Subregion–Subregion Duality in Holography

• Subregion–subregion duality is a framework defining bijective correspondences between local subregions across dual spaces, notably in holography via AdS/CFT.
• It employs causal and entanglement wedge reconstructions alongside an algebraic formulation to map boundary operators to bulk regions and facilitate quantum error correction.
• Extensions into tropical geometry, locale theory, and kinematic space highlight its versatility, with consistency ensured by quantum extremality and wedge-nesting principles.

Subregion–subregion duality encompasses a constellation of ideas connecting local structures (“subregions”) on two sides of a duality: typically, operator or algebraic structures associated to regions of one space are mapped to structures associated to corresponding regions of a dual space. Modern prominence arises from AdS/CFT, where dualities between certain subregions of the conformal field theory and bulk anti-de Sitter space encode deep geometric, information-theoretic, and algebraic correspondences. Subregion–subregion duality generalizes to purely algebraic, convex-geometric, or topological settings, for instance in tropical geometry and locale theory. Across frameworks, the unifying concept is the existence of a compatible, often structure-preserving, bijection or duality between families of subregions or subalgebras on both sides.

1. Subregion–Subregion Duality in AdS/CFT

Subregion–subregion duality in the AdS/CFT correspondence formalizes the principle that to each region $A$ of the CFT, one can associate a corresponding bulk region (usually causal or entanglement wedges), and that the reduced density matrix $\rho_A$ encodes, in a precise sense, the physics of the corresponding bulk subregion (Caceres et al., 2019, Akers et al., 2016).

Key definitions are:

• Causal wedge $C[A]$: For a spatial region $A$ in the boundary CFT with domain of dependence $D(A)$, the causal wedge is $C[A] = I^-(D(A)) \cap I^+(D(A))$, comprising bulk points both influencing and influenced by $D(A)$.
• Entanglement wedge $E[A]$: Defined as the bulk domain spacelike-related to the quantum extremal surface $e(A)$, which extremizes the generalized entropy $$S_{\rm gen}(\sigma) = \operatorname{Area}(\sigma)/4G\hbar + S_{\text{out}}(\sigma) + Q(\sigma)$$.

The central conjecture is that all low-energy operators localized in $E[A]$ (or $C[A]$ for the causal wedge) can be reconstructed from $A$ within a suitable code subspace, and that this map is “dual” in the sense that boundary algebra and bulk algebra agree on $E[A]$ to all relevant orders (Leutheusser et al., 2022, Soni, 2024).

Inclusion relations are tightly constrained: $C[A] \subseteq E[A]$ always, and $E[A]$ satisfies monotonicity under nesting of boundary domains. These relations follow from the Quantum Focusing Conjecture (QFC), Generalized Second Law (GSL), and energy-entropy inequalities (such as the Quantum Null Energy Condition, QNEC) (Akers et al., 2016).

2. Algebraic and Operator-Theoretic Formulation

A precise operator-algebraic formalism is provided by “subregion–subalgebra duality,” which underpins and sharpens geometric forms of subregion–subregion duality (Leutheusser et al., 2022). Given a semiclassical boundary state $\Psi$ with von Neumann algebra $\mathfrak{A}_\Psi$, a boundary subalgebra $\mathfrak{A}$ is dual to a causally complete bulk region $\mathcal{R}$ if there exists an isomorphism

$$\pi_\Psi\left(\mathfrak{A}\right) = \widetilde{\mathfrak{A}(\mathcal{R})}$$

where $\widetilde{\mathfrak{A}(\mathcal{R})}$ is the bulk degree-of-freedom algebra. For boundary algebraic subregions $\mathfrak{A}(\hat R)$ defined on $D(R)$, one recovers standard entanglement wedge duality: $\mathfrak{A}_R = \widetilde{\mathfrak{A}(EW(R))}$.

This algebraic structure admits superadditive and nontrivial intersection/union properties: $$\mathfrak{A}_{R_1}\vee\mathfrak{A}_{R_2} \subsetneq \mathfrak{A}_{R_1\cup R_2}, \quad \mathfrak{A}_{R_1\cap R_2} \subsetneq \mathfrak{A}_{R_1}\wedge\mathfrak{A}_{R_2}$$ with geometric analogs for the entanglement wedges.

Quantum error correction naturally arises in this setting: there exists a recovery channel $\mathcal{R}_{R}$ such that any local bulk operator $O_{\text{bulk}}$ in $EW(R)$ can be reconstructed on the boundary subalgebra $\mathfrak{A}_R$, and erasures on $\bar R$ are correctable.

3. Wedge Hierarchies and Duality Variants

Several wedge constructions make the classification of subregion dualities sharp (Bao et al., 2024). Central notions include:

• Background wedge $B(A)$: The maximal region reconstructible “geometrically” (i.e. for the metric), typically strictly larger than any “operator wedge.”
• Operator-reconstruction wedges (in ascending order, for region $A$):
  1. $W_o^1[A] = D(A)$ (causal wedge)
  2. $W_o^2[A] = G(A)$ (min-entanglement wedge; smallest for which no outside operator is reconstructible)
  3. $W_o^3[A] = E(A)$ (conventional entanglement wedge)
  4. $W_o^4[A] = R(A)$ (max-entanglement wedge; largest region where all operators can be reconstructed)

Nesting always holds, $$D(A) \subseteq G(A) \subseteq E(A) \subseteq R(A) \subseteq B(A)$$, and each wedge underlies a different duality protocol: HKLL (causal), modular-flow/Petz map (entanglement), or one-shot quantum information (min/max EW).

Subregion–subregion duality thus encapsulates a family of dualities labeled by reconstruction method, state, and code subspace, not a unique bijective map.

4. Technical Constraints and Consistency Conditions

The viability of subregion–subregion duality at finite $N$ and near horizons requires the entanglement wedge to be bounded by quantum extremal surfaces (QES), rather than classical extremal surfaces (Soni, 2024, Akers et al., 2016). Any candidate wedge failing the QES condition can be “spoiled” via a complementary modular flow that induces boundary-causality violations.

The logical chain of constraints is:

• Causal-wedge inclusion: $W_E[A] \supseteq W_C[A]$ is required for consistent representation of simple causal bulk unitaries as boundary operators.

• Quantum extremality: Variational stationarity of generalized area, enforced by QHANEC and monotonicity of relative entropy, is a necessity, not merely a sufficiency, for entanglement wedge consistency.

Moreover, in higher-derivative or nonminimal gravity theories, causal and entanglement wedges must be defined using the fastest-propagating bulk mode (e.g., graviton tensor mode) rather than the background metric (Caceres et al., 2019). The demand for causal wedge inclusion yields constraints on higher-derivative couplings as strong as S-matrix causality: in Gauss–Bonnet gravity, it imposes $\lambda = O(\ell_P^2/L^2)$, matching bounds from high-energy graviton scattering.

5. Subregion Complementarity, Observer Dependence, and Black Hole Contexts

Exact one-to-one correspondence between boundary and bulk subregion algebras is violated at $O(N^0)$, even in leading semiclassical limits (Sugishita et al., 2023, Sugishita et al., 2022). In particular, naive AdS–Rindler wedge reconstructions include modes (e.g., trans-Planckian or tachyonic) inconsistent with global CFT constraints, causing breakdown of subregion duality in the strict sense.

Instead, subregion complementarity holds: distinct boundary algebras (e.g., global HKLL and AdS–Rindler HKLL images) may act inequivalently on the global Hilbert space, but agree on all correlators in the shared bulk region, except in neighborhoods near the horizon or in single-sided black holes. In eternal black holes, complementarity recovers smoothness behind the horizon by incorporating a second CFT; for single-sided holes, complementarity fails beyond the stretched horizon, precluding alternate algebraic reconstructions.

This perspective underlines that subregion–subregion duality in holography is inherently observer/patch-dependent, and in general, the global mapping is not unique, but rather comprises a family of patchwise dualities consistent on overlaps.

6. Extensions: Topology, Convexity, and Abstract Dualities

The notion of subregion–subregion duality generalizes to other fields:

• Tropical geometry: The row- and column-spaces of an $n\times n$ tropical matrix $A$, viewed as tropically convex sets, are bijectively related by a duality map $\delta_A$, which is necessary and sufficient for their realization as the row and column space of $A$ (Hollings et al., 2010). This duality map also induces an isometry in the Hilbert projective metric between projective row and column spaces.
• Locale theory and $T_D$–duality: In point-free topology, D-sublocales $S$ of a frame $L$ correspond bijectively to subsets of covered primes $\operatorname{pt}_D(L)$, i.e., $S \leftrightarrow \operatorname{pt}_D(S)$ with the “meet-closure” functor providing the inverse map. When $L$ is $T_D$–spatial, the coframe of D-sublocales is Boolean and atomic, isomorphic to the powerset of the covered primes (Arrieta et al., 2020). This realizes a fully functorial subregion–subregion duality between point-free and point-set regions.

These algebraic and combinatorial dualities mirror, in an abstract setting, the geometric and operator-algebraic dualities of AdS/CFT.

7. Kinematic Space and Information-Theoretic Reconstructions

Kinematic space constructions illustrate how subregion–subregion duality enables bulk reconstruction from knowledge of a boundary region (Huang, 2020). In AdS$_3$/CFT$_2$, geodesics crossing the entanglement wedge boundary can be mapped bijectively to “reflected” geodesics in a canonical purification, so that all bulk geometric information inside the wedge, including reflected entropy, is reconstructible from the reduced density matrix $\rho_A$. The Crofton formula and Radon transform invert the kinematic data into local bulk operators, ensuring the reconstructability of subregions and enforcing subregion–subregion duality at the level of all geometric probes determined by $\rho_A$.

---

In conclusion, subregion–subregion duality admits a technically rigorous, multi-faceted formulation with deep implications for holography, operator algebra, convex geometry, and abstract topology. It is governed by geometrically and algebraically precise criteria (e.g., quantum extremality, wedge inclusion, convex-duality maps, Boolean coframes), and contemporary research demonstrates both its scope and its limitations: subregion–subregion duality is generically patch/observer-dependent, globally non-unique, and sharply constrained by consistency with causal, quantum, algebraic, and geometric structures (Sugishita et al., 2023, Caceres et al., 2019, Leutheusser et al., 2022, Akers et al., 2016, Soni, 2024, Huang, 2020, Sugishita et al., 2022, Hollings et al., 2010, Arrieta et al., 2020, Bao et al., 2024).