---
title: Boundary DHR Category Overview
url: https://www.emergentmind.com/topics/boundary-dhr-category
type: topic
---

# Boundary DHR Category Overview

Boundary DHR category denotes an operator-algebraic superselection structure attached to a boundary observable algebra, or more generally to a lower-dimensional net extracted from a bulk system. In the recent lattice literature, the central idea is that a physical boundary can be encoded by a quasi-local algebra or local net in one lower dimension, and that localized, transportable Hilbert bimodules over this algebra form the relevant category of boundary sectors. In abstract spin chains, this category is a braided \(C^*\)-tensor category of DHR bimodules over the quasi-local boundary algebra; in local topological order, it is the DHR category of the canonically constructed boundary net; in half-infinite symmetric chains, it is a monoidal category of boundary-localized bimodules; and in rational boundary conformal field theory, the analogous boundary sector structure is often expressed instead by module categories over the chiral DHR category rather than by a new modular tensor category [2504.06094] [2307.12552] [2606.19137] [1410.8848].

## 1. Conceptual and operator-algebraic basis

The background notion is the Doplicher–Haag–Roberts superselection formalism, in which sectors are localized and transportable representations or endomorphisms. For lattice systems and quasi-local algebras, several recent works replace localized endomorphisms by localized Hilbert bimodules or correspondences, producing a state-independent version of DHR theory well adapted to boundary observable algebras [2304.00068] [2504.06094].

For a quasi-local algebra \(A\), the bimodule formulation starts from right-finite \(A\)-\(A\) correspondences. In the discrete-net framework, a correspondence \(X\) is localizable in a finite region \(F\) if it admits a projective basis \(\{b_i\}\) such that
\[
ab_i=b_i a \qquad \forall\, a\in A_{F^c}.
\]
An object is then called localizable if it is localizable in all sufficiently large balls, and the resulting full \(C^*\)-tensor subcategory is denoted \(DHR(A)\) [2304.00068]. In the abstract spin-chain formulation, localization is encoded vectorwise: a vector \(\xi\in X\) is localized in \(F\subset \mathbb Z\) if
\[
a\xi=\xi a \qquad \forall a\in \mathcal A_{F^c},
\]
and a DHR bimodule is a dualizable \(\mathcal A\)-\(\mathcal A\) correspondence having projective bases localized in arbitrarily chosen sufficiently large intervals [2504.06094].

This suggests a general boundary pattern: one first identifies the correct boundary quasi-local algebra or boundary net, and then defines boundary sectors as localized transportable bimodules over that algebra. A plausible implication is that the phrase “boundary DHR category” is best understood as a family of closely related constructions adapted to different boundary geometries and different operator-algebraic frameworks.

## 2. Boundary nets from local topological order

A systematic construction appears for locally topologically ordered quantum spin systems on \(\mathbb Z^\ell\). Starting from a net \(\Lambda\mapsto \mathcal A(\Lambda)\subset \mathcal A\) and local ground-state projections \(p_\Lambda\), one defines for suitable bulk regions \(\Lambda_I\Subset_s\Delta_I\) the local boundary algebra
\[
\mathfrak B(I):=\mathfrak B(\Lambda_I \Subset_s \Delta_I),
\]
and then forms the boundary quasi-local algebra
\[
\mathfrak B:=\varinjlim_I \mathfrak B(I).
\]
The resulting net on \(\mathbb Z^{\ell-1}\) is local, and there is a canonical ucp map
\[
\mathbb E:\mathcal A_{\mathbb H}\to \mathfrak B
\]
satisfying
\[
p_\Delta x p_\Delta = \mathbb E(x)\,p_\Delta
\]
for appropriate \(\Lambda\Subset\Delta\) [2307.12552].

Within this framework, the boundary DHR category is the category \(DHR(\mathfrak B)\) of localizable right-finite correspondences over the boundary algebra. The paper argues that in \((2+1)\)D this braided tensor category characterizes the bulk topological order, and proves this claim in the toric code and Levin–Wen examples [2307.12552].

The construction was then extended to systems that already possess a microscopic or topological boundary. For a lattice with boundary
\[
\mathcal L=\mathbb Z^d\times \mathbb N,
\]
one imposes boundary versions of the LTO axioms, written \((\partial\mathrm{LTO}1)\)–\((\partial\mathrm{LTO}4)\), and extracts a physical boundary net with boundary
\[
I\longmapsto \mathfrak B^\partial(I).
\]
Boundary DHR bimodules are then finitely generated projective \(\mathcal A\)-\(\mathcal A\) correspondences localized in every sufficiently large boundary-touching region, and the full subcategory is denoted
\[
DHR^\partial(\mathcal A)\subset Bim(\mathcal A).
\]
In this setting the general result is a \(C^*\)-tensor category, while a braiding is established only in specific models such as the Walker–Wang enriched boundary net [2506.19969].

## 3. Abstract spin chains and the braided boundary DHR category

The most explicit operator-algebraic use of the term occurs for abstract spin chains, which axiomatize the structure of one-dimensional boundary observables of \(2+1\)D topological order. An abstract spin chain is a functor
\[
\mathcal A_\bullet:\mathcal I\to C^*\mathrm{-alg}_{fd}
\]
on finite intervals of \(\mathbb Z\), with commuting local algebras on disjoint intervals and quasi-local algebra
\[
\mathcal A=\varinjlim_I \mathcal A_I.
\]
The boundary DHR category is the full \(C^*\)-tensor subcategory
\[
\mathrm{DHR}(\mathcal A_\bullet)
\]
of dualizable correspondences over \(\mathcal A\) satisfying the basis-localization condition described above [2504.06094].

Its braiding is defined under weak algebraic Haag duality. If \(X,Y\in \mathrm{DHR}(\mathcal A_\bullet)\) have projective bases \((\xi_i)\), \((\eta_j)\) localized in sufficiently separated intervals \(I\ll J\), then
\[
\beta_{X,Y}(\xi_i\boxtimes_{\mathcal A}\eta_j)=\eta_j\boxtimes_{\mathcal A}\xi_i.
\]
The exchange depends on ordered localization along the line rather than planar isotopy. Charge transporters control the passage between different localized bases, and the monodromy is expressed entirely in terms of transporter coefficients [2504.06094].

The structural results are strong. Under rationality, charge-transporter generation, and local alignment, \(\mathrm{DHR}(\mathcal A_\bullet)\) is a unitary modular tensor category. Under algebraic Haag duality, it is braided equivalent to the Drinfeld center of the half-line fusion category:
\[
\mathrm{DHR}(\mathcal A_\bullet)\simeq Z(\mathcal C_-).
\]
Here \(\mathcal C_-\) is obtained by restricting DHR bimodules to the negative half-line. This realizes a precise bulk–boundary relation: the boundary superselection sectors reconstruct the bulk topological order as a center [2504.06094].

A closely related precursor is the state-independent theory of DHR bimodules for quasi-local algebras. In one-dimensional fusion spin chains with fusion categorical symmetry \(\mathcal D\), that theory proves
\[
DHR(A)\cong \mathcal Z(\mathcal D),
\]
again identifying the DHR category of a one-dimensional quasi-local algebra with a Drinfeld center [2304.00068].

## 4. Model realizations and recovered categories

The boundary DHR category is especially explicit in solvable lattice models. In the local-topological-order framework, the boundary nets of Levin–Wen models and the toric code are identified with fusion categorical nets on \(\mathbb Z\), and their DHR categories reproduce the expected bulk modular categories [2307.12552]. In the bulk-boundary extension, Levin–Wen boundaries recover boundary fusion data, while Walker–Wang boundaries recover a braided enriched center [2506.19969].

| Model or framework | Boundary algebra/net | Recovered category |
|---|---|---|
| Levin–Wen boundary net | \(\mathfrak F(I)=\mathrm{End}_{\mathcal C}(X^{\# I})\) | \(DHR(\mathfrak F)\cong Z(\mathcal C)\) |
| Toric code boundary net | Fusion categorical net for \(\mathrm{Hilb}_{fd}(\mathbb Z/2)\) | \(DHR(\mathfrak B)\cong Z(\mathrm{Hilb}_{fd}(\mathbb Z/2))\) |
| Levin–Wen bulk-boundary system | \(\mathfrak M\) | \(DHR^\partial(\mathfrak M)\cong \mathrm{End}(\mathcal M_{\mathcal C})\) |
| Walker–Wang bulk-boundary system | \(\mathfrak X\) | \(DHR^\partial(\mathfrak X)\simeq Z^\mathcal B(\mathcal X)\) |

For Levin–Wen, the boundary algebra is identified with a fusion module spin chain
\[
\mathfrak M(I)=
\begin{cases}
\mathrm{End}_{\mathcal C}(X^{\otimes I}) & I\cap\partial\mathcal L=\emptyset,\\[1mm]
\mathrm{End}_{\mathcal M}(W\lhd X^{\otimes I\setminus \partial\mathcal L}) & I\cap\partial\mathcal L\neq\emptyset,
\end{cases}
\]
and the boundary DHR category is tensor equivalent to \(\mathrm{End}(\mathcal M_{\mathcal C})\), the expected category of boundary excitations [2506.19969].

For Walker–Wang, the physical boundary algebra is a braided categorical net with boundary, and the paper proves a unitary braided equivalence
\[
DHR^\partial(\mathfrak X)\simeq Z^\mathcal B(\mathcal X)=\mathcal B'\subset Z(\mathcal X).
\]
The same work constructs a \(2\)D braided categorical net from a UBFC and shows that, in the canonical state associated with the standard topological boundary, its cone von Neumann algebras are type I with finite-dimensional centers, in contrast with the type II and III cone algebras found earlier for Levin–Wen models [2506.19969].

## 5. Half-infinite chains, boundary symmetry TFT, and boundary conditions

For \((1+1)\)D symmetric gapped phases, the boundary DHR category is formulated for half-infinite fusion spin chains. An abstract spin chain with boundary is a functor
\[
Loc_\bullet:\mathcal I^+\to C^*\text{-alg}
\]
on finite intervals of \(\mathbb Z_{\ge 0}\), with quasi-local algebra
\[
Loc=\varinjlim_{\mathcal I^+} Loc_I.
\]
A boundary DHR bimodule is a semisimple Hilbert \(Loc^{bdy}\)-bimodule admitting a projective basis localized in some boundary interval \([0,r]\), and the full subcategory is
\[
DHR(Loc^{bdy}_\bullet)\subset \mathrm{Bim}(Loc^{bdy}) .
\]
This category is monoidal under relative tensor product, but its defining localization is one-sided, reflecting the fixed physical boundary at the endpoint [2606.19137].

The main identification is
\[
DHR(Loc^{bdy}_\bullet)\simeq (C_M^\vee)^{rev},
\qquad
C_M^\vee:=\mathrm{Fun}_{C^{rev}}(M,M),
\]
for a half-infinite fusion spin chain built from a unitary fusion category \(C\) and an indecomposable semisimple right \(C\)-module category \(M\). The bulk DHR category is
\[
DHR(Loc^{bulk}_\bullet)\simeq Z_1(C^{rev}),
\]
and the canonical action of the bulk DHR category on the boundary DHR category agrees with the categorical action of \(Z_1(C^{rev})\) on \((C_M^\vee)^{rev}\) [2606.19137].

The same framework classifies boundary conditions by right \(Q\)-modules. If \(Q\in C\) is a simple Q-system specifying the bulk phase, then the realization functor
\[
\ReaBCond:M_Q^{op}\to \mathrm{BCond}
\]
is an equivalence, so simple boundary conditions are classified by simple objects of \(M_Q\) and general boundary conditions by finite direct sums [2606.19137]. This makes the boundary DHR category the operator-algebraic realization of boundary SymTFT, while the action on boundary conditions provides the categorical boundary module structure.

## 6. Relation to bulk sectors, boundary CFT, and scope of the notion

The boundary DHR category is not a single invariant with identical formal properties in every setting. A recurrent distinction is between genuine tensor categories of boundary-localized sectors and module-category descriptions of boundaries. In rational \(2\)D conformal nets, the categorical structure associated with a boundary is not introduced as a standalone “boundary DHR category”; instead, for a completely rational chiral net \(\mathcal A\), maximal \(2\)D bulk extensions are classified by Morita equivalence classes of Q-systems in
\[
\mathrm{DHR}(\mathcal A),
\]
and boundary conditions are encoded by Q-systems in that Morita class or, equivalently, by simple objects of the module category \(\mathrm{Mod}(\Theta)\) modulo invertible dual-category symmetries [1410.8848]. This suggests that, in conformal boundary theory, the natural boundary categorical object is often a module category over the chiral DHR UMTC rather than a new modular category.

Bulk operator-algebraic constructions provide the comparison point. For the non-abelian quantum double model, the finite cone-localized DHR category is braided \(C^*\)-tensor equivalent to
\[
\mathrm{Rep}_f\,\mathcal D(G),
\]
showing how bulk anyon braiding arises from cone localization, transportability, and left/right separation in the plane [2503.15611]. A plausible implication is that boundary theories alter precisely this geometry: cones are replaced by boundary-adapted regions, and one may obtain a tensor category, a half-braiding, or a module action rather than the full planar braiding of the bulk.

A further caution comes from \(2+1\)-dimensional AQFT with stringlike Buchholz–Fredenhagen sectors. There the compactly localized DHR sectors form a transparent symmetric subcategory inside the braided BF category, so nontrivial DHR sectors obstruct modularity. Passing to the Doplicher–Roberts field net removes that DHR obstruction, but does not automatically guarantee full modularity [1004.4755]. This clarifies that “boundary DHR category” should not be assumed modular without additional hypotheses.

In current operator-algebraic usage, therefore, the term refers to a boundary-adapted superselection category constructed from a boundary algebra or boundary net, usually via localized bimodules. In \(2+1\)D topological lattice systems it can recover the bulk modular category or the boundary fusion/braided category; in half-infinite symmetric chains it yields the boundary SymTFT; and in rational boundary conformal field theory the comparable role is played by module categories and Q-systems over the chiral DHR category [2307.12552] [2506.19969] [2606.19137] [1410.8848].

Source: https://www.emergentmind.com/topics/boundary-dhr-category