---
title: Haag Duality Violations in QFT
url: https://www.emergentmind.com/topics/haag-duality-violations-hdv
type: topic
---

# Haag Duality Violations in QFT

Searching arXiv for the cited papers and closely related work on Haag duality violations.
Haag duality violations (HDV) are strict failures of the equality between a local algebra and the commutant of the algebra of the causal complement in algebraic quantum field theory. In the Haag–Kastler framework, locality implies an inclusion of the form $\mathcal{A}(O') \subseteq \mathcal{A}(O)'$, or equivalently $\mathcal{A}(O') \subseteq \mathcal{A}(O)'$ in the notation used across the recent literature, but Haag duality requires this inclusion to be saturated. An HDV is therefore the strict inclusion $\mathcal{A}(O) \subsetneq \hat{\mathcal{A}}(O)$ with $\hat{\mathcal{A}}(O)=\mathcal{A}(O')'$, or, in the restricted-sector language, the existence of an operator in the commutant that is not generated by the algebra of the complement [2503.20863], [2511.21810]. Recent work has made HDV a central diagnostic for the interplay between local operator algebras, generalized global symmetries, topological defect operators, and nonlocal observables in both continuum QFT and lattice models, especially in situations involving non-invertible symmetries, higher-form symmetries, and gauge-theoretic confinement [2503.20863], [2509.03589].

## 1. Algebraic definition and operator-algebraic meaning

In the algebraic approach to quantum field theory one assigns to each causally complete spacetime region $O$ a von Neumann algebra $\mathcal{A}(O)$ of bounded operators on a common Hilbert space $\mathcal{H}$. Locality, or Einstein causality, implies
\[
\mathcal{A}(O') \subseteq \mathcal{A}(O)' ,
\]
where $O'$ is the causal complement and the prime on the algebra denotes the commutant in $B(\mathcal{H})$ [2503.20863]. Haag duality is the stronger statement
\[
\mathcal{A}(O')=\mathcal{A}(O)' .
\]
Intuitively, Haag duality means that the algebra on $O$ is maximal subject to locality: nothing more can commute with $\mathcal{A}(O)$ than $\mathcal{A}(O')$ itself [2503.20863].

A complementary formulation used in the higher-form and reconstruction literature introduces the additive algebra
\[
\mathcal{A}(O)=\bigvee_{B\subset O}\mathcal{A}(B)
\]
and the dual algebra
\[
\hat{\mathcal{A}}(O)=\mathcal{A}(O')' .
\]
Haag duality holds when $\mathcal{A}(O)=\hat{\mathcal{A}}(O)$, and an HDV is the strict inclusion
\[
\mathcal{A}(O)\subsetneq \hat{\mathcal{A}}(O) .
\]
The extra operators in $\hat{\mathcal{A}}(O)\setminus \mathcal{A}(O)$ are the non-local or HDV operators of $O$ [2511.21810].

In symmetry-restricted settings the same phenomenon appears as a mismatch between the commutant of the restricted algebra and the restricted algebra of the complement. If a theory carries a global symmetry algebra $S$, which may be a group or a fusion category of topological defect lines, the symmetric subalgebra is defined by
\[
A_S(O)\coloneqq \{X\in A(O)\mid X \text{ commutes with every element of }S\}.
\]
Isotony is preserved and still $A_S(O')\subseteq A_S(O)'$ by locality, but generally one finds
\[
A_S(O')\subset A_S(O)' ,
\]
so Haag duality is violated in the $S$-symmetric sector [2503.20863]. Equivalently, HDV is the existence of an operator $D\in A_S(O)'$ with $D\notin A_S(O')$ [2503.20863].

This operator-algebraic perspective makes HDV a precise statement about incompleteness of a local algebra with respect to all operators commuting with it. In the recent literature, this incompleteness is tied to topological defect operators, Wilson and ’t Hooft operators, bilocal neutral composites, and generalized order/disorder operators [2503.20863], [2509.03589], [2511.21810], [2507.14699], [2508.09172].

## 2. Relation to additivity, disjoint additivity, and locality

HDV is closely connected to the algebraic completeness property known as additivity. In one standard formulation,
\[
\mathcal{A}(O_1\cup O_2)=\mathcal{A}(O_1)\vee \mathcal{A}(O_2),
\]
where $\mathcal{A}(O_1)\vee\mathcal{A}(O_2)$ is the von Neumann algebra generated by the two algebras [2509.03589]. The recent analysis of local quantum physics distinguishes this from the weaker requirement of disjoint additivity, imposed only when $O_1$ and $O_2$ are spatially disjoint [2509.03589].

The central point is that unbreakable extended operators can force a clash between additivity and Haag duality. If an operator $W$ supported on a closed submanifold $\Sigma\subset O$ commutes with $\mathcal{A}(O')$ because it is topological relative to $O$, yet cannot be generated by local operators in an overlapping cover of $O$, then one must either include $W$ in $\mathcal{A}(O)$, in which case additivity fails, or exclude $W$ from $\mathcal{A}(O)$, in which case Haag duality fails [2509.03589]. This is the mechanism by which higher-form symmetries produce algebraic tension without necessarily implying nonlocality in the path-integral or lattice sense [2509.03589].

A more recent proposal resolves this tension by retaining Haag duality while weakening ordinary additivity to disjoint additivity. The proposal is that “Haag duality + disjoint additivity” is the algebraic signature of genuinely local quantum physics, and the paper gives examples in which ordinary additivity fails while disjoint additivity is preserved [2509.03589]. The same work also proves a lattice theorem: for a compact Lie group $S$ acting faithfully on a tensor-product Hilbert space with factorized regional action $U(s)=U_R(s)\otimes U_{R'}(s)$, after imposing the constraint to the $S$-invariant subspace, disjoint additivity holds for “non-adjacent” regions satisfying a specific subgroup factorization criterion [2509.03589].

Within this framework HDV does not merely record a pathology. It differentiates several algebraic possibilities. Free Maxwell theory and toric-code–type lattice gauge theories violate ordinary additivity but satisfy the weaker disjoint additivity, whereas hypersurface truncations and generalized free fields violate Haag duality outright [2509.03589]. This suggests that HDV can separate theories with incomplete local algebras from theories that remain local after replacing additivity by the weaker disjoint form.

## 3. Symmetry restriction and non-invertible sectors in 1+1 dimensions

A major recent source of HDV is restriction to the symmetric sector of a global symmetry algebra, especially when the symmetry is non-invertible. In a bosonic diagonal rational conformal field theory, the chiral modules $\{a\}$ have fusion rules
\[
a\times b=\sum_c N^c_{ab}\,c,
\]
and the associated Verlinde lines $L_a$ satisfy
\[
L_aL_b=\sum_c N^c_{ab}L_c .
\]
The maximal internal symmetry is the fusion category $F$ generated by these lines, and the symmetric subalgebra is the identity sector
\[
A_F(O)=\{X\in A(O)\mid [L_a,X]=0\ \forall a\} .
\]
A line $L_a$ is invertible if $L_aL_{a^*}=1$; otherwise it is non-invertible [2503.20863].

In this setting the paper “Additivity, Haag duality, and non-invertible symmetries” establishes a sharp distinction. Additivity fails when $F$ contains a nontrivial invertible line, because one can build a bilocal pair of charged primaries in two disjoint intervals whose product commutes with $F$ although neither factor does [2503.20863]. By contrast, Haag duality fails whenever there exists a non-invertible line $L_c$ [2503.20863]. The construction uses twisted-sector operators $\phi_{c,1}(z)$ and $\phi_{c^*,1}(w)$ placed at the endpoints of $L_c$. They commute with all local primaries of the identity sector and hence with $A_F(O)$, but if $z,w$ lie in two components of $O$ then their composite is not contained in the restricted algebra of the complement. Concretely,
\[
D_O\coloneqq \phi_{c,1}(z)\;L_c\;\phi_{c^*,1}(w)
\]
obeys
\[
[D_O,A_F(O)]=0\qquad \text{but}\qquad D_O\notin A_F(O') .
\]
Thus for every non-invertible object $c$ in $F$ one finds an HDV [2503.20863].

The same work gives lattice analogues. On the Ising spin chain with periodic boundary conditions, Pauli operators $X_\ell,Z_\ell$, global $\mathbb{Z}_2$ spin-flip $U=\prod_\ell X_\ell$, and Kramers–Wannier MPO operator $\mathfrak{KW}$ acting on the $\mathbb{Z}_2$-even sector, one defines
\[
A_{KW}(R)=\{X\in A(R)\mid [X,U]=[X,\mathfrak{KW}]=[X,T]=0\},
\]
where $T$ is one-site translation [2503.20863]. Because $\mathfrak{KW}$ intertwines $U$ and $T$, any nontrivial operator commuting with the full algebra must have support on all sites. For any proper subregion $R$, one therefore finds $A_{KW}(R)=\mathbb{C}\cdot 1$, so both additivity and Haag duality fail in the restricted sector [2503.20863].

A second spin-chain example uses the non-invertible symmetry $\mathrm{Rep}(D_8)$. With even- and odd-site spin flips $U^e,U^o$ and an MPO operator $\mathfrak{D}$ satisfying $\mathfrak{D}^2=(1+U^e)(1+U^o)$, the symmetric subalgebra is
\[
A_{\mathrm{Rep}(D_8)}(R)=\{X\in A(R)\mid [X,U^e]=[X,U^o]=[X,\mathfrak{D}]=0\}.
\]
Direct Pauli-matrix algebra shows that bilocal disorder operators
\[
U(\ell_1,\ell_2)=\prod_{\ell=\ell_1}^{\ell_2}X_\ell
\]
can lie in the commutant of $A_{\mathrm{Rep}(D_8)}(R)$ without lying in the algebra of the complement when the endpoints belong to different components of a disconnected region. Hence Haag duality fails in the $\mathrm{Rep}(D_8)$-symmetric sector [2503.20863].

These 1+1-dimensional examples make explicit that HDV is not confined to gauge-theory loops or higher-form charge. In restricted identity sectors of non-invertible symmetries, the nonlocal operator can be a defect-line endpoint construction or a bilocal disorder operator, and the obstruction arises from the incompatibility between local symmetry invariance and the full commutant.

## 4. Gauge theories, Wilson lines, and confinement

Gauge theories provide another major class of HDV, now tied to Wilson line operators, Gauss law constraints, and confinement. In the AQFT treatment of the massless Schwinger model, local observable algebras are generated by gauge-invariant smeared currents and electric fields,
\[
j(f)=\int j^\mu(x)f_\mu(x)\,d^2x,\qquad E(h)=\int E(x)h(x)\,dx,
\]
subject to Gauss’s law
\[
\partial_1E(x)=e\,j^0(x)
\]
and the Schwinger anomaly [2507.14699]. One then extends the local net by Wilson line operators
\[
W(\gamma)=\exp\Bigl(i\,e\int_\gamma A_\mu\,dz^\mu\Bigr),
\]
defining an extended algebra
\[
A_{\mathrm{ext}}(O)=C^*\bigl(A(O)\cup \{W(\gamma):\gamma\cap O\neq \varnothing\}\bigr) .
\]
When a path $\gamma$ lies outside $O$, the Wilson line commutes with all local operators in $A(O)$, yet if the path crosses the complement in a nontrivial way it is not contained in $A(O')$. Hence
\[
A(O')\subsetneq A(O)' ,
\]
so Haag duality is violated in the extended net [2507.14699].

The physical interpretation is that charged fields fail to be localizable in bounded regions and the only DHR sector is the vacuum sector, which is presented as an exact algebraic statement of confinement [2507.14699]. In net-cohomological language the Wilson line defines a nontrivial $1$-cocycle whose class in $H^1_{\mathrm{net}}(A,U(1))\neq 0$ obstructs duality [2507.14699].

A non-Abelian analogue appears in the 1+1-dimensional $SU(N)$ gauge theory studied in “Confinement, Nonlocal Observables, and Haag Duality Violation in the Algebraic Structure of 1+1-Dimensional Non-Abelian Gauge Theories” [2508.09172]. There the local observable net is generated by gauge-invariant composite operators, including the fermion current and traces of non-Abelian electric fields, with the non-Abelian Gauss law
\[
D_1^{ab}E^b(x)=g\,j_0^a(x)
\]
imposed as an operator constraint [2508.09172]. To capture global gauge flux one introduces path-ordered Wilson lines
\[
W(x,y)=\mathcal{P}\exp\Bigl(i\,g\int_x^y A_\mu^a(z)T^a\,dz^\mu\Bigr) .
\]
Their commutator with the electric field is
\[
[E^a(x),W(y,z)]
=\delta(x-y)\,T^aW(y,z)-\delta(x-z)\,W(y,z)T^a ,
\]
showing that the line shifts color flux at its endpoints but otherwise commutes with local gauge-invariant observables supported away from those endpoints [2508.09172].

For a bounded interval $\mathcal{O}$ and two points $y,z\in \mathcal{O}'$ in the two spacelike wedges of the complement, the Wilson line $W(y,z)$ commutes with every element of $\mathcal{A}(\mathcal{O})$ but cannot be approximated by elements of $\mathcal{A}(\mathcal{O}')$, because operators in the complement factorize across the two disconnected components whereas the Wilson line stretches between them as a single operator [2508.09172]. This yields a strict inclusion
\[
\pi_0(\mathcal{A}(\mathcal{O}))' \supset \pi_0(\mathcal{A}(\mathcal{O}'))
\]
and therefore an HDV [2508.09172]. The paper identifies the topological origin through Roberts’ net cohomology and relates the obstruction to the non-trivial center $\mathbb{Z}_N$ of $SU(N)$ [2508.09172].

Across both Abelian and non-Abelian 1+1-dimensional gauge theories, HDV is therefore tied to the same structural triad: Gauss law, absence of localizable charged sectors, and nonlocal Wilson operators carrying flux between disconnected components of the complement [2507.14699], [2508.09172].

## 5. Higher-form symmetries, DHR reconstruction, and classification results

HDV also enters a broader classification program for generalized symmetries in dimensions $D>2$. The paper “A generalization of the DHR theorem for higher form symmetries” reformulates the DHR theorem in terms of HDV for regions with non-trivial homotopy [2511.21810]. For disconnected regions with $\pi_0(O)\neq 1$, finite-index subfactor theory yields a canonical endomorphism
\[
\rho_O \simeq \bigoplus_{r\in \mathrm{Irreps}(G)} d_r\cdot \rho_r ,
\]
where the $\rho_r$ are irreducible transportable endomorphisms localized in one component, with $\sum_r d_r^2=[\hat{\mathcal{A}}(O):\mathcal{A}(O)] =:\mu$ [2511.21810]. One then reconstructs an extension $\mathcal{B}\supset \mathcal{A}$ with a compact group $G$ of global symmetries such that
\[
\mathcal{A}(O)=\mathcal{B}(O)^G
\]
and the HDV for disconnected regions become trivial in the parent theory [2511.21810].

In this formulation, the category associated with $\pi_0$-type HDV is equivalent to the dual of a compact group. This provides a local-algebraic route to the group-like origin of $0$-form internal symmetries in $D>2$ [2511.21810]. The same framework extends to spontaneous symmetry breaking by passing first to the dual net $\mathcal{A}^d(O)=\hat{\mathcal{A}}(O)$, reconstructing a larger net $\mathcal{B}$, and identifying a chain
\[
\mathcal{A}\subset \mathcal{A}^d \subset \mathcal{B}
\]
with fixed-point relations $\mathcal{B}^H=\mathcal{A}^d$ and $\mathcal{B}^G=\mathcal{A}$ [2511.21810].

After eliminating $\pi_0$ sectors by reconstruction, the same paper treats HDV for regions with non-trivial higher homotopy, such as loops $S^k$ with $0<k<D-2$. By dimensional reduction to a lower-dimensional problem, it obtains a finite-index classification theorem: for each such $k$, the subfactor associated with a region homotopic to $S^k$ has index $|G|$ with $G$ a finite abelian group [2511.21810]. Concretely, for a loop $L\simeq S^k$ one may write
\[
\hat{\mathcal{A}}(L)=\mathcal{A}(L)\rtimes^\wedge G,
\]
with unitary flux operators $a_g$ satisfying
\[
a_{g_1}a_{g_2}=a_{g_1g_2}.
\]
In the special self-dual case $D=4n$ and $k=2n-1$, the character matrix
\[
\chi(g,h)=a_g a_h /(a_h a_g)
\]
must satisfy the Hermitian condition
\[
\chi(g,h)=\chi(h,g)^* .
\]
The paper further states that in this case $G\simeq H\times H^*$ up to a possible extra $\mathbb{Z}_2$ factor [2511.21810].

The examples include Yang–Mills Wilson and ’t Hooft loops. In an $SU(N)$ gauge theory without matter, the center $\mathbb{Z}_N$ produces HDV for ring regions with
\[
\hat{\mathcal{A}}(\mathrm{Loop})=\mathcal{A}(\mathrm{Loop})\rtimes^\wedge (\mathbb{Z}_N\times \mathbb{Z}_N^*) ,
\]
and the mutual commutation of Wilson and ’t Hooft loops is controlled by the linking number [2511.21810]. The same analysis extends to knot and link order parameters, where generalized knot non-local operators are classified by the unknot order parameters and their commutator is determined by the linking number [2511.21810].

These results place HDV at the center of a structural classification theorem: in $D>2$, $\pi_0$-type HDV reconstruct compact groups, while higher-form HDV are associated with finite abelian groups, with an additional Hermitian-character constraint in the self-dual case [2511.21810].

## 6. Physical significance, common confusions, and scope

A recurring interpretation across the recent literature is that Haag duality is a hallmark of completeness of local operator algebras, and an HDV indicates the presence of operators that are topological or nonlocal relative to a chosen region but are not generated by the local algebra of the complement [2503.20863]. In theories with non-invertible symmetries, HDV is described as generic whenever non-invertible topological sectors exist, because the very existence of these sectors forces the local algebra of the complement to be strictly smaller than its commutant [2503.20863]. In gauge theory, Wilson lines and related flux operators play the same role [2507.14699], [2508.09172].

One common confusion is to treat every violation of ordinary additivity as evidence of nonlocality. The 2025 analysis of disjoint additivity argues against this identification. Free Maxwell theory and related systems with higher-form symmetries are presented as perfectly local from the path-integral and lattice perspectives even though they violate ordinary additivity; the proposed resolution is to keep Haag duality and replace additivity by disjoint additivity [2509.03589]. This suggests that HDV and additivity failure must be interpreted together rather than in isolation.

A second confusion concerns whether HDV is always a defect of an incomplete description. Some examples do fit that pattern: hypersurface restrictions and generalized free fields obey microcausality but violate Haag duality outright [2509.03589]. But in other settings HDV is not merely a truncation artifact; it is the algebraic manifestation of a genuine global or topological structure, such as a non-invertible defect sector, a Wilson line, or a higher-form order parameter [2503.20863], [2511.21810]. A plausible implication is that the significance of HDV depends on whether the relevant nonlocal operator should be incorporated into the physical net or instead eliminated by a reconstruction to a larger parent theory.

The literature also connects HDV to broader structural themes. The 1+1-dimensional RCFT and spin-chain analysis states that HDV provides an operator-algebraic diagnostic of non-invertibility and connects to modular invariance in RCFTs and to completeness in gauge theories [2503.20863]. The higher-form reconstruction program uses HDV to give a new proof of the group-like origin of internal symmetries and a classification of generalized order parameters [2511.21810]. The Schwinger-model treatment conjectures that Haag duality is related to the quantum error-correcting property behind entanglement wedge reconstruction in holography, so its failure in a confining gauge theory may signal an obstruction to local quantum information recovery [2507.14699].

Taken together, these developments establish HDV as a unifying notion linking algebraic completeness, generalized symmetry, confinement, and topology. In 1+1-dimensional conformal and lattice models, HDV detects non-invertible sectors in symmetry-restricted algebras [2503.20863]. In gauge theories, it identifies Wilson and flux operators that commute with local observables yet evade localization in the complement [2507.14699], [2508.09172]. In $D>2$, it supplies the local-algebraic input for reconstructing compact groups and classifying higher-form order/disorder operators by finite abelian groups [2511.21810]. The resulting picture is that HDV is neither an incidental anomaly nor a single phenomenon with a uniform interpretation; it is a precise structural indicator of how global topological data fail to be exhausted by naive local operator algebras.

Source: https://www.emergentmind.com/topics/haag-duality-violations-hdv