Entanglement Wedge Cross-Section in Holography

Updated 27 February 2026

Entanglement Wedge Cross-Section is a geometric measure in holography that defines the minimal partitioning surface within the bulk, separating boundary subregions.
It underlies key holographic dualities by corresponding to entanglement of purification, reflected entropy, and odd entanglement entropy, thereby linking geometry to mixed-state correlations.
EWCS dynamically tracks quantum and classical correlations, exhibiting characteristic time-dependent growth and phase transition behavior in response to bulk deformations and quenches.

The entanglement wedge cross-section (EWCS) is a geometric quantity in the AdS/CFT correspondence that characterizes mixed-state correlations between boundary subregions in terms of minimal-area surfaces in the bulk spacetime. It underlies a range of holographic dualities equating bulk geometric data with boundary information-theoretic measures, and serves as a key probe of multipartite correlations, entanglement phase transitions, quantum information scaling, and the response of correlation measures to both local and global perturbations.

1. Definition and Holographic Construction

The EWCS is defined for two (or more) disjoint boundary subregions, typically labeled $A$ and $B$ , in a holographic CFT. The union $A\cup B$ is associated, via the Ryu–Takayanagi (RT) or HRT prescription, to a bulk codimension-2 extremal surface $\gamma_{A\cup B}$ , and the entanglement wedge $\mathcal{W}[A\cup B]$ is the bulk domain bounded by $A\cup B$ and $\gamma_{A\cup B}$ . The EWCS is the minimal-area codimension-2 surface $\Sigma_{A:B}^{\mathrm{min}}$ inside $\mathcal{W}[A\cup B]$ which splits the wedge into two regions homologous to $A$ and $B$ , respectively:

$E_W(A:B) = \min_{\Sigma\subset \mathcal{W}[A\cup B]} \frac{\mathrm{Area}(\Sigma)}{4 G_N}$

Here, $G_N$ is the bulk Newton constant. The EWCS can be formulated for more general multipartitions: for $n$ regions $A_1,\ldots,A_n$ with connected $\mathcal{W}[A_1\cup\cdots\cup A_n]$ , the EWCS is the minimal-area partitioning surface dividing the wedge into $n$ subwedges homologous to each $A_i$ (Bao et al., 2021).

2. Relation to Mixed-State Information Measures

EWCS serves as the holographic dual of several boundary correlation measures that go beyond von Neumann entropy. Key identifications, supported by replica and analytic constructions, are:

Entanglement of purification: $E_P(A:B) = E_W(A:B)$ for holographic states (Akers et al., 2019).
Reflected entropy: $S_R(A:B) = 2E_W(A:B)$ , where $S_R$ is the von Neumann entropy in a canonical purification (GNS state) of $\rho_{AB}$ (Dutta et al., 2019).
Odd entanglement entropy: For CFTs, $S_o(A:B) - S(A\cup B) = E_W(A:B)$ in holographic limits (Tamaoka, 2018, Mollabashi et al., 2020).
Logarithmic negativity: $E_W(A:B)$ is proportional (with a dimension-dependent factor) to the negativity between $A$ and $B$ in the large- $c$ limit (Sahraei et al., 2021).

For example, in $d=2$ the explicit expressions for two disjoint intervals of lengths $\ell$ and separation $h$ demonstrate these relations:

$E_W = \frac{c}{6} \log\left(\frac{1+\sqrt{x}}{1-\sqrt{x}}\right), \quad x = \frac{(u_1-v_1)(u_2-v_2)}{(u_1-u_2)(v_1-v_2)}$

where $c$ is the CFT central charge (Tamaoka, 2018).

3. Dynamical Evolution in Quenched and Time-Dependent Backgrounds

EWCS exhibits rich time-dependent behavior in holographic global and local quenches, governed by the geometry of infalling shells or local operator shock waves. The typical dynamical regimes observed are:

Early time: EWCS grows quadratically with time, $\Delta E_W(t)\propto t^2$ for relativistic cases, or $\propto (zt)^{1+1/z}$ for Lifshitz exponent $z$ (Velni et al., 2023).
Intermediate time: Linear growth regime, $\Delta E_W(t)\approx v_E t$ , with $v_E$ matching the entanglement velocity from HEE growth (Velni et al., 2020, Velni et al., 2023).
Late time: EWCS saturates when the entanglement wedge reaches its equilibrium configuration; this can be continuous or discontinuous (first-order–like transition), depending on the geometry and separation.
Quenches in higher-derivative/higher-spin gravity: Gauss–Bonnet corrections slow the onset of thermal correlation build-up and modify the disentanglement time and profile (Li et al., 2021).

The dynamical EWCS tracks not only quantum but also classical correlations generated during the evolution, exhibiting distinct behaviors from mutual information (MI) and standard HEE, for example in the presence of thermalization plateaus, saturation plateaus, and non-monotonic response to Gauss–Bonnet or other couplings.

4. Phase Structure, Quantum Information Scaling, and Universality

EWCS acts as a diagnostic for entanglement phase transitions. For two strips of width $\ell$ separated by $h$ , the connectedness of the entanglement wedge determines whether $E_W>0$ or $E_W=0$ . Discontinuities in $E_W$ at critical separation or system parameter values are a robust signature of purification transitions (Jokela et al., 2019, Velni et al., 2019):

First-order phase transitions: Discontinuous drop of $E_W$ at separations where MI vanishes; EWCS and MI may exhibit opposite monotonicity at criticality (Liu et al., 2021).
Second-order transitions: EWCS shows a continuous profile with a singular derivative; near criticality, all geometric measures (HEE, MI, EWCS) scale with the same critical exponent, e.g., $\alpha=1/3$ in van der Waals–type transitions, $\alpha=1$ when driven by scalar condensation (Liu et al., 2021).
High/low-temperature regimes: The leading correction to EWCS is area-law dominated, and EWCS vanishes via a first-order–like transition as separation or $T$ is increased (Velni et al., 2019).

EWCS can distinguish mechanisms of critical scaling: “metric-driven” vs. “scalar-condensate–driven” (Liu et al., 2021).

5. Multipartite EWCS, Inequalities, and the Entanglement Structure

Extensions to multipartite settings define the $n$ -partite EWCS as the minimal-area partitioning surface inside the bulk wedge associated to $A_1 \cup \cdots \cup A_n$ . Using replicated bulk geometries, all multipartite EWCS can be re-expressed as standard RT surfaces in the replicated spacetime, and inherit the full suite of entropy cone inequalities (Bao et al., 2021). Key results include:

Universal inequalities: For any (possibly disconnected) regions,
- $E_W(A:B) \geq \frac12 I(A:B)$ ,
- $E_W(A:B) \leq \min\{S(A), S(B)\}$ ,
- $E_W(A:BC) \geq E_W(A:B)$ ,
- Monogamy/polygamy properties are geometry- and dimension-dependent; only weaker, “mutual-information–improved” or quadratic (squared-monogamy) inequalities hold universally (Jain et al., 2022).
AdS/BCFT generalization: All main inequalities for EWCS in AdS/CFT also extend to AdS/BCFT geometries, with the added feature that extremal surfaces may end on branes. The set-based proof via wedge nesting is unaffected by this additional phase structure (2206.13417).
Tripartite entanglement: Holographic states must have $\mathcal{O}(1/G_N)$ tripartite entanglement, as the difference $2E_W - I(A:B) \sim \mathcal{O}(1/G_N)$ cannot be reproduced by (almost) bipartite tensor network states (Akers et al., 2019).
New multipartite bounds: Replicated geometry techniques yield a hierarchical set of new inequalities mixing different party numbers in the multipartite EWCS zoo (Bao et al., 2021).

6. Sensitivity to Gravity Modifications, Matter Fields, and Deformations

EWCS is highly sensitive to both ultraviolet and infrared properties of the bulk geometry and responds distinctly to different types of deformations:

Higher-derivative corrections: Gauss–Bonnet terms enhance EWCS and delay disentanglement, affecting both critical separation and time evolution post-quench (Li et al., 2021).
Massive gravity and axion backgrounds: EWCS effectively probes momentum dissipation, scaling behaviors, and phase transitions, with universal growth in e.g., axion–Maxwell coupling—behavior not shared with HEE or MI (Cheng et al., 2021, Liu et al., 2021).
Einstein–Aether gravity: Lorentz-violating couplings induce non-monotonic (“U-shaped”) dependence of EWCS on system parameters, a richer behavior than seen in HEE or MI (Chen et al., 2021).
Excited states and matter backreaction: Purely gravitational excitations lower EWCS, while gauge fields and scalar condensates can enhance or suppress it, depending on operator dimension (Sahraei et al., 2021).
Nonrelativistic and hyperscaling-violating geometries: Hamiltonian scaling exponents govern both the growth rate and the persistence of $E_W$ at late times, as observed in dynamical and equilibrium studies (Velni et al., 2019, Velni et al., 2023).

Such sensitivity makes EWCS a detailed probe of both quantum and classical correlations in holographic quantum matter, and a powerful diagnostic for distinguishing the physical content of different holographic phases and RG flows.

7. Quantum Information Interpretation and Physical Implications

EWCS is fundamentally a mixed-state correlation measure, connecting to physical entanglement features such as the spread of quantum/chaotic information, the distinction between classical and quantum correlations, and tripartite/beyond entanglement. In the presence of time-dependent driving or operator quenches, EWCS discriminates between chaos (growth of classical and quantum correlations) and integrability (step-function behavior, pure entanglement). At strong coupling and large $c$ , the universal relations between EWCS and reflected/odd entropy or entanglement of purification are apparent, while for free or weakly coupled models, classicality and monotonicity may break down (Mollabashi et al., 2020, Kusuki et al., 2019, Kusuki et al., 2019).

Multipartite generalizations of EWCS, triangle information, and entanglement of assistance further quantify the structure of assisted and shared entanglement in holography, align with quantum information theoretic objects such as conditional mutual information, and bring new phase structure and operational meaning to the distribution and interconversion of entanglement resources in AdS/CFT (Ju et al., 25 Dec 2025).

Key References: (Li et al., 2021, Liu et al., 2021, Bao et al., 2021, Jain et al., 2022, 2206.13417, Dutta et al., 2019, Akers et al., 2019, Tamaoka, 2018, Kusuki et al., 2019, Velni et al., 2019, Jokela et al., 2019, Velni et al., 2023, Sahraei et al., 2021, Cheng et al., 2021, Chen et al., 2021, Ju et al., 25 Dec 2025, Mollabashi et al., 2020, Velni et al., 2020, Kusuki et al., 2019).