---
title: Crossed n-Cubes of Groups
url: https://www.emergentmind.com/topics/crossed-n-cubes-of-groups
type: topic
---

# Crossed n-Cubes of Groups

Crossed \(n\)-cubes of groups are algebraic structures used to organize local symmetry data, overlap data, commutator data, and higher compatibility conditions into a model of a higher group. In the setting of lattice gauge theory, they arise from the operator-algebraic analysis of symmetry transformations localized on spatial subregions and of the ambiguities that appear when such transformations are restricted to intersections, triple intersections, and higher overlaps. In two spatial dimensions, the relevant structure is a crossed square of groups, which is an algebraic model of a connected homotopy \(3\)-type, hence of a \(3\)-group; in general spatial dimension \(d\), the corresponding structure is a crossed \(d\)-cube [2507.16966].

## 1. Localized symmetries and the need for crossed \(n\)-cubes

The operator-algebraic starting point is that a symmetry of a lattice system is treated as an automorphism of the algebra of observables \(\mathcal A\), rather than as a unitary operator on a global Hilbert space. For an infinite system, one cannot usually implement a symmetry by a single global unitary operator, so the natural objects are automorphisms of \(\mathcal A\); these are the kinematic symmetries in the formulation under discussion [2507.16966].

The decisive additional feature is locality. If \(U\subset V\), then a symmetry supported in \(U\) gives one supported in \(V\), but this restriction is not unique. There are ambiguities on pairwise overlaps, ambiguities of ambiguities on triple overlaps, and further coherence data at higher intersections. The resulting structure is not adequately described by an ordinary group. Instead, the data of localized transformations, their restrictions, their commutators, and the higher compatibility conditions among them are packaged by a nonabelian Čech-type construction into a crossed \(n\)-cube.

This viewpoint leads to a precise conceptual claim: higher-form symmetries and ’t Hooft anomalies in lattice gauge theory are best understood by examining how symmetry transformations localize to subregions. The higher-group structure is therefore not an auxiliary reformulation but an algebraic expression of locality itself. A plausible implication is that whenever a lattice model admits sufficiently rich regionwise localization data, one should expect higher-group structures to emerge systematically rather than exceptionally.

## 2. Homotopy-theoretic interpretation

The homotopy-theoretic motivation is expressed through Postnikov towers. A connected homotopy type \(\mathfrak X\) is built from its homotopy groups \(\pi_i\) and Postnikov classes \(k_i\), with a tower
\[
\cdots \to \mathfrak X_n \xrightarrow{f_n} \mathfrak X_{n-1}\to \cdots \to \mathfrak X_1\to pt,
\]
and fibrations
\[
K(\pi_n,n)\to \mathfrak X_n \xrightarrow{f_n} \mathfrak X_{n-1}.
\]
Each fibration is classified by a class
\[
k_n \in H^{n+1}(\mathfrak X_{n-1},\pi_n).
\]

Within this framework, ordinary symmetries correspond to \(K(G,1)=BG\), while a \(p\)-form symmetry with group \(G\) corresponds to \(K(G,p+1)\). Nontrivial Postnikov classes encode the interaction between lower and higher symmetries. The same language also describes anomalies: an ’t Hooft anomaly is interpreted as an obstruction to splitting or lifting such fibrations [2507.16966].

This interpretation places crossed \(n\)-cubes within the standard homotopy-theoretic description of higher groups. Their role is to provide algebraic models of higher homotopy types whose homotopy groups are identified with symmetry sectors, while the higher commutator and coherence data encode the Postnikov information. In this sense, crossed \(n\)-cubes furnish the algebraic realization of the statement that localized symmetries assemble into a higher group.

## 3. Crossed squares as the two-dimensional case

In two spatial dimensions, the relevant structure is a crossed square, which models a connected homotopy \(3\)-type and hence a \(3\)-group [2507.16966]. The definition is preceded by the crossed module \((H,G,\partial)\), with
\[
\partial:H\to G,
\]
together with an action of \(G\) on \(H\) satisfying
\[
\partial({}^g h)=g(\partial h)g^{-1},\qquad {}^{\partial h_0}\!h_1=h_0h_1h_0^{-1}.
\]
This models a \(2\)-group, with
\[
\pi_1 = G/\partial(H),\qquad \pi_2=\ker \partial.
\]

A crossed square is a diagram
\[
\begin{CD}
L @>f>> M \\
@VgVV @VVvV \\
N @>u>> P
\end{CD}
\]
equipped with an action of \(P\) on \(L,M,N\), and a function
\[
\eta : M\times N \to L
\]
satisfying the compatibility relations
\[
\begin{aligned}
f(\eta(m,n))&=m\,{}^{u(n)}\!m^{-1}, & g(\eta(m,n))&={}^{v(m)}\!n\,n^{-1}, \\
\eta(f(l),n)&=l\,{}^{u(n)}\!l^{-1}, & \eta(m,g(l))&={}^{v(m)}\!l\,l^{-1}, \\
\eta(mm',n)&={}^{v(m)}\!\eta(m',n)\eta(m,n), & \eta(m,nn')&=\eta(m,n)\,{}^{u(n)}\!\eta(m,n'), \\
\eta({}^p\!m,{}^p\!n)&={}^p\!\eta(m,n).
\end{aligned}
\]

The map \(\eta\) measures the failure of the left and right local pieces to commute strictly; it is a coherent commutator-like defect. From the crossed square one forms
\[
Q=M\rtimes N,
\]
and defines
\[
\delta:L\to Q,\qquad \partial:Q\to P
\]
by
\[
\delta(l)=(f(l)^{-1},g(l)),\qquad \partial(m,n)=v(m)u(n).
\]
This gives the complex of nonabelian groups
\[
L \xrightarrow{\delta} M\rtimes N \xrightarrow{\partial} P.
\]
The associated homotopy groups are
\[
\pi_1 = \mathrm{coker}(\partial),\qquad \pi_2 = \ker(\partial)/\mathrm{im}(\delta),\qquad \pi_3 = \ker(\delta).
\]

The crossed square contains additional coherence data. From \(\eta\) one defines
\[
\{(m,n),(m',n')\}=\eta(m,n n' n^{-1})^{-1},
\]
and for \(x\in \ker\partial\),
\[
q(x)=\{x,x\}.
\]
This induces a quadratic function
\[
q:\pi_2 \to \pi_3.
\]
The paper proves that \(q\) is well defined on \(\pi_2=\ker\partial/\mathrm{im}\,\delta\), lands in \(\pi_3=\ker\delta\), and satisfies the quadratic identities. In the physical interpretation developed there, this quadratic datum detects anomalies.

## 4. Nonabelian Čech structure in operator algebras

The operator-algebraic realization of crossed \(n\)-cubes is formulated in terms of groups of localized symmetries. For a region \(U\), let \((U)\) denote the group of symmetries localized near \(U\), more precisely approximately localized on a thickening of \(U\). If \(U\subset U'\), then \((U)\to (U')\) is an inclusion of a normal subgroup. Moreover, if \(g\in (U)\) and \(g'\in(U')\), then
\[
[g,g'] \in (U\cap U').
\]
This is exactly the kind of algebraic behavior encoded by a crossed cube [2507.16966].

For localized finite-support transformations, one may represent the transformation by a unitary observable \(u\), and the corresponding automorphism is
\[
{}_u(a)=uau^{-1}.
\]
However, when one considers region-restricted transformations and their overlap behavior, ordinary group structure ceases to be sufficient. One needs groups attached to regions, overlaps, and higher overlaps, together with boundary maps and commutator maps. This is why the construction is described as a nonabelian version of the usual Čech picture.

The resulting interpretation is geometric as well as algebraic. If space is covered by suitably chosen cones or simplex-based regions, then the groups of automorphisms localized on their intersections assemble into a crossed \(n\)-cube. The structure maps \(\mu_i\) are inclusions or conjugation maps, and the higher pairings record commutators or their automorphism-level analogues. This suggests that crossed \(n\)-cubes are adapted specifically to the locality structure of lattice models rather than being imposed externally.

## 5. Crossed squares in \((2+1)\)-dimensional lattice gauge theory

The construction is developed concretely for the \((2+1)\)-dimensional \(\mathbb Z_2\) lattice gauge theory on the square lattice. There is a qubit on each edge \(e\), with Pauli operators \(X_e,Y_e,Z_e\). The Gauss law at a vertex \(v\) is
\[
A_v=\prod_{e\supset v} X_e,
\]
gauge-invariant observables are generated by the \(X_e\) and the plaquette operators
\[
B_p=\prod_{e\subset p} Z_e,
\]
and the Hamiltonian is
\[
H=-g^2\sum_e X_e-\frac{1}{g^2}\sum_p B_p.
\]

The crossed square is built from the following groups: \(P\), gauge-invariant circuits approximately localized in the lower half-plane; \(M\), circuits localized on the left half of the \(x\)-axis; \(N\), circuits localized on the right half of the \(x\)-axis; and \(L\), unitary gauge-invariant observables. Then \(M\) and \(N\) are normal subgroups of \(P\), \(P\) acts on \(L,M,N\), the maps \(M\to P\) and \(N\to P\) are inclusions, and the maps \(L\to M,N\) are conjugations \(u\mapsto {}_u\). The map \(\eta:M\times N\to L\) is defined by the commutator of the left- and right-localized circuits; it is the localized defect produced by splitting a global circuit into regional pieces. This is the kinematic crossed square [2507.16966].

Its homotopy groups have direct physical interpretations. One has
\[
\pi_3 = \ker\delta = U(1),
\]
reflecting the scalar center. The group \(\pi_1\) is large and nontrivial, containing global automorphisms such as those that flip infinitely many plaquettes. The group \(\pi_2\) contains a distinguished \(\mathbb Z_2\) subgroup generated by a string automorphism. For a dual-lattice path \(\gamma\), the \(X\)-string automorphism \(\mu^X_\gamma\) flips the sign of plaquettes at the endpoint of \(\gamma\), and the generator of the \(1\)-form symmetry is represented by
\[
(\mu^X_{-},\mu^X_{+})\in M\rtimes N,
\]
where \(\mu^X_{+}\) and \(\mu^X_{-}\) are strings extending to \(+\infty\) and \(-\infty\). This gives a concrete realization of a higher-form symmetry as an element of \(\pi_2\).

For the untwisted theory, the anomaly is trivial. The distinguished \(\mathbb Z_2\) element \(x\in \pi_2\) satisfies
\[
q(x)=\{x,x\}=1.
\]
Accordingly, the \(1\)-form \(\mathbb Z_2\) symmetry is anomaly-free.

## 6. Twists, anomalies, and higher-dimensional generalization

A modified Gauss law produces a different quadratic datum. In the twisted \(\mathbb Z_2\) theory,
\[
A'_v=B_{NE(v)}\prod_{e\supset v}X_e.
\]
The gauge-invariant observables are then generated by \(B_p\) and modified edge operators
\[
U_e = X_e Z_{r(e)}.
\]
The same crossed-square construction applies, but now the distinguished string automorphism has nontrivial commutator behavior. For the generator \(\mathbf x\in \pi_2\),
\[
q(\mathbf x^n)=\exp(i\pi n^2).
\]
Equivalently, the basic commutator is \(-1\), so the \(1\)-form symmetry is anomalous [2507.16966].

The same paper identifies a Euclidean counterpart,
\[
S=\frac{1}{g^2}\sum_p (\delta a)_p + i\pi \int a\cup \delta a,
\]
with anomaly inflow action
\[
S_{\text{top}=i\pi \int B\cup B.
\]
It interprets the resulting defect-line structure as the category of super vector spaces, so that the generator of the symmetry behaves like an odd line. Within the framework at hand, this is the operator-algebraic manifestation of the familiar statement that the theory has fermionic vortex excitations.

The construction extends to \(\mathbb Z_n\) gauge theory, with clock and shift operators \(T,S\) satisfying
\[
T^n=S^n=1,\qquad ST=\lambda TS,\qquad \lambda=e^{2\pi i/n},
\]
and modified Gauss law
\[
B^s_{NE(v)}\prod_{e\supset v}S_e=1.
\]
The crossed square again yields a distinguished \(\mathbb Z_n\subset \pi_2\), and the anomaly is
\[
q(\mathbf x^m)=\exp\!\left(\frac{2 i\pi s m^2}{n}\right).
\]
For \(\mathbb Z_n\times \mathbb Z_n\) gauge theory there are two kinds of edge variables and two Gauss laws,
\[
{}^s_{NE(v)}\prod_{e\supset v} S_e=1,\qquad B^s_{NE(v)}\prod_{e\supset v} \hat S_e=1,
\]
producing a \(\mathbb Z_n\times \mathbb Z_n\) \(1\)-form symmetry with anomaly quadratic form
\[
q(\mathbf x^m \hat{\mathbf x}^k)=\exp\!\left(\frac{4 i\pi s mk}{n}\right).
\]
The paper notes that this vanishes for \(n=2\), but for general \(n\) it may be nontrivial.

In general spatial dimension \(d\), the relevant object is a crossed \(d\)-cube of groups. Such a crossed \(n\)-cube consists of groups \(M_A\) indexed by subsets \(A\subset \langle n\rangle=\{1,\dots,n\}\), together with structure maps
\[
\mu_i : M_A \to M_{A\setminus\{i\}},
\]
and commutator-like pairings
\[
h: M_A\times M_B \to M_{A\cup B},
\]
subject to identities generalizing those of crossed squares. A crossed \(1\)-cube is the same as a crossed module, and a crossed \(2\)-cube is the same as a crossed square. From a crossed \(d\)-cube one constructs a chain complex
\[
C_d \xrightarrow{\partial_d} C_{d-1}\xrightarrow{\partial_{d-1}\cdots \xrightarrow{\partial_1} C_0,
\]
with
\[
\pi_1(C_*)=\operatorname{coker}\partial_1,\qquad \pi_k(C_*)=\ker\partial_{k-1}/\operatorname{im}\partial_k \quad (2\le k\le d),\qquad \pi_d(C_*)=\ker\partial_d.
\]
These groups are interpreted as higher-form symmetry groups, while the higher commutator data \(h(a,b)\) encodes the higher Postnikov data and therefore the possible anomalies. The overall conclusion is that crossed \(n\)-cubes provide the algebraic mechanism by which local operator data becomes a higher group of symmetries, with anomalies appearing as the higher Postnikov or quadratic data of that higher group [2507.16966].

Source: https://www.emergentmind.com/topics/crossed-n-cubes-of-groups