmetry in d = 2, there is an emergent
1-form U(1) symmetry whose charge counts the winding number of the phase variable
H
φ around an arbitrary closed loop C, Q[C] = C dφ. In d spatial dimensions, this is a
(d − 1)-form symmetry. The charged operator creates a vortex (in d = 2, or a vortex
line or sheet in d > 2). Unlike the examples above, this symmetry is generally not an

ordering is obtained by placing the support of these operators on successive time slices. Since U
is topological, from a spacetime point of view, the same result obtains if instead we deform the
surfaces Σ and −Σ to a single surface S in spacetime that surrounds the locus C, as illustrated here
in cross-section:
Σ
t C = C S 6.
−Σ
The variation of the operator then depends on the linking number of S and C in spacetime.
5 For discussion of p = −1, see (22).

www.annualreviews.org • Generalized Symmetries 7

 exact symmetry of a microscopic Hamiltonian for a superfluid; it is explicitly broken
by the presence of vortex configurations. More on this example and its consequences
for superfluid physics in §3.2.
• There is a sense in which the 3d Ising model has a Z2 1-form symmetry reflecting
the integrity of domain walls between regions of up spins and regions of down spins.
The charged line operator is the Kadanoff-Ceva disorder line (23) – the boundary
of a locus along which the sign of the Ising interaction is reversed (for a review, see
(24)). But because a domain wall is always the boundary of some region, no states
are charged; relatedly, the disorder line is not a local string operator. If we gauge the
Z2 symmetry of the Ising model, the disorder line becomes the Wegner-Wilson line
of the resulting Z2 gauge theory, and this theory has a genuine 1-form symmetry.

2.2. Spontaneous symmetry breaking

Anything we can do with ordinary (0-form) symmetries, we can do with higher-form sym-
metries. In particular, they can be spontaneously broken.
One way to characterize the unbroken phase of a 0-form symmetry is that correlations
of charged operators are short-ranged, meaning that they decay exponentially with the
separation between the operators
D E
O(x)† O(0) ∼ e−m|x| . 7.

A language that will generalize is to regard the two points at which we insert a charged
operator and its conjugate as an S 0 , a zero-dimensional sphere, and the separation between
the points as the size of the sphere. The broken phase for 0-form symmetry can be diagnosed
by long-range correlations:
D E D ED E
O(x)† O(0) = O† (x) O (0) + · · · , 8.

independent of the size of the S 0 .

For a p-form symmetry, the unbroken phase is also when correlations of charged opera-
tors are short-ranged, and decay when the charged object grows. For a 1-form symmetry,
this is when the charged loop operator exhibits an area law:

⟨W (C)⟩ ∼ e−Tp+1 Area(C) , 9.

where Area(C) is the area of the minimal surface bounded by the curve C. In the case of
electricity and magnetism, an area law for W E (C) is the superconducting phase.
The broken phase for a p-form symmetry is signalled by a failure of the expectation
value of the charged operator to decay with size. For a 1-form symmetry, this is when the
charged loop operator exhibits a perimeter law:

⟨W (C)⟩ = e−Tp Perimeter(C) + · · · . 10.

The coefficient Tp can be set to 0 by modifying the definition of W (C) by counterterms

local to C, so Eq. 10 says that a large loop has an expectation value.
The fact that charged operators have long-range correlations means that the generators
of the symmetry act nontrivially on the groundstate. More precisely, ψ is not stationary

8 McGreevy
under the symmetry (SSB) if and only if there exists a charged operator O with ψ O ψ ̸=
0 (long-range order). So, LRO ⇔ SSB. I put a proof of this statement in Appendix A.
SSB of higher-form symmetry has been a fruitful idea. In the next two subsections, I’ll
illustrate the consequences in the case of discrete and continuous symmetries, respectively.

2.3. Topological order as SSB

One definition of topological order is the presence of a groundstate subspace of locally
indistinguishable states, as in Eq. 1. This means that no local operator takes one groundstate
to another; instead the operator that takes one groundstate to another is necessarily an
extended operator. But this is a description of the spontaneous breaking of a higher-form
symmetry (16, 25): the generators of the broken part of the higher-form symmetry commute
with the Hamiltonian (at least at low energies) and act nontrivially on the groundstates6 .
Let’s think about the example of Zp gauge theory (whose solvable limit is the toric
code (28)) in D spacetime dimensions. This is a system with Zp 1-form symmetry with
symmetry operators U (MD−2 ), supported on a (D − 2)-dimensional manifold, and charged
operators V (C1 ), supported on a curve. In terms of the toric code variables, we can be
completely explicit. On each link we have a p-state system on which act the Pauli operators
Z = pk=0 ω k |k⟩⟨k| and X = pk=0 |k + 1⟩⟨k| (where ω ≡ e2πi/p and the arguments of the
P P
Q Q
kets are understood mod p). Then V (C) = ℓ∈C Xℓ , U (M ) = ℓ⊥M Zℓ , where we regard
M as a surface in the dual lattice, and ℓ ⊥ M indicates all links crossed by the surface M .
Their algebra is
2πi mn #(C,M ) n
U m (M )V n (C) = e p V (C)U m (M ) 11.
where #(C, M ) is the intersection number of the curve C with the surface M . This is
the algebra of electric strings and magnetic flux surfaces in Zp gauge theory. Deep in this
gapped phase, H = 0, and there is a description in terms of topological field theory. A
simple realization is BF theory of a 1-form potential a and (D − 2)-form potential b, with
action Z Z
p p
S[b, a] = b ∧ da = dD xϵµ1 ···µD bµ1 ···µD−2 ∂µD−1 aµD 12.
2π D 2π
in terms of which H H
U n (M ) = ein M bD−2
, V m (C) = eim C a
. 13.
The algebra Eq. 11 follows from canonical commutation relations in this gaussian theory.
Since V (C) has a perimeter law in the deconfined phase, the charged objects whose con-
densation breaks the 1-form symmetry are the lines of electric flux.
Another example is the Laughlin fractional quantum Hall states. So far in our discussion
the symmetry operators for a 1-form symmetry with group A form a representation of A
on the 1-cycles of space, Z, i.e. a linear map U : Z → U (1), where the representation
operators commute U (M )U (M ′ ) = U (M ′ )U (M ). This relation can be generalized to allow

6 Addendum in v3: However, SSB of higher-form symmetry by itself is not sufficient to imply

topological order (26). The loophole is the following. Just because two groundstates are related
by the action of a nonlocal operator does not mean that there isn’t some linear combination of
them that are related by a local operator. And in fact there is a counterexample, a state with SSB
of a 1-form symmetry which does not have topological order (27, 26). I have added an appendix
B explaining some details of this counterexample. In the discussion of the toric code below, the
extra assumption is that the charged operators V (C) are also topological. Alternatively, we could
describe the extra assumption as requiring SSB of an anomalous higher-form symmetry.

www.annualreviews.org • Generalized Symmetries 9

 for phases – i.e. a projective representation. Consider a system in D = 2 + 1 with a Zk
1-form symmetry that is realized projectively in the following sense:
2πimn#(C,C ′ )
U m (C)U n (C ′ ) = e k U n (C ′ )U m (C) 14.

where #(C, C ′ ) is the intersection number of the two curves C, C ′ in space. Regarding
U (C) as the holonomy of a charged particle along the loop C, this is the statement that
flux carries charge. Representing this algebra nontrivially gives k groundstates on T 2 . This
k
R
algebra, too, has a Hsimple realization via abelian Chern-Simons theory, S[a] = 4π a ∧ da,
with U m (C) = eim C a .
The algebra in Eq. 14 is a further generalization of 1-form symmetry, in that the group
law is only satisfied up to a phase. As we will discuss in §3, it is an example of a 1-form
symmetry anomaly.
The preceding discussion applies to abelian topological orders. In this context, abelian
means that the algebra of the line operators transporting the anyons forms a group, which
must be abelian by the argument above. In §5 we discuss the further generalization that
incorporates non-abelian topological orders.

2.4. Photon as Goldstone boson

What protects the masslessness of the photon? The case of quantum electrodynamics
(QED) is the most visible version of this question; the same question arises in condensed
matter as: why are there U(1) spin liquid phases, with an emergent photon mode?
Higher-form symmetries provide a satisfying answer to this question (unlike appeals
to gauge invariance, which is an artifact of a particular description): the gaplessness of
the photon can be understood as required by spontaneously-broken U(1) 1-form symmetry
(12, 16, 17, 29), as a generalization of the Goldstone phenomenon.
Here is a perspective on the zero-form version of the Goldstone theorem. Given a
continuous zero-form symmetry with current jµ , we can couple to a background gauge field
A by adding to the Lagrangian ∆L ∋ jµ Aµ . If the symmetry is spontaneously broken,
the effective Lagrangian will contain a Meissner term proportional to A2 . But the effective
action must be gauge invariant, and this requires the presence of a field that transforms
nonlinearly under the U(1) symmetry: φ → φ + λ, A → A − dλ; this is a global symmetry
if dλ = 0. Altogether, the effective Lagrangian must contain a term of the form
1
Leff = − (dφ + A)2 15.
4πg
(where by (ω)2 I mean ωp ∧ ⋆ωp = p! 1
ωµ1 ···µp ω µ1 ···µp ). The coefficient 4πg
1
is the superfluid
stiffness.
The analog for a continuous 1-form symmetry works as follows. The current is now a
two-form, so the background field must be a two-form gauge field Bµν and the coupling is
∆L ∋ Jµν Bµν . The same logic implies that the effective action for the broken phase must
contain a term
1
Leff = − 2 (da + B)2 16.
2g
where the Goldstone mode a is a 1-form that transforms nonlinearly a → a + λ, B → B − dλ;
this is a global symmetry if dλ = 0. Setting the background field B = 0, we recognize this as
a Maxwell term for a. The coupling strength g is determined by the analog of the superfluid
stiffness.

10 McGreevy
 For p-form U(1) symmetry, we conclude by the same logic that there is a massless p-form
field a with canonical kinetic term
Z
1
SMax [a] = − 2 da ∧ ⋆da. 17.
2g

Returning to QED, we see that the familiar Coulomb phase is the SSB phase for the
U(1) 1-form symmetry. The unbroken phase is the superconducting phase, where the photon
has short-ranged correlations. (In an ordinary superconductor, where the Cooper pair has
charge two, a Z2 subgroup of the 1-form symmetry remains broken.)
As in the case of 0-form SSB, the broken phase can be understood via the condensation
of charged objects; in this case the charged objects are the strings of electric flux (30, 31).
Notice that the presence of charged matter, on which these strings can end, and which
therefore explicitly breaks this symmetry, does not necessarily destroy the phase. We’ll
comment on this robustness more in §2.6. In fact, because of electromagnetic duality, the
Coulomb phase is the broken phase for either the electric 1-form symmetry or the magnetic
1-form symmetry (16).

2.5. Effects of IR fluctuations

The analog of the Hohenberg-Coleman-Mermin-Wagner (HCMW) theorem for higher-form
symmetries (10, 11, 16, 29) is interesting. As in the proof of the HCMW theorem, we
suppose that a p-form U(1) symmetry in D spacetime dimensions is spontaneously broken
and that there is therefore a Goldstone mode, a massless p-form field a. Then we ask if
indeed the symmetry is broken by evaluating the expectation value of a charged operator
WC , including the fluctuations of the would-be-Goldstone mode a. We can choose C to
be a copy of Rp ⊂ RD so that we can do the integrals, and the result is (see (29) for a
discussion of a convenient gauge choice)
Z  Z D−p 
R
d̄ k
⟨WC ⟩ = Z −1 [Da]e−SMax [a]+i C a
≃ exp − 12 g 2 Lp 2
18.
k⊥

dk
where d̄k ≡ 2π and k⊥ is the momentum transverse to C. The integral in the exponent
of Eq. 18 is IR divergent when D − p ≤ 2. As in the p = 0 case, we interpret this as the
statement that the long-wavelength fluctuations of the would-be-Goldstone mode necessarily
destroy the order. (For D − p ≥ 2, the integral is UV divergent. This divergence can be
R p
absorbed in a counterterm locally redefining the operator WC → WC e−δT C d x , which can
be interpreted as a renormalization of the tension T of the charged brane.) In the marginal
case of p = D − 2, the long-range order is destroyed, but ⟨WC ⟩ decays as a power-law in the
loop size, rather than an exponential; this is a higher-form analog of algebraic long-range
order in D = 2.
The calculation above is independent of compactness properties of the Goldstone form
field, in the sense that in Eq. 18 we just did a Gaussian integral over the topologically-trivial
fluctuations of a. In the marginal case D = 2 + 1, p = 1, if we treat a as a compact U(1)
gauge field, SSB of the 1-form symmetry is avoided instead because monopole instantons
generate a potential for the dual photon dσ = ⋆da/2π (32). This mechanism generalizes to
any case with D − p = 2 (29).
(10, 11) interpret such results as a generalization of Elitzur’s theorem on the unbreak-
ability of local gauge invariance (33).

www.annualreviews.org • Generalized Symmetries 11

 2.6. Robustness of higher-form symmetries
We are used to the idea that consequences of emergent (aka accidental) symmetries are
only approximate: explicitly breaking a spontaneously-broken continuous 0-form symmetry
gives a mass to the Goldstone boson.
This raises a natural question. The existence of magnetic monopoles with m =
Mmonopole explicitly breaks the 1-form symmetry of electrodynamics:

∂ µ Jµν
E
= jνmonopole .

If the photon is a Goldstone for this symmetry, does this mean the photon gets a mass?
Perhaps surprisingly, the answer is ‘no’ (early discussions of the robustness of broken higher-
form symmetries using different words include (34, 2, 35)). This is a way in which zero-form
and higher-form symmetries are quite distinct.
A cheap way to see that ‘no’ is the right answer is by dimensional analysis. How does
the mass of the photon mγ depend on the mass of the magnetic monopole, Mmonopole ?
Suppose all the electrically charged matter (such as the electron) is very heavy or massless.
We must have mγ → 0 when Mmonopole → ∞. But there is no other mass in the problem
to make up the dimensions.
A slightly less cheap way to arrive at this answer is by dimensional reduction. If we
put quantum electrodynamics (QED) on a circle of radius R, we arrive at low energies at
abelian gauge theory in D = 2 + 1, which is confined by monopole instantons (32). The
monopole instantons arise from euclidean worldlines of magnetic monopoles wrapping the
circle, and so their contribution to the mass of the (2+1)d photon is

mγ (R) ∼ e−RMmonopole . 19.

The polarization of the photon along the circle gets a mass from euclidean worldlines of
charged matter (like the electron) wrapping the circle, so its mass is

m4 (R) ∼ e−Rme . 20.

But now the point is simply that when R → ∞, both of these effects go away and the
(3+1)d photon is massless.
A third argument is that operators charged under a 1-form symmetry are loop operators
– they are not local. We can’t add non-local operators to the action at all. This argument
is not entirely satisfying, since on the lattice even the action for pure gauge theory is a
sum over (small) loop operators. The question is whether the dominant contributors in this
ensemble of charged loop operators grow under the RG. (36) describes a toy calculation
to address this question: begin in a phase with a perimeter law ⟨W [C]⟩ ∼ tlength[C] and
R
consider adding to the action g [dC]W [C] + h.c. in perturbation theory in g. Regularizing
on the lattice and neglecting collisions of loops, the result is the same as integrating out a
charged particle whose mass is determined by the parameter t. For small enough t there is
an IR divergence indicating a transition to a phase where the charged particle is condensed.
Until that happens, the SSB phase survives. A useful slogan extracted from this calculation
is that a loop operator becoming relevant (changing the IR physics) indicates the onset of
a Higgs transition.
The discrete analog of this phenomenon is instructive. In the solvable toric code model,
the discrete 1-form symmetries are exact. But in the rest of the deconfined (spontaneously

12 McGreevy
broken) phase, they are emergent, but still spontaneously broken, and still imply a topology-
dependent groundstate degeneracy that becomes exact in the thermodynamic limit. A
rigorous proof of this (35) constructs (slightly thickened) string operators by quasi-adiabatic
continuation.
Known forms of topological order in d ≤ 3 + 1 have the property that at any T > 0
they are smoothly connected to T = ∞ (a trivial product state). If the 1-form symmetry
is emergent, then as soon as T > 0, a mass is generated for the photon (by the argument
above, with the circle regarded as the thermal circle, so that R = 1/T ), and the state is
smoothly connected to T = ∞.
We do know an example of a topologically ordered phase that is stable at T > 0,
namely the two-form toric code in D = 4 + 1 (37). In the U(1) version of this theory, the
masslessness of the two-form gauge field should survive explicit short-distance breaking of
the U(1) two-form symmetry, even at finite temperature. The reason is that a theory with
a two-form symmetry on a circle still has a 1-form symmetry.
We conclude that the consequences of higher-form symmetries are more robust to ex-
plicit breaking than zero-form symmetries.

2.7. Mean field theory

Landau-Ginzburg mean field theory is our zeroth-order tool for understanding symmetry-
breaking phases and their neighbors. It is therefore natural to ask whether it has an analog
for higher-form symmetries (36). We focus on the simplest nontrivial case of a U(1) 1-form
symmetry.
It is worthwhile to review the logic that produces this weapon. If we take the Landau
paradigm seriously, then the only low-energy modes we require are those swept out by the
symmetry. The key idea is to introduce a degree of freedom ϕ(x) at each point in space that
transforms linearly under the symmetry. ϕ should be regarded as a coarse-grained object,
and this is an effective long-wavelength description. In the example of a magnet, ϕ(x) can
be the magnetization averaged over a small cell at x. Now, because there are no other light
degrees of freedom (by assertion 1), the effective action for ϕ should be given by an analytic
functional of ϕ which is local in spacetime. This functional can therefore be expanded in a
series consisting of all symmetric local functionals of ϕ, organized in a derivative expansion
of terms of decreasing relevance. The length scale suppressing higher derivates is the short
distance over which we averaged in constructing ϕ(x).
The 1-form analog of the order parameter field ϕ(x) (which is a function from the
space of points into a linear representation of G), is a functional ψ[C] from the space of
loops into a linear representation of G, a ‘string field’. While ϕ(x) transforms under the
zero-form symmetry
H
as ϕ(x) → ϕ(x)eiα , with dα = 0, the 1-form analog transforms like
i CΓ
ψ[C] → ψ[C]e , with dΓ = 0.
To write an action for such a field requires the analog of a derivative, which compares its
values on nearby loops. Such an ‘area derivative’ was discovered in the study of loop-space
formulations of gauge theory (38) (see Fig. 3, left). The analog of integrating the action
over spacetime dD x is integrating over the space of loops [dC]. The most general action
R R

consistent with the symmetries then takes the form

δψ ⋆ [C] δψ[C]
Z  I 
1
S[ψ] = [dC] V (|ψ[C]|2 ) + ds + · · · + Sr [ψ] . 21.
2L[C] δCµν (s) δC µν (s)

www.annualreviews.org • Generalized Symmetries 13

 The last ‘recombination term’
Z
Sr [ψ] = [dC1,2,3 ]δ[C1 − (C2 + C3 )] (λψ[C1 ]ψ ⋆ [C2 ]ψ ⋆ [C3 ] + h.c.) + · · · 22.

is not local in loop space, but is local in real space since it involves only a single integral
over the center-of-mass of the loops. Here the delta function imposes the equality of loops
regarded as integration domains (see Fig. 3, right). The · · · denote terms with more deriva-
tives or more powers of ψ. Models similar to this Mean String Field Theory (MSFT) have

C1
C2
<latexit sha1_base64="b342B6lxS9u4MYepTTuvsFWGG+U=">AAAB6nicbVBNS8NAEJ34WetX1aOXxSJ4KkkV9FjsxWNF+wFtKJvtpF262YTdjVBCf4IXD4p49Rd589+4bXPQ1gcDj/dmmJkXJIJr47rfztr6xubWdmGnuLu3f3BYOjpu6ThVDJssFrHqBFSj4BKbhhuBnUQhjQKB7WBcn/ntJ1Sax/LRTBL0IzqUPOSMGis91Ptev1R2K+4cZJV4OSlDjka/9NUbxCyNUBomqNZdz02Mn1FlOBM4LfZSjQllYzrErqWSRqj9bH7qlJxbZUDCWNmShszV3xMZjbSeRIHtjKgZ6WVvJv7ndVMT3vgZl0lqULLFojAVxMRk9jcZcIXMiIkllClubyVsRBVlxqZTtCF4yy+vkla14l1WqvdX5dptHkcBTuEMLsCDa6jBHTSgCQyG8Ayv8OYI58V5dz4WrWtOPnMCf+B8/gC8MY1v</latexit>

<latexit sha1_base64="1qkECvOB+R2I9+RxbCR24LYk+kg=">AAAB6nicdVDLSsNAFL3xWeur6tLNYBFchaQPWnfFblxWtA9oQ5lMJ+3QySTMTIQS+gluXCji1i9y5984bSOo6IELh3Pu5d57/JgzpR3nw1pb39jc2s7t5Hf39g8OC0fHHRUlktA2iXgkez5WlDNB25ppTnuxpDj0Oe360+bC795TqVgk7vQspl6Ix4IFjGBtpNvmsDQsFB27WqpfVitoRcr1jFRqyLWdJYqQoTUsvA9GEUlCKjThWKm+68TaS7HUjHA6zw8SRWNMpnhM+4YKHFLlpctT5+jcKCMURNKU0Gipfp9IcajULPRNZ4j1RP32FuJfXj/RQd1LmYgTTQVZLQoSjnSEFn+jEZOUaD4zBBPJzK2ITLDERJt08iaEr0/R/6RTst2yXbqpFBtXWRw5OIUzuAAXatCAa2hBGwiM4QGe4Nni1qP1Yr2uWtesbOYEfsB6+wRSr43X</latexit>

C3
<latexit sha1_base64="3iDpH6MRg0ngr4gVSKGCmFgZBDA=">AAAB6nicdVDLSsNAFL3xWeur6tLNYBFchaQPWnfFblxWtA9oQ5lMJ+3QySTMTIQS+gluXCji1i9y5984bSOo6IELh3Pu5d57/JgzpR3nw1pb39jc2s7t5Hf39g8OC0fHHRUlktA2iXgkez5WlDNB25ppTnuxpDj0Oe360+bC795TqVgk7vQspl6Ix4IFjGBtpNvmsDwsFB27WqpfVitoRcr1jFRqyLWdJYqQoTUsvA9GEUlCKjThWKm+68TaS7HUjHA6zw8SRWNMpnhM+4YKHFLlpctT5+jcKCMURNKU0Gipfp9IcajULPRNZ4j1RP32FuJfXj/RQd1LmYgTTQVZLQoSjnSEFn+jEZOUaD4zBBPJzK2ITLDERJt08iaEr0/R/6RTst2yXbqpFBtXWRw5OIUzuAAXatCAa2hBGwiM4QGe4Nni1qP1Yr2uWtesbOYEfsB6+wRUM43Y</latexit>

Figure 3
Left: a sketch of the definition of the area derivative δC δ (s) . Right: The arrangement of loops
µν
involved in the topology-changing term Sr in the MSFT action.

been considered before in various specific contexts (39, 40, 41, 42, 43).
The potential term V (|ψ[C]|2 ) = r|ψ[C]|2 + u|ψ[C]|4 + · · · √controls the low-energy
behavior. If r > 0, we find an unbroken phase where ψ[C] ≃ e− rA[C] . When r < 0, the
strings want to condense. The fluctuations around nonzero |ψ| are all massive, except for
H µ
the geometric mode ψ[C] = vei C dsaµ (x(s))ẋ (s) , which describes a slowly-varying 1-form
symmetry transformation, and in terms of which the action Eq. 21 reduces to the Maxwell
action for a, with coupling g 2 = 2v12 .
As in the zero-form case, another application of this mean field theory is to classify
topological defects of the resulting ordered media. The conclusion for G = U(1) is that the
only topological defect is the codimension-three magnetic monopole.
So far, we have discussed the case of a U(1) 1-form symmetry. The case of discrete
symmetries can be approached by explicitly breaking the U(1) 1-form symmetry to a discrete
subgroup. A term of the form Z
h [dC]ψ k [C] + h.c. 23.

breaks it down to Zk . In the broken phase, the effective action reduces to a continuum (BF)
description of Zk gauge theory.
The action Eq. 21 can be given a lattice definition and contact can be made with mi-
croscopic Hamiltonians in the following way. Zero-form mean field theory arises from a
variational using a product state ansatz Ψϕ = ⊗x ϕ(x) ; given a microscopic Hamilto-
nian, the variational energy Ψϕ Ĥ Ψϕ = H[ϕ] takes the form of the Landau-Ginzburg
Hamiltonian.
Consider for definiteness a Z2 lattice gauge theory Hamiltonian, in the form
X X X
HTC = −∞ As − Γ Bp − g Zℓ . 24.
sites,s plaquettes,p links,ℓ

This acts on a Hilbert space that is a tensor product of qubits on the links of a cell complex;
Q Q
As = ℓ∈s Zℓ and Bp = ℓ∈p Xℓ . X and Z denote the Pauli operators. In the Z eigenbasis,

14 McGreevy
we regard a link as covered by a segment of string if Z = −1. We take the coefficient of the
star term As to infinity so that the loops are closed and there is an exact (electric) 1-form
Q
symmetry generated by U (C) = ℓ∈C Xℓ . When g = 0, the groundstate is the uniform
superposition over all collections of closed loops. g represents a tension for the electric
strings; for large enough g/Γ, there is a transition to a confined phase.
The analog of a product state for the 1-form case is a many-body wavefunction on
collections of loops determined by a function ψ[C] on a single loop:
P
ψ[c]U [c]
Ψψ =: e c,connected : 0 25.

where U [c] 0 = c creates the loop c, and the normal-ordering symbol : · · · : is a pre-
scription for dealing with overlapping loops. The variational energy for this state is a lattice
Hamiltonian for the action Eq. 21 plus Eq. 23.
Brief comments on phase transitions. As in the 0-form case, we expect the mean-
field description to break down near critical points, below the upper critical dimension. (The
extension of the renormalization group to MSFT has not yet been attempted.) Dimensional
analysis says that the string field ψ has dimension (D−4)/2 and hence u has mass dimension
8 − D, and λ has dimension 6 − D/2, which puts the upper critical dimension at 8 or
12, depending on which coupling matters. More significantly, the recombination term is
a symmetric term cubic in the order parameter field, and we expect that it generically
renders the transition first order. This is consistent with numerical work on deconfinement
transitions in gauge theory in D > 3 (see e.g. (44) and references therein).
Notice that the string field has engineering dimension zero in D = 4. There are two
possible notions of lower critical dimension, which coincide at D = 2 in the case of 0-form
symmetries. One is the largest dimension where the HCMW theorem forbids symmetry
breaking, which is D = 3 for 1-form symmetry. The other is the dimension in which the
linearly-transforming field is classically dimensionless, which is D = 4 for 1-form symme-
tries. In the case of 0-form symmetry, this allows for the rich physics of the Berezinsky-
Kosterlitz-Thouless transition, where there is a line of (free) fixed points (parameterized
by g in Eq. 15) that terminates when a symmetry-allowed operator becomes relevant. A
universal prediction is the value of the stiffness at the transition, since as in Eq. 15, the
stiffness determines the coupling.
For the special case of 1-form symmetry-breaking in D = 4, there is again a line of
(free) fixed points, parameterized by the Maxwell coupling, as in Eq. 16. Consider the
application of this picture to Zk gauge theory, described by perturbing the MSFT action
by Eq. 23. In the free theory, this operator can be argued (45) to have an anomalous
g 2 k2
√
dimension ∆k (g) = 32π 2 ; for large enough k, ∆p (g) passes through 4 at some gc < 4π,
and we can interpret this as the location of a transition in the low-energy physics from a
Coulomb phase to a phase with Zk topological order. The prediction is again a universal
jump in the ‘superfluid stiffness’, namely the value of the gauge coupling at the transition.
Many of these ideas were anticipated by Cardy (46) without the benefit of the language
higher-form symmetry.
There is a catch: this transition is observed in Monte Carlo simulations to actually
be weakly first order (see (47) and references therein). Does that mean there is nothing
universal to say? There is a reason the transition is weakly first order. The magnetic charge
whose condensation drives the transition has a good dual description via the Abelian Higgs
model. In this model, fluctuations drive the transition first order (48). If the coupling is
weak at the transition, this description is good and the transition is weakly first order. But,

www.annualreviews.org • Generalized Symmetries 15

 using the slogan of §2.6 (a loop operator becoming relevant means a Higgs transition), the
mechanism of the previous paragraph determines the critical coupling and shows that it
should be small at large k.

3. Anomalies
My motivation for including a discussion of anomalies here is twofold. One is that anomalies
are a necessary ingredient in a suitably-generalized Landau Paradigm that incorporates all
phases, in particular topological insulators and SPT phases. A second motivation is that, as
I will review, the existence of anomalies makes symmetries much more useful for constraining
the dynamics of a physical system, and their generalization to higher-form symmetries is
therefore an essential step.
The historical, high-energy-physics perspective on anomalies starts from specifying a
quantum field theory by a path integral
Z
Z= [D(fields)]eiS[fields] . 26.

An anomaly is a symmetry of the action S that is not a symmetry of the path-integral

measure. The first example found was the chiral anomaly, the violation of the axial current
of a charged Dirac field (the symmetry that rotates left-handed and right-handed fermions
with opposite phases)
µ e2
∂µ jA =N ϵµνρσ F µν F ρσ , 27.
16π 2
which controls the decay of the neutral pion into two photons.

ϵ ϵ

ΔNR
{
ΔNL
ϵF
} ϵF

2 π4 π
k 2 π4 π
k
-kF ... kF -kF-kF +Δp ... kF kF +Δp
L L L L

Figure 4
Left: Spectrum of a free-fermion tight-binding model in one dimension, near the bottom of the
band at some small filling. Green circles indicate filled states. Right: The result of adiabatically
applying an electric field. NL/R indicate the number of left-moving and right-moving excitations.

A more concrete perspective arises if we consider the same kind of system on the lattice,
in one dimension for simplicity: consider a tight-binding model of fermions hopping on a
chain, at some small filling as in Fig. 4. In this case, there is no chiral symmetry at all at
the lattice scale. It is an emergent symmetry, violated by the UV physics in a definite way.
At low energies, the system is approximately described by the neighborhood of the two
boundaries of the Fermi sea, giving a 1d massless Dirac fermion, with a chiral symmetry.

16 McGreevy
But if we adiabatically apply an electric field Ex , every fermion increases its momentum
and the chiral charge changes by
Z Z
∆p L e
∆QA = ∆(NR − NL ) = 2 = e dtEx (t) = ϵµν F µν . 28.
2π/L π 2π

The left hand side is ∆QA = ∂ µ jµA , and so Eq. 28 is the 2d version of the chiral anomaly:
R

µ e
∂µ jA = ϵµν F µν . 29.
2π
A reason for excitement about this phenomenon is that the coefficient N in Eq. 27 is
an integer. This is the first hint that an anomaly is a topological phenomenon, a quantity
that is RG invariant (49). The idea is that the existence of the anomaly means that the
partition function varies by some particular phase under the anomalous symmetry, but an
RG transformation must preserve the partition function. Much of physics is about trying
to match microscopic (UV) and long-wavelength (IR) descriptions. That is, we are often
faced with questions of the form “what could be a microscopic Hamiltonian that produces
these phenomena?” and “what does this microscopic Hamiltonian do at long wavelengths?”.
Anomalies are precious to us, because they are RG-invariant information: any anomaly in
the UV description must be realized somehow in the IR description.
Another useful perspective on anomaly is as an obstruction to gauging the symmetry.
Gauging a symmetry means creating a new system where the symmetry is a redundancy
of the description, by coupling to gauge fields. If the symmetry is not conserved in the
presence of background gauge fields, the resulting theory would be inconsistent.
Above I’ve described an example of an anomaly of a continuous symmetry. Discrete
symmetries can also be anomalous.
Anomaly is actually a more basic notion than phase of matter: The anomaly is a
property of the degrees of freedom (of the Hilbert space) and how the symmetry acts on
them, independent of a choice of Hamiltonian. Multiple phases of matter can carry the
same anomaly.

3.1. SPT phases and anomalies

The definition of gapped phases can be refined by studying only the space of Hamiltonians
preserving some particular symmetry group G. Two phases that are distinct in this smaller
space may nevertheless be connected by a gapped path in the larger space of non-symmetric
Hamiltonians.
One way to define (50) a Symmetry-Protected Topological (SPT) phase is as a nontrivial
phase of matter (with some symmetry G) without topological order (for a review, see (51)).
SPT phases can be characterized by their edge states. The idea is that the edge theory has
to represent an anomaly of the symmetry G. It is really this anomaly that labels the bulk
phase. This phenomenon is called anomaly inflow.
As a simple example, consider an effective field theory for the integer quantum Hall
effect, regarded as an SPT for charge conservation symmetry7 . The charge conservation
symmetry is associated, by Noether’s theorem, with a conserved current j µ , with ∂µ j µ = 0.

7 Actually, the integer quantum Hall phase is more robust, and survives explicit breaking of

the charge conservation symmetry. It is protected by the gravitational anomaly manifested in the
nonzero chiral central charge.

www.annualreviews.org • Generalized Symmetries 17

 In D = 2 + 1, this equation can be solved by writing j µ = ϵµνρ ∂ν aρ /(2π), in terms of a
1-form gauge field aµ , with redundancy a → a + dα. The leading effective action for such
a field, in the absence of parity symmetry, is a Chern-Simons term (52, 53):
Z
1
SIQH [a, A] = ϵµνρ (aµ ∂ν aρ + 2Aµ ∂ν aρ ) 30.
4π M

where A is a background field for the charge conservation symmetry. Under A → A + dλ,
ij
δSIQH = ∂M ϵ4π fij λ. This is the contribution to the chiral anomaly from a single right-
R

moving edge mode.

In terms of the definition of the anomaly as a variation of the partition function of the
edge theory in the presence of background fields, the variation of the bulk action cancels
the anomaly of the edge theory, so that the whole system is G symmetric. The edge theory
cannot be trivial, since it has to cancel the variation of the bulk under the symmetry
transformation: it has to be either (54)

• gapless
• symmetry-broken
• or topologically ordered.

In particular, there cannot be a trivial gapped groundstate. These are the same conditions
arising from the Lieb-Schultz-Mattis-Oshikawa-Hastings (LSMOH) theorem (55, 56) (for
more recent developments, see e.g. (57)), and we can call this an LSMOH constraint.
A perhaps-simpler example is the free fermion topological insulator in D = 3 + 1,
protected by charge conservation and time-reversal symmetry. In this case, the bulk effective
action governs a single massive Dirac fermion; a boundary is an interface where the mass
changes sign, at which a single Dirac cone arises. A single Dirac cone in D = 2+1 realizes the
so-called parity anomaly. The fact that anomaly transcends a phase of matter is illustrated
by the fact that, in the presence of interactions or disorder, there are other possible edge
theories for the topological insulator.
There is by now a sophisticated (still conjectural) mathematical classification of SPTs
for various G in various dimensions (58, 59) about which I will not say more here. My point
is that we are still using their realization of symmetries to label these phases!

3.2. Anomalies of higher-form symmetries

Let’s return to the example from §2.1 of the (d − 1)-form symmetry that arises in any
superfluid phase (17, 18, 60). The current can be written as (⋆J)µ = ∂µ φ. However, in the
presence of a background gauge field A for the U(1) symmetry, the gauge-invariant current
is instead
(⋆J)µ = Dµ φ 31.

where Dµ φ = ∂µ φ − qAµ is the covariant derivative. But this current is not conserved:

d ⋆ J = −qF 32.

with F ≡ dA. This equation has a simple interpretation: applying an electric field leads to
a supercurrent that increases linearly in time.
The symmetry violation in Eq. 32 is an example of a mixed anomaly between a 0-form
symmetry and a (d − 1)-form symmetry, that arises automatically from SSB. Reference (18)

18 McGreevy
shows a converse statement: any system with U(1)(0) × U(1)(D−2) symmetry with anomaly
Eq. 32 contains a Goldstone boson in its spectrum. Since no long-range order is assumed,
this is a more general statement than Goldstone’s theorem – it applies even in D = 2.
This perspective can be used to demonstrate the existence of equilibrium states with
non-dissipating current (60).
A direct 1-form generalization of Oshikawa’s argument (55) appears in (61). This is an
example of a mixed anomaly between a 1-form symmetry and lattice translation symmetry.
We should give an example of an anomaly of a higher-form symmetry that does not
involve zero-form symmetries. An example is provided by the theory of abelian anyons in
D = 2 + 1, and is best understood by regarding an anomaly as an obstruction to gauging.
Gauging a continuous 1-form symmetry means coupling the conserved current J µν to a
dynamical two-form gauge field bµν by a term like bµν J µν . That is, gauging a symmetry
means summing over all possible background fields. In the discrete case, this is the same as
summing over the insertions of all possible symmetry operators. (In the continuous case, it
also requires summing over connections that are not flat.)
Thus, gauging a 1-form symmetry in 2+1 dimensions means proliferating the worldlines
of the associated anyons (16, 62); this is ‘anyon condensation’ (63). But it only makes
sense to condense particles with bosonic self-statistics: condensation means essentially that
the many-particle wavefunction is a constant, which has bosonic statistics. Therefore,
a subgroup of a 1-form symmetry generated by line operators with nontrivial statistics
cannot be gauged. We conclude that, in 2+1 dimensions, the ’t Hooft anomaly of a 1-form
symmetry is encoded in the self-statistics of the line operators, i.e. of the anyons. Thus,
the algebra Eq. 14 is an example of a 1-form symmetry with an ’t Hooft anomaly. Notice
that from this point of view, non-trivial mutual statistics of a pair of anyon types a and
b is a mixed ’t Hooft anomaly: it does not stop us from gauging (i.e. condensing) a, but
we cannot condense both simultaneously, since in the presence of the a condensate, b is
confined. The algebra for discrete gauge theory Eq. 11 can also be regarded an example of
an anomaly for higher-form symmetry because the charged operators Vn are also topological;
so this is a 1-form symmetry and a (D − 2)-form symmetry with a mixed anomaly. In fact,
the generalized symmetry that emerges and is spontaneously broken in any topologically
ordered groundstate is always anomalous: this is the statement of braiding nondegeneracy,
which is an axiom of topological field theory, and a theorem of Entanglement Bootstrap
(64).

3.3. SPT phases of higher-form symmetries

We can combine the ingredients of the above discussions, and consider SPT phases protected
by higher-form symmetries (21). This is a slightly awkward subject because higher-form
symmetries tend to be emergent, and it therefore might be artificial to restrict ourselves to
the subspace of Hamiltonians with exact higher-form symmetry.
In D = 2 + 1, an ’t Hooft anomaly for a 1-form symmetry is diagnosed by the self-
statistics of the line operators. So the edge of a 1-form G SPT in D = 3 + 1 just needs
to have G topological order with quasiparticles that aren’t bosons. Lattice models for
higher-form SPTs have been written down in (65, 66) and effective theories studied in (67).

www.annualreviews.org • Generalized Symmetries 19

 4. Subsystem symmetries and fracton phases
Above we have discussed p-form symmetries, described by symmetry operators acting on
codimension-(p + 1) submanifolds of spacetime. These operators were deformable, in the
sense that their correlations only depend on their deformation class in spacetime (avoiding
any charged operator insertions).
A distinct generalization of the notion of symmetry arises by defining symmetry oper-
ators acting independently on rigid subspaces of the space on which the system is defined.
That is, we can imagine that there is a different symmetry operator for each subspace, even
in the same homology class, so that the symmetry operators are not topological, but still
commute with the Hamiltonian. This is sometimes called a “faithful” symmetry (68) or
subsystem symmetry. This generalization is not compatible with Lorentz invariance.
An object charged under such a subsystem symmetry cannot leave the locus on which
the symmetry is defined. This sort of restricted mobility of excitations is a defining property
of fracton phases (5, 6). A fracton phase can be identified as one that spontaneously breaks
such a “faithful” higher-form symmetry (68, 69, 70). Foliated fracton phases (71) like the
X-cube model (72) spontaneously break a ‘foliated 1-form symmetry’ acting independently
on each plane of a lattice (68).
A closely-related concept is that of multipole symmetries (e.g. (73, 74, 75, 76, 77, 78)).
A multipole symmetry is one where the continuity equation involves extra derivatives, like
∂0 J 0 + ∂i ∂j J ij = 0 (a dipole symmetry). Such a conservation law produces conserved
charges that need not be integrated over all of space, and act independently of each other.
(For example (75), consider the continuity equation ∂0 J 0 + ∂x ∂y J = 0 in D = 2 + 1; then
Qx (x) = dyJ 0 (x, y) is conserved for each x.) The simplest example is that conservation
R

of dipole moment implies that charges are immobile (73).

Models with such symmetries have been studied for a long time in the condensed matter
literature (79). Efforts to understand how the rules of ordinary field theory must be relaxed
to accommodate such systems and their symmetries have been vigorous (see e.g. (80, 71,
81, 82, 83, 75, 76, 77, 84) and references therein and thereto). Attempts have been made
to classify subsystem-symmetry-protected topological phases (85) and their anomalies (86),
and to understand subsystem-symmetry-enriched topological order (87). A subsystem-
symmetry-based understanding of Haah’s code (88) appears in (89).
An important issue is the robustness of such phases, especially in the gapless case,
upon breaking the large symmetry group. At least in examples, the scaling dimensions
of operators charged under the subsystem symmetry is large, and in fact diverges in the
continuum limit (79, 83, 75, 76, 77) (see in particular Eq. (121) of the first reference). This
shows that there is at least a small open set in the space of subsystem-symmetry-breaking
couplings in which such phases persist.
Fractal symmetry. The subsystem on which a symmetry acts can be more interesting
than just a line or a plane. For example, it can be a fractal (90, 91). The Newman-Moore
model (92) is a simple example of a model with a symmetry operator supported on a fractal
subset of space. Put qubits on the sites i of the triangular lattice and consider
X X
H= Zi Zj Zk + g Xi , 33.
ijk∈∆ i

where the sum is only over up-pointing triangles. To see that this has a fractal symmetry,
pick a spin to flip, say the circled spin in Fig. 5. Moving outward from that starting point
and demanding that each up-triangle contains an even number of flipped spins, there are

20 McGreevy
many possible self-similar subsets of the lattice we can choose to flip. In fact, there is an
extensive number.
This transverse-field Newman-Moore model Eq. 33 has a number of interesting prop-
erties. It has a self-duality mapping g → 1/g, obtained by defining dual spins X̃∆ ≡
Q
i∈∆ Zi Zj Zk on a new lattice with sites corresponding to the up-pointing triangles. The
exotic critical point at g = 18 (94) separates a gapped paramagnetic phase from a gapless
phase in which the fractal Z2 symmetry is spontaneously broken. Such critical points were
claimed (94) to be ‘beyond renormalization’; rather, what is broken is the connection be-
tween short distances and high energies (84). Other models with such fractal symmetry
have been studied in (95).

Figure 5
An example of the support of a fractal symmetry operator in the Newman-Moore model. If we flip
only the red spins, it preserves the Hamiltonian Eq. 33. That is, every up-triangle has an even
number of red dots. There are many ways to accomplish this.

5. Categorical symmetries
Our understanding of what is a symmetry of a quantum many body system or quantum
field theory (QFT) has evolved quite a bit. The above discussion shows that the presence
of a symmetry means the existence of topological defect operators9 . (I believe the word
‘defect’ in this name just refers to the fact that these operators have positive codimen-
sion.) In the case of an ordinary symmetry, these are the symmetry operators, Ug (Σd ),
that we discussed above; they are labelled by a group element g ∈ G, and supported on
a codimension-one (e.g. fixed-time) slice Σd , and are topological in the sense that their
correlation functions do not change under continuous deformations. These operators sat-
isfy a ‘fusion rule’ in the sense that for two symmetry operators associated with the same
time-slice, limϵ→0+ Ug (t + ϵ)Uh (t) = Ugh (t). When a local operator crosses such a Ug , it
gets acted on by the transformation g.
If the surface Σd is not a fixed-time slice, such an operator implements a modification of

8 Earlier work (90, 93) found indications of a first-order transition.

9A sufficient condition for this conclusion is Lorentz symmetry. In its absence, we have already
seen examples of systems with subsystem symmetries, where there are operators that commute with
the Hamiltonian that are not fully topological.

www.annualreviews.org • Generalized Symmetries 21

 the Hamiltonian, such as a change of boundary conditions.A good example to keep in mind
is the defect operator U−1 (Σ) in the classical Ising model. It is an instruction to flip the
sign of the coupling along any bond crossing the codimension-one locus Σ. This operator
is topological: deforming Σ through a region R is accomplished by redefining all the spins
in R by σ → −σ. This shows that the charged operator is the spin.
A useful perspective is to reverse the logic, and regard the existence of topological
defect operators as the definition of a symmetry. One of the advantages of this perspective
is that it treats continuous and discrete symmetries uniformly; it also makes no reference
to transformations of fields, and so treats Noether symmetries and topological symmetries
uniformly. And from this perspective it is easy to see some generalizations. The first
generalization is that a p-form symmetry is associated with (unitary) topological operators
whose support has codimension p + 1. In the low-energy theory describing an abelian
topological order in D = 2 + 1, these operators U are the holonomies of anyon worldlines.
For two operators on the same submanifold, Uα (MD−p−1 )Uβ (MD−p−1 ) = Uα+β (MD−p−1 )
where for p > 1, the order does not matter, and we adopt an additive notation.
One way in which an anomaly can appear in this language is when, in the presence of
background fields F (which could include curvature of spacetime), the symmetry operator
fails to be topological in the sense that
R
Uα (M ) = eiα D Γ(F )
Uα (M ′ ), 34.

where D is a (D − p)-dimensional surface bounding the difference of M and M ′ : ∂D = M −

M ′ , and Γ(F ) is some polynomial in background fluxes and curvatures of the appropriate
form degree (96).
The preceding discussion suggests a further generalization, which we will need in order
to describe non-abelian topological order as SSB: what about the worldlines of non-abelian
anyons? This is a dramatic step because the algebra of topological operators Ta that
transport non-abelian anyons is no longer a group. Rather, they satisfy the fusion algebra:
X c
Ta Tb = Nab Tc . 35.
c

By definition, a topological order is non-abelian if there is more than one term on the
RHS of this equation for some choice of a, b. Whereas multiplication of two elements of a
group always produces a unique third element, here we produce a superposition of elements,
c
weighted by fusion multiplicities Nab . Further, there is some tension between the fusion
algebra Eq. 35 and unitarity of the operators Tc . The trivial anyon corresponds to the
identity operator, T1 = 1. Each type of anyon a has an antiparticle ā. Since Tā corresponds
to transporting a in the opposite direction, we expect that Tā = Ta† , and therefore Eq. 35
says in particular X c
Ta Ta† = Naā Tc . 36.
c
1
If the RHS here has a term other than Naā , then Ta is not unitary. As an example, consider
the Ising topological order, with three anyon types {1, ψ, σ} and the fusion rules

Tσ Tσ = 1 + Tψ , Tσ Tψ = Tψ Tσ = Tσ , Tψ Tψ = T1 . 37.

Note that σ is its own antiparticle. Eq. 37 implies that the topological line operator Tσ
cannot be unitary, and moreover cannot be inverted by any linear combination of Ta . Such
symmetries are called categorical symmetries or fusion category symmetries.

22 McGreevy
 α β α β γ α β γ

µ ν
X 
a) b) = Fδαβγ
µν
ν

γ δ δ
Figure 6
γ
a) Fusion of symmetry operators: this junction is allowed if Nαβ ̸= 0. b) Associativity data of
γ
fusion of symmetry operators (in the simpler case where the fusion coefficients Nαβ are only 0 or
1).

More generally, any algebra of topological operators acting on a physical system can be
regarded as encoding some kind of generalized symmetry.
At the moment, condensed matter applications of the idea of fusion category symmetries
remain in the realm of relatively formal developments, as opposed to active phenomenology
of real materials. One application is to understand non-abelian topological order as spon-
taneous symmetry breaking10 . A concrete example of a (2+1)d model with non-invertible
symmetries is Gk Chern-Simons (CS) theory, with non-Abelian gauge group G at level
k > 1. The non-invertible symmetry operators are the Wegner-Wilson lines. The spe-
cific example of SU(2)2 CS theory can describe the Ising topological order, and is possibly
realized as part of the effective low-energy description of ν = 25 quantum Hall states.
More generally, any topological field theory for non-Abelian topological order enjoys
such a non-invertible symmetry. A nice example of the application of this perspective on
anyon worldlines as symmetry operators is (99) which provides a condition on the anyon
data required for a general 2+1D topological order to admit a gapped boundary condition,
beyond vanishing chiral central charge.
Part of the reason for the nomenclature ‘categorical symmetry’ is that such a collection
of symmetry operators comes with some additional data. Besides putting two symmetry
operators right on top of each other, we can also consider symmetry operators associated
with branched manifolds, as in Fig. 6a. Once we allow such objects, we must also consider
more complicated objects related to the associativity of the product, as in Fig. 6b, which
relates the two ways of resolving a 4-valent junction of topological operators into two 3-
valent junctions. This associativity information (creatively called F -symbols) is part of
the specification of the categorical symmetry, and must satisfy the pentagon identities (see
e.g. Fig. 1 of (100)). In the case of 1-form symmetry in (2+1)-D, there is further information
associated with braiding.
A good example of a non-invertible line operator appears in the critical Ising conformal
field theory (CFT) in D = 1 + 1, in the form of the duality wall (Fig. 7). The definition
of such an object is: when we pass through the wall, we act by the Kramers-Wannier self-
duality interchanging the spin and the disorder operator. The latter is not a local operator,
but rather must be attached to a branch cut across which the Z2 symmetry acts. Moving a

10 A related perspective appears via the ‘pulling-though’ operators in the tensor network descrip-

tion of topological orders reviewed in (97). For a study of categorical symmetries realized as matrix
product operators, see (98).

www.annualreviews.org • Generalized Symmetries 23

 N
σ(x) µ(x)
⇝
η
N N
Figure 7
When a spin σ moves through the duality wall N , it turns into a disorder operator µ, attached by
η
a topological line η to the duality wall. The right figure illustrates the fact that NN N ̸= 0.

local spin operator through such a duality wall then turns it into an operator attached by
a topological defect line to the duality wall. The fusion algebra of the duality wall operator
N and the ordinary Z2 symmetry line operator η can be summarized as

ηη = 1, N η = ηN = N , N N = 1 + η.

(These are a relabelling of the Ising fusion rules above.) The last, non-abelian, relation
comes from the fact that the Kramers-Wannier duality only keeps track of the locations of
domain walls, and erases the information about the overall spin flip. In a theory with such
a symmetry operator, RG flows generated by a perturbation by a local operator can only
generate operators that pass freely through the wall (96, 101). Examples of duality walls
in D = 3 + 1 were studied in (102, 103).
Categorical symmetries have been studied most extensively in 1+1d QFTs (e.g. (104, 96,
100, 105, 106, 101, 107, 108, 109)), where they can be used constrain RG flows. It was shown
in (96) that certain non-invertible symmetries can forbid a trivial gapped groundstate, as in
the LSMOH theorem. The idea is to consider the partition function on T 2 with a symmetry
line operator L wrapping one of the circles, and argue by contradiction. If there is a gap,
we can evaluate this quantity in the effective low-energy topological theory. Demanding
modular invariance (i.e. that we get the same answer whichever circle we regard as time)
relates the trace over the Hilbert space with twisted boundary conditions

L L − 2π
trHL e−β(H−E0 ) =
(H−E0 )
= = trLe β , 38.
t x x t

to the ordinary trace with the insertion of the symmetry operator. In a topological field
theory, the former quantity is just trL 1, the number of states in the twisted sector. If
there is furthermore a unique groundstate, then the latter quantity is just ⟨L⟩. Since the
former is a non-negative integer, we can conclude that if ⟨L⟩ is not a non-negative integer,
then there cannot be a unique gapped groundstate. For example, a certain perturbation
of the tricritical Ising model has a symmetry operator W with (Fibonacci) fusion algebra
√
W 2 = 1 + W . This algebra implies that the eigenvalues of W are (1 ± 5)/2, and there
must therefore be an even number of vacua for its expectation value to be an integer. More
generally, this argument shows that (in a system with modular invariance) the existence of
a symmetry operator with no integer eigenvalues forbids a unique groundstate (96): if the
unique groundstate were not an eigenvector of L, we could make another groundstate by
acting with L, therefore if there is a unique groundstate, then ⟨L⟩ must be an eigenvector
of L. A related argument shows (110) that all irreps of G appear in the spectrum of a 1+1d
CFT with finite symmetry group G. An extension of this modular-invariance argument to
3+1-D can be found in (102).

24 McGreevy
 The edge theory of the Gk CS theory is the Gk WZW model; it inherits the categorical
symmetry from the Wegner-Wilson lines running parallel to the boundary. These ingredients
are used by (101) to construct massless 2d QCD with adjoint fermions by coupling CS theory
on an interval to 2d Yang-Mills theory; the construction makes manifest some surprising
non-invertible symmetries of the theory, which guarantee deconfinement.
(100, 106) argue that a 1+1d system with fusion category symmetry can always be
realized as a boundary condition of a gapped 2+1 dimensional topological order with anyon
types carrying the associated labels of the topological operators. They wish to study anoma-
lies of the fusion category symmetry, to use them as RG invariants, and to label SPTs
protected by such a symmetry: the bulk is a realization of anomaly inflow. Gapped edge
theories are realized if the bulk theory admits gapped boundary conditions; such a bulk the-
ory has an exactly solvable description as a string-net model (30). Explicit lattice models
for gapped phases in D = 1 + 1 with fusion category symmetries appear in (111).
Examples of systems with categorical symmetries include the anyon chain models stud-
ied in (112), which uses the categorical symmetry to explain the gaplessness of the model.
(113, 114, 115) build classical lattice models whose defects realize a fusion category.
The terms ‘categorical symmetry’ and ‘non-invertible symmetry’ are not used in a unique
way in the literature. In (116, 117, 20), the term is used in the context of gapped phases in
D = 2 + 1 with gapless boundaries; the idea is that such edge theories can have anomalies
that go beyond those associated with invertible phases, which are therefore called non-
invertible anomalies. The term ‘algebraic higher symmetry’ is used in (116, 117) for the
concept I called categorical symmetry above. (116, 117, 118) argue that the most general
notion of symmetry of a D-dimensional system is labelled by a topological order in one
higher dimension.

6. Gapless states
6.1. Critical points
The second part of the Landau paradigm (Item 2) says that at a critical point, the critical
degrees of freedom are the fluctuations of an order parameter. Apparent exceptions to this
statement come in several varieties.
First, any transition out of a phase without a local order parameter presents an imme-
diate problem. Consider the case of Z2 gauge theory in D = 2 + 1, which spontaneously
breaks a Z2 1-form symmetry, with a charged loop operator W [C]. Can we understand the
critical theory in terms such a string order parameter field? By Wegner’s duality (2), the
local physics of the critical theory is in the same universality class as the 3d Ising model.
This is yet another point of view from which the 3d Ising model should have a string theory
dual (119, 19).
Second, there are direct transitions between states that break different symmetries,
known as deconfined quantum critical points (DQCP) ((120) has a useful summary and
references). Does this require a revision to Item 2 of the Landau Paradigm as stated
above? There is a sense in which the degrees of freedom of the critical theory are simply
the order parameters of both of the neighboring phases, coupled by a WZW term (121,
122). The presence of the WZW term is required by a mixed anomaly between the two
symmetries. It says that defects of the order in one phase carry charge under the other
(123). This perspective predicts a dramatic enlargement of symmetry at the critical point,
not obvious from other points of view, and borne out by numerical work. This symmetry-

www.annualreviews.org • Generalized Symmetries 25

 based description as a non-linear sigma model has the serious shortcoming that it is strongly
coupled, but so is the more-familiar description in terms of abelian gauge theory.
Independently of the extended Landau Paradigm, I should also mention that the study
of order parameters for higher-form symmetry at various critical points has been instructive
(20, 20, 124, 125). In particular, their study has provided independent evidence that the
2+1d DQCP between Neel and VBS phases is a weakly first order transition (125).

6.2. Gapless phases

Gapless phases are a wild frontier of our understanding, and we certainly do not have a
symmetry-based (or any other) understanding of all possibilities at the moment. I limit
myself to remarks on two illuminating examples.
First, I mention a set of exotic gapless fractonic phases that can be constructed by
assembling layers of quantum Hall states. They can be described at low energies by an
IJ R
abelian Chern-Simons theory S[aI ] = K4π aI ∧ daJ with a nearly-diagonal K matrix
whose size grows with the number of layers (126, 127, 128). For some choices of K matrix,
this represents a new class of gapped fracton phase, with irrational particle statistics and a
large-order fusion group. For other choices of K matrix, the spectrum is gapless. Ref (129)
shows that the gapless examples of such states can be understood in terms of weak symmetry
breaking (130). This means that the charged operator that condenses is not local, but rather
an extended operator, in this case extended along the direction of the stack of layers.
Second, among the list above of apparent exceptions to the Landau Paradigm, it re-
mains to discuss the Landau Fermi Liquid. (131) gives something like a symmetry-based
understanding of both Fermi liquids and a large class of non-Fermi liquids (for a review of
the latter, see (132)). First we assume translation symmetry, so that we may speak about
a well-defined Fermi surface in momentum space. The key ingredient is an emergent sym-
metry representing independent particle number conservation at each point on the Fermi
surface. In 2+1 dimensions, where the simplest Fermi surface is a circle, this is a loop
group symmetry; that is, the symmetry transformation is a map from the circle to U(1).
Such a loop-group symmetry emerges in the Landau theory, as well as in a large class of
non-Fermi liquids obtained by coupling a Fermi surface to gapless modes. (131) shows that
a state with a fractional and continuously-variable filling must have such a large symmetry.
From this starting point, the authors develop an understanding of Luttinger’s theorem as
an anomaly of this loop symmetry. (A related anomaly-based perspective on Luttinger’s
theorem appears in (133).) It shows that in a system with such a loop group symmetry, a
literal Fermi arc, i.e. a boundary of the Fermi surface, would imply a violation of charge
conservation: the Fermi surface must be the boundary of some region of the Brillouin zone.

7. Concluding remarks
Topological local operators. What about the case of (D − 1)-form symmetries in D
spacetime dimensions? This means that there are local operators that are topological. This
case is studied in (13, 14, 15) and more generally in (101, 134, 135). The conclusion is that
the Hilbert space of such a system is divided into superselection sectors with different values
of the topological operators. An example where this arises is in gauge theory in D = 1 + 1
without minimally-charged matter, where sectors represent different values of the electric
flux. (135) considers what happens when the action is perturbed by such operators, which

26 McGreevy
are always relevant. The perturbation changes the difference of the vacuum energies between
different sectors.
Higher groups. The concept of higher groups can be regarded as a natural extension of
higher-form symmetry (see e.g. (136) for a broader mathematical perspective). For example,
a 2-group structure can be defined in a physical context as follows: it is a modification of
the current algebra of a 1-form symmetry and a 0-form symmetry, so that the 0-form gauge
transformation acts nontrivially on the 2-form background field B for the 1-form symmetry:

A → A + dλ, B → B + κλdA, 39.

where A is the background 1-form field for the 0-form symmetry, and κ can be regarded as
a structure constant. This construction is closely related to the Green-Schwarz mechanism
of anomaly cancellation: suppose, for example, the effective action of a (1 + 1)D theory
with the above ingredients has an anomalous variation δλ S = κλ dA
R
2π
under a 0-form gauge
R B
transformation. Then the modified action S − 2π is invariant under the transformation
Eq. 39. Though it has not yet explicitly played a role in the condensed matter literature
to my knowledge, it appears in many places in QFT (e.g. (137, 138, 104, 139)) and we can
expect that it will be useful.
Other applications. In the preceding discussion, we have focussed on generalizations
of notions of symmetry as applied to zero-temperature groundstates of quantum matter. I
should mention that these same generalized symmetries have a number of other applications:

• A new organizing principle for magnetohydrodynamics (140, 141, 142, 143). More
generally, many kinds of exotic hydrodynamics can be understood by applying the
systematic logic of hydrodynamics to a system with generalized symmetries (see, for
example, (144)).
• (145) provides a nice example using both anomalies and generalized symmetries to
understand the spectrum of Goldstone modes of the Standard Model in a magnetic
field, and suggests a realization of the same physics in Dirac semimetals.
• More generally, more symmetry means more possible anomalies, and therefore new
anomaly constraints on IR behavior of QFT. For example, a mixed anomaly between
time-reversal symmetry and a 1-form symmetry implies an LSMOH constraint on
the groundstate of Yang-Mills theory at θ = π (146, 147, 148, 149). Work using
anomalies involving higher-form symmetries to constrain dynamics of QFT includes
(150, 151, 152, 153, 154, 155, 156, 157, 158, 159) and many others.

Disorder. I have not spoken about systems with disorder. Even if we are generally
interested in clean systems, it is important to ask about the stability of our statements to
the introduction of disorder. In the case of zero-form symmetries, the Imry-Ma argument
for stability of SSB proceeds by coupling the local order parameter to the disorder. Naively,
the inability to write such a coupling corroborates our expectation that higher-form SSB is
even more robust (78).
Dynamics. I have focused entirely on equilibrium phases of matter. Dynamics of
quantum matter is a current frontier, in which of course symmetries continue to play a
crucial role. A generalization of the notion of symmetry that has appeared in this context
is the phenomenon of Hilbert space fragmentation: this is what happens when the algebra of
operators that commute with each term of the Hamiltonian grows exponentially with system
size (160) (for systems with ordinary symmetries, this algebra grows only polynomially with
system size).

www.annualreviews.org • Generalized Symmetries 27

 Still beyond Landau? In this review, I’ve tried to motivate the following question:
Does the enlarged Landau paradigm (including all generalizations of symmetries, and their
anomalies) incorporate all equilibrium quantum phases of matter (and transitions between
them) as consequences of symmetry? Even if the answer is ‘no’, I think it has already
been a fruitful question. I close by enumerating some outstanding possible exceptions to
even the most generous interpretation in hopes of encouraging some further thought in this
direction.

• Symmetries can forbid all relevant operators that would lift gapless modes that are
however not Goldstones. An example is chiral symmetry in QCD, which forbids
fermion masses. A condensed matter example is the Dirac spin liquid – a phase
described by a CFT with no symmetric relevant operators.
• Above I argued that the DQCP between two distinct symmetry-breaking phases sat-
isfies Item 2 of the Landau Paradigm because it admits a description in terms of
a nonlinear sigma model whose fields are the order parameters of the two phases.
Ref. (161) generalizes this description to a sigma model on the Stiefel manifold, the
coset space SO(N + 4)/SO(4). For N = 1 this is the DQCP, for N = 2 they give evi-
dence that this is a description of a Dirac spin liquid in terms of only gauge-invariant
variables. The case N = 2 is called Stiefel liquid, and (161) provides a candidate
microscopic realization and argues that it has no weakly-coupled limit.
• An extremely interesting example of a claimed exception to Item 2 of even the Gener-
alized Landau Paradigm is provided by phase transitions described by IR-free gauge
theory (162). The claim of (162) is that SU(N ) gauge theory with adjoint fermions
(take N = 2) has a Z2 symmetry, and describes, as the fermion mass changes sign,
a completely novel critical theory for the transition from the trivial phase to the or-
dinary SSB phase. The degrees of freedom of this theory certainly go beyond the
fluctuations of the order parameter. Notice that for any nonzero mass there is an
extra emergent 1-form symmetry associated with the center of the gauge group. A
physical consequence of this symmetry (and a mixed anomaly), were it exact, would
be that a domain wall between the two Z2 -breaking vacua would satisfy an LSMOH
constraint: that is, the domain walls of the ordered phase would carry some extra de-
grees of freedom, and this would distinguish this phase from the ordinary SSB phase.
This symmetry is, however, explicitly broken by the massive charged matter of the
gauge theory.

Acknowledgements
I am deeply grateful to Nabil Iqbal for our collaboration, which has had a decisive influence
on the perspective advocated by this article. I would also like to thank Tarun Grover, Diego
Hofman, Jin-Long Huang, Zohar Nussinov, Mike Ogilvie, Gerardo Ortiz, Leo Radzihovsky,
T. Senthil, Shu-Heng Shao, Zhengdi Sun and David Tong for conversations about the ideas
in this review, and Xiang Li, Dachuan Lu, and Yi-Zhuang You for helpful comments on the
manuscript. This work was supported in part by funds provided by the U.S. Department of
Energy (D.O.E.) under cooperative research agreement DE-SC0009919, and by the Simons
Collaboration on Ultra-Quantum Matter, which is a grant from the Simons Foundation
(652264).

28 McGreevy
A. Spontaneous symmetry breaking and long-range order
ψ is not stationary under the symmetry (SSB) if and only if there exists a charged
operator O with ψ O ψ ̸= 0 (long-range order).
Proof:
⇐ Suppose the state is stationary under the symmetry, meaning

S ψ = eiα ψ . A.40.

Then for any charged operator O = eiγ S † OS, γ ∈

/ 2πZ,

ψ O ψ = eiγ ψ S † OS ψ = eiγ ψ O ψ , A.41.

which says ψ O ψ = 0, there is no long-range order.

11
⇒ Consider the reduced density matrix of a region X in the g roundstate:

X
ρX ≡ trX̄ |ψ⟩⟨ψ| = ⟨OI ⟩ OI , A.42.
I

Here {OI } is a basis of Hermitian operators on X, which we can choose to be orthonormal

with respect to the Hilbert-Schmidt inner product trOI OJ = δIJ . If no charged operator
has an expectation value, then the sum in Eq. A.42 only contains neutral operators. But
then SρX S † = ρX , so the state is invariant.
■
Notice that this argument makes no reference to the support of the symmetry operator
or its invertibility, and so works also for generalized symmetries.

B. SSB of higher-form symmetry without topological order, a confession

In this bonus appendix (added in v3), I want to explain a counterexample (27) to the
statement that SSB of a discrete one-form symmetry implies topological order, based very
closely on (26). The key point is that SSB of one-form symmetry

gs1 = WC gs2 B.43.

(or WC has a perimeter law, as we saw in the preceding appendix) is not quite a sufficient
condition for topological order. That is, Eq. B.43 with WC topological says that the second
groundstate can be obtained from the first by the action of an extended operator; but this
is not enough to guarantee that there isn’t also some local operator that relates them!
Here is a model, a deformation of the toric code, that provides a counterexample to
many simple and nice statements. It is due to Chamon and Castelnovo (27). The Hilbert
space is qubits on the links of an arbitrary cell complex, which let’s take to be the square
lattice for definiteness. The Hamiltonian is
X X
Hβ = + Qi − Bp B.44.
vertices i plaquettes p

11 I learned this argument from Tarun Grover.

www.annualreviews.org • Generalized Symmetries 29

 Q
where Bp ≡ ℓ∈∂p Zℓ is the usual toric code plaquette term and
P X
Qi ≡ e−β ℓ∈v(i) Zℓ
− Xℓ B.45.
ℓ∈v(i)

is a deformation of the star term that depends on a real parameter β. For β → 0 this
reduces to the usual toric code Hamiltonian (up to an additive constant).
Here are some facts about this model.
√

• The model has a phase transition at β = βc = 21 ln 1 + 2 . The TEE goes from
log 2 for β < βc to zero for β > βc . Thus, the phase at large β is not topologically
ordered12 .
• For any groundstate of the toric code gs(0) ,
Y
gs(β) ∝ eβZℓ /2 gs(0) B.46.
ℓ

is a groundstate of Hβ . This means that on the torus, there is a fourfold degeneracy

for every real β.
• In fact the system orders magnetically for β > βc . For one thing, the magnetic sus-
P
ceptibility ∂β ℓ Zℓ diverges at the transition. Further, there is magnetic ordering
in the sense that the different groundstates gs(β)ab , that come from the eigenstates
Q
of W (Cx,y ) = ℓ∈Cx,y Zℓ for various winding actually have expectation values of the
P
magnetization ℓ Zℓ that differ by amounts of order L. You can see this from the
fact that the wavefunctions weight contributions with different numbers of up spins
differently.
Thus, the different groundstates are actually distinguishable by local operators!
• The electric one-form symmetry operators WC = ℓ∈C Zℓ (for C a closed curve) still
Q

commute with Hβ (while the magnetic ones do not). For every β, there is SSB of this
one-form symmetry in the sense that the charged operators have a perimeter law

gs(β) VĈ gs(β) ∼ e−βℓ(Ĉ) B.47.

where ℓ(Ĉ) is the length of the curve Ĉ. (In fact WC also satisfies a perimeter law, but
VĈ is not a symmetry of Hβ , so this is not spontaneously breaking any symmetry.)

To see what is happening here, consider the following two requirements of topological
order (26). For all local operators O, and candidate orthonormal topological groundstates,

1. gs1 O gs1 = gs2 O gs2 .

2. gs1 O gs2 = 0.
Q
To see that these are distinct conditions, consider the states ⇑ ≡ x ↑ x and
Q
⇓ ≡ x ↓ x (which describe SSB of a 0-form Z2 symmetry). These satisfy condition
2 (since we must flip every spin to get from ⇑ to ⇓ , but not condition 1 (since
e.g. ⇑ Z1 ⇑ = 1 = − ⇓ Z1 ⇓ ). On the other hand, the fact that condition 2 is

12 Further evidence for this statement is the fact that the operator that creates a pair of e par-
Q
ticles W (C) = ℓ∈C Zℓ , for an open curve C, has a nonzero expectation value for β > βc . This
expectation value can be mapped to a correlation function in the Ising model between two spins at
the endpoints of C, which becomes long-ranged for β > βc .

30 McGreevy
satisfied is a basis-dependent statement – it is not true for a linear combination of these



states: Z1 ⇑ + ⇓ =− ⇑ − ⇓ .
In contrast, for SSB of 1-form symmetry we’ll see that we have condition 1 (in some
basis) but not necessarily condition 2. Note that if condition 1 holds in every basis of the
groundstate subspace, then condition 2 holds in every basis, and vice versa. Moreover, if
both 1 and 2 hold in some basis, then they both hold in any basis.
Proposition: If gs1 = WC gs2 for a one-form symmetry operator WC , then condition
1 holds.
Proof: For all local operators O

gs2 O gs2 = gs1 WC† OWC gs1 B.48.

= gs1 WC† ′ OWC ′ gs1 = gs1 WC† ′ WC ′ O gs1
= gs1 O gs1 .

In the second step we used the topological property of WC to deform it (if necessary) to
avoid the support of the local operator O. ■
So SSB of 1-form symmetry implies that, in some basis, condition 1 holds.
Proposition: If in addition, there’s a second 1-form symmetry VĈ that acts as an order
parameter to distinguish the states gs1 = WC gs2 , in the sense that

VĈ gsa = eiθa gsa , a = 1, 2 B.49.

with eiθ1 ̸= eiθ2 , then condition 2 also holds.

Another way to state the hypothesis is: some operator charged under the first one-form
symmetry is also topological.
Proof: The idea is similar to the relation between SSB and long-range order: For all
local operators O,

gs2 O gs1 = gs2 VĈ† OVĈ gs1 ei(θ2 −θ1 )

= gs2 VĈ† VĈ O gs1 ei(θ2 −θ1 ) = gs2 O gs1 ei(θ2 −θ1 ) . B.50.

Since ei(θ2 −θ1 ) ̸= 1 by assumption, we conclude that gs2 O gs1 = 0 . ■

A state very much like the Chamon-Castelnovo state is realized upon subjecting the toric
code to enough decoherence (37, 163, 164, 165, 166, 167). For weak-enough decoherence, the
topological order survives (37), a good sign for its usefulness for real quantum information
processing. At a certain threshold of decoherence strength, there is a phase transition
to a phase where the quantum information stored in the groundstates is lost. Strangely,
expectation values of observables are completely unchanged (since the decoherence can be
described by a finite-depth quantum channel), so for example, the Wilson loop still has a
perimeter law, and there is still SSB of a one-form symmetry. (In fact, the only way to
tell that something happened is by looking at quantities nonlinear in the density matrix.)
But beyond the threshold, the other one-form symmetry, with which the spontaneously-
broken one-form symmetry has a mixed anomaly, does not emerge. So in this phase, there
is no algebraic guarantee of a topological degeneracy, and indeed one can check (by various
measures nonlinear in the density matrix) (163, 164, 165, 166, 167) that the quantum
information is gone.

www.annualreviews.org • Generalized Symmetries 31

 LITERATURE CITED
1. X. G. Wen, “Topological orders in rigid states,” Int. J. Mod. Phys. B 04 (feb, 1990)
239–271, https://www.worldscientific.com/doi/abs/10.1142/S0217979290000139.
2. F. J. Wegner, “Duality in Generalized Ising Models and Phase Transitions without Local
Order Parameters,” Journal of Mathematical Physics 12 (1971), no. 10 2259–2272,
http://scitation.aip.org/content/aip/journal/jmp/12/10/10.1063/1.1665530.
3. J. Haah, “A degeneracy bound for homogeneous topological order,” SciPost Phys. 10 (2021),
no. 1 011, 2009.13551.
4. T. Grover, Y. Zhang, and A. Vishwanath, “Entanglement entropy as a portal to the physics
of quantum spin liquids,” New Journal of Physics 15 (2013), no. 2 025002, 1302.0899.
5. R. M. Nandkishore and M. Hermele, “Fractons,” Annual Review of Condensed Matter
Physics 10 (2019) 295–313, 1803.11196.
6. M. Pretko, X. Chen, and Y. You, “Fracton phases of matter,” International Journal of
Modern Physics A 35 (2020), no. 06 2030003, 2001.01722.
7. M. E. Peskin, “Mandelstam ’t Hooft Duality in Abelian Lattice Models,” Annals Phys. 113
(1978) 122.
8. C. Dasgupta and B. I. Halperin, “Phase Transition in a Lattice Model of
Superconductivity,” Phys. Rev. Lett. 47 (1981) 1556–1560.
9. N. Seiberg, “Electric - magnetic duality in supersymmetric nonAbelian gauge theories,”
Nucl. Phys. B 435 (1995) 129–146, hep-th/9411149.
10. Z. Nussinov and G. Ortiz, “Sufficient symmetry conditions for Topological Quantum Order,”
Proc. Nat. Acad. Sci. 106 (2009) 16944–16949, cond-mat/0605316.
11. Z. Nussinov and G. Ortiz, “A symmetry principle for topological quantum order,” Annals
Phys. 324 (2009) 977–1057, cond-mat/0702377.
12. A. Kovner and B. Rosenstein, “New look at QED in four-dimensions: The Photon as a
Goldstone boson and the topological interpretation of electric charge,” Phys. Rev. D 49
(1994) 5571–5581, hep-th/9210154.
13. T. Pantev and E. Sharpe, “Notes on gauging noneffective group actions,” hep-th/0502027.
14. S. Hellerman, A. Henriques, T. Pantev, E. Sharpe, and M. Ando, “Cluster decomposition,
T-duality, and gerby CFT’s,” Adv. Theor. Math. Phys. 11 (2007), no. 5 751–818,
hep-th/0606034.
15. E. Sharpe, “Notes on generalized global symmetries in QFT,” Fortsch. Phys. 63 (2015)
659–682, 1508.04770.
16. D. Gaiotto, A. Kapustin, N. Seiberg, and B. Willett, “Generalized Global Symmetries,”
JHEP 02 (2015) 172, 1412.5148.
17. D. M. Hofman and N. Iqbal, “Goldstone modes and photonization for higher form
symmetries,” 1802.09512.
18. L. V. Delacretaz, D. M. Hofman, and G. Mathys, “Superfluids as Higher-form Anomalies,”
1908.06977.
19. N. Iqbal and J. McGreevy, “Toward a 3d Ising model with a weakly-coupled string theory
dual,” SciPost Phys. 9 (2020), no. 2 019, 2003.04349.
20. J. Zhao, Z. Yan, M. Cheng, and Z. Y. Meng, “Higher-form symmetry breaking at Ising
transitions,” Physical Review Research 3 (July, 2021) 033024, 2011.12543.
21. A. Kapustin and R. Thorngren, “Higher symmetry and gapped phases of gauge theories,”
1309.4721.
22. C. Córdova, D. S. Freed, H. T. Lam, and N. Seiberg, “Anomalies in the Space of Coupling
Constants and Their Dynamical Applications I,” SciPost Phys. 8 (2020), no. 1 001,
1905.09315.
23. L. P. Kadanoff and H. Ceva, “Determination of an operator algebra for the two-dimensional
Ising model,” Physical Review B 3 (1971), no. 11 3918.
24. E. Fradkin, “Disorder Operators and their Descendants,” J. Statist. Phys. 167 (2017) 427,

32 McGreevy
 1610.05780.
25. X.-G. Wen, “Emergent anomalous higher symmetries from topological order and from
dynamical electromagnetic field in condensed matter systems,” Phys. Rev. B 99 (2019),
no. 20 205139, 1812.02517.
26. J. Huxford, D. X. Nguyen, and Y. B. Kim, “Gaining insights on anyon condensation and
1-form symmetry breaking across a topological phase transition in a deformed toric code
model,” SciPost Phys. 15 (2023), no. 6 253, 2305.07063.
27. C. Castelnovo and C. Chamon, “A quantum topological phase transition at the microscopic
level,” Physical Review B 77 (2008), no. 5 054433, arXiv:0707.2084.
28. A. Y. Kitaev, “Fault-tolerant quantum computation by anyons,” Annals of Physics 303
(Jan., 2003) 2–30, quant-ph/9707021.
29. E. Lake, “Higher-form symmetries and spontaneous symmetry breaking,” 1802.07747.
30. M. A. Levin and X.-G. Wen, “String net condensation: A Physical mechanism for
topological phases,” Phys. Rev. B71 (2005) 045110, cond-mat/0404617.
31. M. A. Levin and X.-G. Wen, “Colloquium: Photons and electrons as emergent phenomena,”
Rev. Mod. Phys. 77 (2005) 871–879, cond-mat/0407140.
32. A. Polyakov, “Quark confinement and topology of gauge theories,” Nuclear Physics B 120
(1977) 429–458.
33. S. Elitzur, “Impossibility of spontaneously breaking local symmetries,” Phys. Rev. D 12
(Dec, 1975) 3978–3982, https://link.aps.org/doi/10.1103/PhysRevD.12.3978.
34. D. Forster, H. B. Nielsen, and M. Ninomiya, “Dynamical Stability of Local Gauge
Symmetry: Creation of Light from Chaos,” Phys. Lett. B 94 (1980) 135–140.
35. M. B. Hastings and X.-G. Wen, “Quasiadiabatic continuation of quantum states: The
stability of topological ground-state degeneracy and emergent gauge invariance,” Physical
review b 72 (2005), no. 4 045141, cond-mat/0503554.
36. N. Iqbal and J. McGreevy, “Mean string field theory: Landau-Ginzburg theory for 1-form
symmetries,” 2106.12610.
37. E. Dennis, A. Kitaev, A. Landahl, and J. Preskill, “Topological quantum memory,” J. Math.
Phys. 43 (2002) 4452–4505, quant-ph/0110143.
38. A. A. Migdal, “Loop equations and 1/N expansion,” Physics Reports 102 (1983), no. 4
199–290.
39. T. Banks, “The Gaussian transformation converts lattice gauge theory into a field theory of
strings,” Phys. Lett. B 89 (1980) 369–372.
40. T. Yoneya, “A Path Functional Field Theory of Lattice Gauge Models and the Large N
Limit,” Nucl. Phys. B 183 (1981) 471–496.
41. S.-J. Rey, “The Higgs Mechanism for Kalb-ramond Gauge Field,” Phys. Rev. D 40 (1989)
3396.
42. M. Franz, “Vortex-boson duality in four space-time dimensions,” EPL 77 (2007), no. 4
47005, cond-mat/0607310.
43. A. J. Beekman, D. Sadri, and J. Zaanen, “Condensing Nielsen-Olesen strings and the vortex
boson duality in 3+1 and higher dimensions,” New Journal of Physics 13 (mar, 2011)
033004, https://doi.org/10.1088/1367-2630/13/3/033004.
44. A. Florio, J. a. M. V. P. L. J. Matos, and J. a. Penedones, “Searching for continuous phase
transitions in 5D SU(2) lattice gauge theory,” 2103.15242.
45. A. Kapustin, “Wilson-’t Hooft operators in four-dimensional gauge theories and S-duality,”
Phys. Rev. D74 (2006) 025005, hep-th/0501015.
46. J. L. Cardy, “UNIVERSAL PROPERTIES OF U(1) GAUGE THEORIES,” Nucl. Phys. B
170 (1980) 369–387.
47. M. Vettorazzo and P. de Forcrand, “Electromagnetic fluxes, monopoles, and the order of the
4-d compact U(1) phase transition,” Nucl. Phys. B 686 (2004) 85–118, hep-lat/0311006.
48. S. R. Coleman and E. J. Weinberg, “Radiative Corrections as the Origin of Spontaneous

www.annualreviews.org • Generalized Symmetries 33

 Symmetry Breaking,” Phys. Rev. D7 (1973) 1888–1910.
49. G. ’t Hooft, “Naturalness, chiral symmetry, and spontaneous chiral symmetry breaking,”
NATO Adv.Study Inst.Ser.B Phys. 59 (1980) 135.
50. X. Chen, Z.-C. Gu, Z.-X. Liu, and X.-G. Wen, “Symmetry protected topological orders and
the group cohomology of their symmetry group,” Physical Review B 87 (2013), no. 15
155114, 1106.4772.
51. T. Senthil, “Symmetry Protected Topological phases of Quantum Matter,”
Ann.Rev.Condensed Matter Phys. 6 (2015) 299, 1405.4015.
52. X.-G. Wen and A. Zee, “Classification of Abelian quantum Hall states and matrix
formulation of topoological fluids,” Phys. Rev. B 46 (1992) 2290–2301.
53. A. Zee, “Quantum Hall fluids,” Lect.Notes Phys. 456 (1995) 99–153, cond-mat/9501022.
54. A. Vishwanath and T. Senthil, “Physics of Three-Dimensional Bosonic Topological
Insulators: Surface-Deconfined Criticality and Quantized Magnetoelectric Effect,” Phys. Rev.
X 3 (Feb, 2013) 011016, 1209.3058, http://link.aps.org/doi/10.1103/PhysRevX.3.011016.
55. M. Oshikawa, “Topological Approach to Luttinger’s Theorem and the Fermi Surface of a
Kondo Lattice,” Phys. Rev. Lett. 84 (Apr., 2000) 3370–3373, cond-mat/0002392.
56. M. B. Hastings, “Lieb-Schultz-Mattis in higher dimensions,” Phys. Rev. B 69 (2004) 104431,
cond-mat/0305505.
57. D. V. Else and R. Thorngren, “Topological theory of Lieb-Schultz-Mattis theorems in
quantum spin systems,” Phys. Rev. B 101 (2020), no. 22 224437, 1907.08204.
58. A. Kapustin, “Bosonic Topological Insulators and Paramagnets: a view from cobordisms,”
1404.6659.
59. C. Z. Xiong, “Minimalist approach to the classification of symmetry protected topological
phases,” Journal of Physics A: Mathematical and Theoretical 51 (2018), no. 44 445001,
1701.00004.
60. D. V. Else and T. Senthil, “Critical drag as a mechanism for resistivity,” Phys. Rev. B 104
(2021), no. 20 205132, 2106.15623.
61. R. Kobayashi, K. Shiozaki, Y. Kikuchi, and S. Ryu, “Lieb-Schultz-Mattis type theorem with
higher-form symmetry and the quantum dimer models,” Phys. Rev. B 99 (2019), no. 1
014402, 1805.05367.
62. P.-S. Hsin, H. T. Lam, and N. Seiberg, “Comments on One-Form Global Symmetries and
Their Gauging in 3d and 4d,” SciPost Phys. 6 (2019), no. 3 039, 1812.04716.
63. F. J. Burnell, “Anyon condensation and its applications,” Ann. Rev. Condensed Matter
Phys. 9 (2018) 307–327, 1706.04940.
64. B. Shi, J.-L. Huang, and J. McGreevy, “Remote detectability from entanglement bootstrap I:
Kirby’s torus trick,” 2301.07119.
65. B. Yoshida, “Topological phases with generalized global symmetries,” Phys. Rev. B 93 (Apr,
2016) 155131, 1508.03468, https://link.aps.org/doi/10.1103/PhysRevB.93.155131.
66. L. Tsui and X.-G. Wen, “Lattice models that realize Zn -1-symmetry-protected topological
states for even n,” Phys. Rev. B 101 (2020), no. 3 035101, 1908.02613.
67. P.-S. Hsin, W. Ji, and C.-M. Jian, “Exotic Invertible Phases with Higher-Group
Symmetries,” 2105.09454.
68. M. Qi, L. Radzihovsky, and M. Hermele, “Fracton phases via exotic higher-form
symmetry-breaking,” Annals Phys. 424 (2021) 168360, 2010.02254.
69. X. Shen, Z. Wu, L. Li, Z. Qin, and H. Yao, “Fracton Topological Order at Finite
Temperature,” Phys. Rev. Res. 4 (2022) L032008, 2109.06887.
70. B. C. Rayhaun and D. J. Williamson, “Higher-Form Subsystem Symmetry Breaking:
Subdimensional Criticality and Fracton Phase Transitions,” 2112.12735.
71. W. Shirley, K. Slagle, Z. Wang, and X. Chen, “Fracton Models on General
Three-Dimensional Manifolds,” Physical Review X 8 (July, 2018) 031051, 1712.05892.
72. S. Vijay, J. Haah, and L. Fu, “Fracton topological order, generalized lattice gauge theory,

34 McGreevy
 and duality,” Phys. Rev. B 94 (Dec., 2016) 235157, 1603.04442.
73. M. Pretko, “Subdimensional Particle Structure of Higher Rank U(1) Spin Liquids,” Phys.
Rev. B 95 (2017), no. 11 115139, 1604.05329.
74. A. Gromov, “Towards classification of Fracton phases: the multipole algebra,” Phys. Rev. X
9 (2019), no. 3 031035, 1812.05104.
75. N. Seiberg and S.-H. Shao, “Exotic Symmetries, Duality, and Fractons in 2+1-Dimensional
Quantum Field Theory,” SciPost Phys. 10 (2021), no. 2 027, 2003.10466.
76. N. Seiberg and S.-H. Shao, “Exotic U (1) Symmetries, Duality, and Fractons in
3+1-Dimensional Quantum Field Theory,” SciPost Phys. 9 (2020), no. 4 046, 2004.00015.
77. N. Seiberg and S.-H. Shao, “Exotic ZN symmetries, duality, and fractons in 3+1-dimensional
quantum field theory,” SciPost Phys. 10 (2021), no. 1 003, 2004.06115.
78. C. Stahl, E. Lake, and R. Nandkishore, “Spontaneous breaking of multipole symmetries,”
2111.08041.
79. A. Paramekanti, L. Balents, and M. P. Fisher, “Ring exchange, the exciton Bose liquid, and
bosonization in two dimensions,” Phys. Rev. B 66 (Aug., 2002) 054526, cond-mat/0203171.
80. K. Slagle and Y. B. Kim, “Quantum Field Theory of X-Cube Fracton Topological Order and
Robust Degeneracy from Geometry,” Phys. Rev. B 96 (2017), no. 19 195139, 1708.04619.
81. D. Bulmash and M. Barkeshli, “The Higgs Mechanism in Higher-Rank Symmetric U (1)
Gauge Theories,” Phys. Rev. B 97 (2018), no. 23 235112, 1802.10099.
82. D. Bulmash and M. Barkeshli, “Generalized U (1) Gauge Field Theories and Fractal
Dynamics,” 1806.01855.
83. N. Seiberg, “Field Theories With a Vector Global Symmetry,” SciPost Phys. 8 (2020), no. 4
050, 1909.10544.
84. E. Lake, “Renormalization group and stability in the exciton Bose liquid,” Phys. Rev. B 105
(2022), no. 7 075115, 2110.02986.
85. T. Devakul, D. J. Williamson, and Y. You, “Classification of subsystem symmetry-protected
topological phases,” Phys. Rev. B 98 (2018), no. 23 235121, 1808.05300.
86. F. J. Burnell, T. Devakul, P. Gorantla, H. T. Lam, and S.-H. Shao, “Anomaly Inflow for
Subsystem Symmetries,” 2110.09529.
87. D. T. Stephen, J. Garre-Rubio, A. Dua, and D. J. Williamson, “Subsystem symmetry
enriched topological order in three dimensions,” Phys. Rev. Res. 2 (2020), no. 3 033331,
2004.04181.
88. J. Haah, “Local stabilizer codes in three dimensions without string logical operators,”
Phys. Rev. A 83 (Apr., 2011) 042330, 1101.1962.
89. D. J. Williamson, “Fractal symmetries: Ungauging the cubic code,” Phys. Rev. B 94 (2016),
no. 15 155128, 1603.05182.
90. B. Yoshida, “Exotic topological order in fractal spin liquids,” Phys. Rev. B 88 (2013), no. 12
125122, 1302.6248.
91. T. Devakul, Y. You, F. Burnell, and S. Sondhi, “Fractal Symmetric Phases of Matter,”
SciPost Physics 6 (2018), no. 1 007, 1805.04097.
92. M. Newman and C. Moore, “Glassy dynamics and aging in an exactly solvable spin model,”
Physical Review E 60 (1999), no. 5 5068.
93. L. M. Vasiloiu, T. H. Oakes, F. Carollo, and J. P. Garrahan, “Trajectory phase transitions in
noninteracting spin systems,” Physical Review E 101 (2020), no. 4 042115, 1911.11739.
94. Z. Zhou, X.-F. Zhang, F. Pollmann, and Y. You, “Fractal Quantum Phase Transitions:
Critical Phenomena Beyond Renormalization,” 2105.05851.
95. N. E. Myerson-Jain, S. Liu, W. Ji, C. Xu, and S. Vijay, “Pascal’s Triangle Fractal
Symmetries,” 2110.02237.
96. C.-M. Chang, Y.-H. Lin, S.-H. Shao, Y. Wang, and X. Yin, “Topological Defect Lines and
Renormalization Group Flows in Two Dimensions,” JHEP 01 (2019) 026, 1802.04445.
97. I. Cirac, D. Perez-Garcia, N. Schuch, and F. Verstraete, “Matrix Product States and

www.annualreviews.org • Generalized Symmetries 35

 Projected Entangled Pair States: Concepts, Symmetries, and Theorems,” 2011.12127.
98. J. Garre-Rubio, L. Lootens, and A. Molnár, “Classifying phases protected by matrix product
operator symmetries using matrix product states,” 2203.12563.
99. J. Kaidi, Z. Komargodski, K. Ohmori, S. Seifnashri, and S.-H. Shao, “Higher central charges
and topological boundaries in 2+1-dimensional TQFTs,” 2107.13091.
100. R. Thorngren and Y. Wang, “Fusion Category Symmetry I: Anomaly In-Flow and Gapped
Phases,” 1912.02817.
101. Z. Komargodski, K. Ohmori, K. Roumpedakis, and S. Seifnashri, “Symmetries and strings of
adjoint QCD2 ,” JHEP 03 (2021) 103, 2008.07567.
102. Y. Choi, C. Cordova, P.-S. Hsin, H. T. Lam, and S.-H. Shao, “Non-Invertible Duality Defects
in 3+1 Dimensions,” 2111.01139.
103. J. Kaidi, K. Ohmori, and Y. Zheng, “Kramers-Wannier-like duality defects in (3+1)d gauge
theories,” 2111.01141.
104. L. Bhardwaj and Y. Tachikawa, “On finite symmetries and their gauging in two dimensions,”
JHEP 03 (2018) 189, 1704.02330.
105. Y.-H. Lin and S.-H. Shao, “Duality Defect of the Monster CFT,” J. Phys. A 54 (2021), no. 6
065201, 1911.00042.
106. D. Gaiotto and J. Kulp, “Orbifold groupoids,” JHEP 02 (2021) 132, 2008.05960.
107. R. Thorngren and Y. Wang, “Fusion Category Symmetry II: Categoriosities at c = 1 and
Beyond,” 2106.12577.
108. K. Kikuchi, “Symmetry enhancement in RCFT,” 2109.02672.
109. I. M. Burbano, J. Kulp, and J. Neuser, “Duality Defects in E8 ,” 2112.14323.
110. S. Pal and Z. Sun, “High Energy Modular Bootstrap, Global Symmetries and Defects,”
JHEP 08 (2020) 064, 2004.12557.
111. K. Inamura, “On lattice models of gapped phases with fusion category symmetries,”
2110.12882.
112. A. Feiguin, S. Trebst, A. W. W. Ludwig, M. Troyer, A. Kitaev, Z. Wang, and M. H.
Freedman, “Interacting anyons in topological quantum liquids: The golden chain,” Phys.
Rev. Lett. 98 (2007), no. 16 160409, cond-mat/0612341.
113. D. Aasen, R. S. K. Mong, and P. Fendley, “Topological Defects on the Lattice I: The Ising
model,” J. Phys. A 49 (2016), no. 35 354001, 1601.07185.
114. D. Aasen, P. Fendley, and R. S. K. Mong, “Topological Defects on the Lattice: Dualities and
Degeneracies,” 2008.08598.
115. R. Vanhove, L. Lootens, M. Van Damme, R. Wolf, T. Osborne, J. Haegeman, and
F. Verstraete, “A critical lattice model for a Haagerup conformal field theory,” 2110.03532.
116. W. Ji and X.-G. Wen, “Categorical symmetry and noninvertible anomaly in
symmetry-breaking and topological phase transitions,” Phys. Rev. Res. 2 (2020), no. 3
033417, 1912.13492.
117. L. Kong, T. Lan, X.-G. Wen, Z.-H. Zhang, and H. Zheng, “Algebraic higher symmetry and
categorical symmetry: A holographic and entanglement view of symmetry,” Physical Review
Research 2 (Oct., 2020) 043086, 2005.14178.
118. A. Chatterjee and X.-G. Wen, “Algebra of local symmetric operators and braided fusion
n-category – symmetry is a shadow of topological order,” 2203.03596.
119. A. M. Polyakov, “Gauge Fields and Strings,” Contemp. Concepts Phys. 3 (1987) 1–301.
120. C. Wang, A. Nahum, M. A. Metlitski, C. Xu, and T. Senthil, “Deconfined quantum critical
points: symmetries and dualities,” Phys. Rev. X 7 (2017), no. 3 031051, 1703.02426.
121. A. Tanaka and X. Hu, “Many-body spin berry phases emerging from the π-flux state:
Competition between antiferromagnetism and the valence-bond-solid state,” Physical review
letters 95 (2005), no. 3 036402, cond-mat/0501365.
122. T. Senthil and M. P. Fisher, “Competing orders, nonlinear sigma models, and topological
terms in quantum magnets,” Physical Review B 74 (2006), no. 6 064405, cond-mat/0510459.

36 McGreevy
123. M. Levin and T. Senthil, “Deconfined quantum criticality and Néel order via dimer
disorder,” Phys. Rev. B 70 (Dec., 2004) 220403, cond-mat/0405702.
124. Y.-C. Wang, M. Cheng, and Z. Y. Meng, “Scaling of the disorder operator at (2+1)d U(1)
quantum criticality,” Phys. Rev. B 104 (2021), no. 8 081109, 2101.10358.
125. Y.-C. Wang, N. Ma, M. Cheng, and Z. Y. Meng, “Scaling of disorder operator at deconfined
quantum criticality,” 2106.01380.
126. X. Qiu, R. Joynt, and A. MacDonald, “Phases of the multiple quantum well in a strong
magnetic field: Possibility of irrational charge,” Physical Review B 40 (1989), no. 17 11943.
127. J. D. Naud, L. P. Pryadko, and S. L. Sondhi, “Fractional Quantum Hall Effect in
Infinite-Layer Systems,” Phys. Rev. Lett. 85 (Dec., 2000) 5408–5411, cond-mat/0006432.
128. X. Ma, W. Shirley, M. Cheng, M. Levin, J. McGreevy, and X. Chen, “Fractonic order in
infinite-component Chern-Simons gauge theories,” 2010.08917.
129. J. Sullivan, A. Dua, and M. Cheng, “Weak symmetry breaking and topological order in a 3D
compressible quantum liquid,” 2109.13267.
130. C. Wang and M. Levin, “Weak symmetry breaking in two-dimensional topological
insulators,” Phys. Rev. B 88 (Dec., 2013) 245136, 1311.0767.
131. D. V. Else, R. Thorngren, and T. Senthil, “Non-Fermi liquids as ersatz Fermi liquids: general
constraints on compressible metals,” Phys. Rev. X 11 (2021), no. 2 021005, 2007.07896.
132. S.-S. Lee, “Recent Developments in Non-Fermi Liquid Theory,” Annual Review of
Condensed Matter Physics 9 (Mar., 2018) 227–244, 1703.08172.
133. R. Ma and C. Wang, “Emergent anomaly of Fermi surfaces: a simple derivation from Weyl
fermions,” 2110.09492.
134. D. Delmastro, J. Gomis, and M. Yu, “Infrared phases of 2d QCD,” 2108.02202.
135. A. Cherman, T. Jacobson, and M. Neuzil, “Universal Deformations,” 2111.00078.
136. J. C. Baez and A. D. Lauda, “Higher-Dimensional Algebra V: 2-Groups,” arXiv
Mathematics e-prints (July, 2003) math/0307200, math/0307200.
137. C. Córdova, T. T. Dumitrescu, and K. Intriligator, “Exploring 2-Group Global Symmetries,”
JHEP 02 (2019) 184, 1802.04790.
138. F. Benini, C. Córdova, and P.-S. Hsin, “On 2-Group Global Symmetries and their
Anomalies,” JHEP 03 (2019) 118, 1803.09336.
139. N. Iqbal and N. Poovuttikul, “2-group global symmetries, hydrodynamics and holography,”
2010.00320.
140. S. Grozdanov, D. M. Hofman, and N. Iqbal, “Generalized global symmetries and dissipative
magnetohydrodynamics,” Phys. Rev. D 95 (2017), no. 9 096003, 1610.07392.
141. J. Armas and A. Jain, “One-form superfluids & magnetohydrodynamics,” 1811.04913.
142. J. Armas and A. Jain, “Magnetohydrodynamics as superfluidity,” Phys. Rev. Lett. 122
(2019), no. 14 141603, 1808.01939.
143. S. Grozdanov and N. Poovuttikul, “Generalised global symmetries and
magnetohydrodynamic waves in a strongly interacting holographic plasma,” 1707.04182.
144. S. Grozdanov and N. Poovuttikul, “Generalized global symmetries in states with dynamical
defects: The case of the transverse sound in field theory and holography,” Phys. Rev. D97
(2018), no. 10 106005, 1801.03199.
145. N. Sogabe and N. Yamamoto, “Triangle Anomalies and Nonrelativistic Nambu-Goldstone
Modes of Generalized Global Symmetries,” Phys. Rev. D99 (2019), no. 12 125003,
1903.02846.
146. D. Gaiotto, A. Kapustin, Z. Komargodski, and N. Seiberg, “Theta, Time Reversal, and
Temperature,” JHEP 05 (2017) 091, 1703.00501.
147. Z. Wan, J. Wang, and Y. Zheng, “New higher anomalies, SU(N) Yang–Mills gauge theory
and CPN−1 sigma model,” Annals Phys. 414 (2020) 168074, 1812.11968.
148. Z. Wan, J. Wang, and Y. Zheng, “Quantum 4d Yang-Mills Theory and Time-Reversal
Symmetric 5d Higher-Gauge Topological Field Theory,” Phys. Rev. D 100 (2019), no. 8

www.annualreviews.org • Generalized Symmetries 37

 085012, 1904.00994.
149. C. Córdova and K. Ohmori, “Anomaly Obstructions to Symmetry Preserving Gapped
Phases,” 1910.04962.
150. D. Gaiotto, Z. Komargodski, and N. Seiberg, “Time-reversal breaking in QCD4 , walls, and
dualities in 2 + 1 dimensions,” JHEP 01 (2018) 110, 1708.06806.
151. R. Kitano, T. Suyama, and N. Yamada, “θ = π in SU (N )/ZN gauge theories,” JHEP 09
(2017) 137, 1709.04225.
152. Y. Tanizaki, Y. Kikuchi, T. Misumi, and N. Sakai, “Anomaly matching for phase diagram of
massless ZN -QCD,” 1711.10487.
153. Z. Komargodski, A. Sharon, R. Thorngren, and X. Zhou, “Comments on Abelian Higgs
Models and Persistent Order,” 1705.04786.
154. M. M. Anber and E. Poppitz, “Two-flavor adjoint QCD,” Phys. Rev. D 98 (2018), no. 3
034026, 1805.12290.
155. Z. Wan and J. Wang, “Adjoint QCD4 , Deconfined Critical Phenomena, Symmetry-Enriched
Topological Quantum Field Theory, and Higher Symmetry-Extension,” Phys. Rev. D 99
(2019), no. 6 065013, 1812.11955.
156. C. Córdova and T. T. Dumitrescu, “Candidate Phases for SU(2) Adjoint QCD4 with Two
Flavors from N = 2 Supersymmetric Yang-Mills Theory,” 1806.09592.
157. A. Cherman, T. Jacobson, Y. Tanizaki, and M. Ünsal, “Anomalies, a mod 2 index, and
dynamics of 2d adjoint QCD,” SciPost Phys. 8 (2020), no. 5 072, 1908.09858.
158. A. A. Cox, E. Poppitz, and F. D. Wandler, “The mixed 0-form/1-form anomaly in Hilbert
space: pouring the new wine into old bottles,” JHEP 10 (2021) 069, 2106.11442.
159. M. Nguyen, Y. Tanizaki, and M. Ünsal, “Noninvertible 1-form symmetry and Casimir scaling
in 2D Yang-Mills theory,” Phys. Rev. D 104 (2021), no. 6 065003, 2104.01824.
160. S. Moudgalya and O. I. Motrunich, “Hilbert Space Fragmentation and Commutant
Algebras,” arXiv e-prints (Aug., 2021) arXiv:2108.10324, 2108.10324.
161. L. Zou, Y.-C. He, and C. Wang, “Stiefel Liquids: Possible Non-Lagrangian Quantum
Criticality from Intertwined Orders,” Phys. Rev. X 11 (2021), no. 3 031043, 2101.07805.
162. Z. Bi, E. Lake, and T. Senthil, “Landau ordering phase transitions beyond the Landau
paradigm,” Phys. Rev. Res. 2 (2020), no. 2 023031, 1910.12856.
163. R. Fan, Y. Bao, E. Altman, and A. Vishwanath, “Diagnostics of Mixed-State Topological
Order and Breakdown of Quantum Memory,” PRX Quantum 5 (2024), no. 2 020343,
2301.05689.
164. Y. Bao, R. Fan, A. Vishwanath, and E. Altman, “Mixed-state topological order and the
errorfield double formulation of decoherence-induced transitions,” 2301.05687.
165. Y.-H. Chen and T. Grover, “Separability Transitions in Topological States Induced by Local
Decoherence,” Phys. Rev. Lett. 132 (2024), no. 17 170602, 2309.11879.
166. J. Y. Lee, “Exact Calculations of Coherent Information for Toric Codes under Decoherence:
Identifying the Fundamental Error Threshold,” 2402.16937.
167. Y.-H. Chen and T. Grover, “Unconventional topological mixed-state transition and critical
phase induced by self-dual coherent errors,” Phys. Rev. B 110 (2024), no. 12 125152,
2403.06553.

38 McGreevy