Entanglement Wedge Cross Section

• EWCS is defined as the minimal-area codimension-2 surface within the entanglement wedge, extending the Ryu–Takayanagi prescription to mixed states.
• It is closely linked to measures like reflected entropy and entanglement negativity, capturing both quantum and classical correlations in holography.
• Computations of EWCS involve solving orthogonality conditions and adapting methods beyond Einstein gravity to probe phase transitions and critical dynamics.

Searching arXiv for recent and foundational EWCS papers to ground the article. Entanglement wedge cross section (EWCS), usually denoted $E_W(A:B)$ for boundary subregions $A$ and $B$, is the minimal-area codimension-2 surface inside the entanglement wedge of $A\cup B$ that separates the bulk region into parts homologous to $A$ and $B$. In static AdS/CFT it is a geometric generalization of the Ryu–Takayanagi surface from pure-state entanglement entropy to mixed-state correlations, while in covariant settings it is replaced by an extremal cross section inside the HRT wedge. Across the literature surveyed here, EWCS is treated as a mixed-state probe related to entanglement of purification, reflected entropy, logarithmic negativity, odd entanglement entropy, and balanced partial entanglement, but the same literature also emphasizes that its precise information-theoretic interpretation is subtler than a simple entanglement monotone (Tamaoka, 2018, Wen et al., 2022, Umemoto, 2019).

1. Definition, geometric meaning, and basic kinematics

For a bipartite mixed state $\rho_{AB}$ on boundary subregions $A$ and $B$, the standard holographic prescription used throughout these works is

$$E_W(\rho_{AB})=\min_{\Sigma_{AB}\subset \mathcal{E}_{AB}}\frac{\mathrm{Area}(\Sigma_{AB})}{4G_N},$$

where $A$0 lies inside the entanglement wedge $A$1 of $A$2 and cuts the wedge into two parts attached to $A$3 and $A$4 respectively (Liu et al., 2021, Chen et al., 2021). In AdS$A$5, the codimension-2 surface is a geodesic segment, so EWCS reduces to a length divided by $A$6 (Wen et al., 2022). If the entanglement wedge is disconnected, the minimal cross section does not exist and one sets $A$7; this is repeatedly tied to the vanishing of holographic mutual information (Tamaoka, 2018, Liu et al., 2021).

The same sources treat EWCS as a geometric measure of mixed-state correlations rather than a direct analog of von Neumann entropy. For pure bipartite states, it collapses to the ordinary entanglement entropy because the entanglement wedge cross section reduces to the RT surface (Tamaoka, 2018). For genuinely mixed states, however, it probes structure inside the entanglement wedge and is therefore sensitive to how the bulk region associated with $A$8 decomposes. This is the sense in which it refines subregion/subregion duality (Kusuki et al., 2019).

The covariant formulation in AdS$A$9 replaces the static minimal cut by an extremal geodesic chord in the HRT wedge. In the covariant treatment of two-dimensional CFTs, the relevant object is a saddle geodesic chord whose endpoints lie on the HRT surfaces bounding the wedge; in Einstein gravity its contribution is $B$0 (Wen et al., 2022). In AdS/BCFT, the definition is extended so that the relevant extremal surfaces and cross sections may end on the end-of-the-world brane $B$1, which generates a much richer phase structure while preserving the same formal role for EWCS (2206.13417).

A standard geometric consequence, already built into the definition, is the inequality

$B$2

which is used repeatedly as a consistency check and later generalized in AdS/BCFT (Umemoto, 2019, 2206.13417).

2. Boundary duals and interpretive status

Several distinct boundary quantities have been proposed to reproduce EWCS. One direct construction is the odd entanglement entropy $B$3, defined through an odd-replica continuation of the partially transposed density matrix. In AdS$B$4/CFT$B$5, the proposal is

$B$6

with explicit checks in vacuum AdS$B$7 and planar BTZ and a conjectural extension to higher dimensions (Tamaoka, 2018). This route is notable because it does not minimize over purifications.

A second line of work identifies EWCS with reflected entropy and balanced partial entanglement. In covariant two-dimensional CFT setups, including theories with gravitational anomaly, balanced partial entanglement, reflected entropy, and EWCS are found to coincide, with the normalization convention of that work implying $B$8 and $B$9 (Wen et al., 2022). The same analysis extends the discussion to topologically massive gravity and gives a prescription for a Chern–Simons correction to EWCS beyond Einstein gravity (Wen et al., 2022).

A third strand links EWCS to holographic negativity. In AdS$A\cup B$0/CFT$A\cup B$1, the proposal based on backreacted EWCS and reflected entropy is shown to be equivalent, modulo constants related to the Markov gap, to earlier geodesic-sum prescriptions for entanglement negativity. In that context, an alternative construction of the EWCS for a single interval at finite temperature is introduced to remove the thermal entropy contamination and match large-$A\cup B$2 CFT results up to the Markov-gap constant (Basak et al., 2020).

The interpretive status of EWCS is not settled by these dualities alone. One sustained argument is that EWCS is not a purely quantum entanglement measure. In particular, it can become larger than standard quantum entanglement measures and even larger than quantum discord in holographic states; this motivated the introduction of the entanglement wedge mutual information as a dual of the $A\cup B$3-correlation and led to the conclusion that EWCS captures classical correlations as well as quantum entanglement (Umemoto, 2019). A related result shows that if EWCS is dual to either reflected entropy or entanglement of purification, then the fact that those quantities can differ from mutual information at $A\cup B$4 implies that holographic CFT states require $A\cup B$5 tripartite entanglement, in direct tension with a “mostly bipartite” picture (Akers et al., 2019). This suggests that EWCS is best viewed as a geometric mixed-state quantity whose classical and multipartite content is essential rather than incidental.

3. Computation and extensions beyond Einstein gravity

For static homogeneous backgrounds of the form

$A\cup B$6

the strip-based EWCS problem is reduced to a codimension-2 surface $A\cup B$7 with area functional

$A\cup B$8

The decisive algorithmic point in several numerical studies is that the globally minimal cross section must be orthogonal to the RT surfaces that bound the wedge at its endpoints. The endpoint parameters are therefore determined by solving the orthogonality conditions $A\cup B$9, typically by Newton–Raphson iteration, after the minimal surfaces themselves have been obtained using spectral or Gauss–Lobatto discretization (Liu et al., 2021, Chen et al., 2021, Yang et al., 2023).

This orthogonality-based method is used in holographic massive gravity, Einstein–Aether gravity, and p-wave superconductors. In all three settings the geometry is static and translationally invariant along one boundary direction, so the problem reduces effectively to a two-dimensional profile, but the numerical difficulty is controlled by the same perpendicularity condition at the cross-section endpoints (Liu et al., 2021, Chen et al., 2021, Yang et al., 2023). In symmetric strip configurations, some perturbative analyses reduce EWCS to a radial integral between turning points of RT surfaces, which makes the role of shifted turning points especially transparent in excited geometries (Sahraei et al., 2021).

Beyond Einstein gravity, the definition itself becomes nontrivial. In five-dimensional Einstein–Gauss–Bonnet gravity, one study adopts a GB-corrected definition for EWCS by treating the cross section as a segment of a holographic entanglement entropy extremal surface and using the same higher-derivative functional rather than a pure area (Li et al., 2021). In topologically massive gravity, the Chern–Simons term induces an additional contribution

$A$0

and the proposed prescription fixes the relevant normal frame by using both the EWCS geodesic and the RT surfaces of the mixed state. The resulting anomaly contribution matches the anomalous part of balanced partial entanglement and reflected entropy (Wen et al., 2022). A plausible implication is that, beyond Einstein gravity, EWCS is not specified by the bulk metric alone.

AdS/BCFT adds a different type of complication. There, the candidate RT surfaces and candidate EWCS surfaces are allowed to end on the brane $A$1, so both the entanglement wedge and its cross section admit many more phases than in pure AdS/CFT. The relevant proofs are therefore organized in terms of candidate sets and wedge nesting rather than by enumerating pictures (2206.13417).

4. Inequalities, multipartite structure, and recent EWCS combinations

In AdS/BCFT, EWCS satisfies the same inequalities as in AdS/CFT despite the enlarged phase space. The proofs are algebraic and rely on wedge nesting, disjointness, and the definition of candidate sets for RT surfaces and cross sections. Besides the basic bounds $A$2 and $A$3, the same framework yields

$A$4

$A$5

and the multipartite inequality

$A$6

with $A$7 (2206.13417). The significance of this result is structural: the brane introduces abundant phase structures, yet the inequalities survive unchanged.

A complementary perspective comes from the claim that holographic mutual information behaves more like an entanglement-only quantity, whereas EWCS and the entanglement wedge mutual information encode optimized total correlations including classical pieces (Umemoto, 2019). In that sense the inequalities above do not merely bound a bipartite entanglement monotone; they constrain a more general correlation measure intrinsic to the entanglement wedge.

A recent multipartite extension is the EWCS triangle information

$A$8

This quantity is non-negative, is presented as an analogue of conditional mutual information, and is upper bounded by the entanglement of assistance in the canonical purification state (Ju et al., 25 Dec 2025). In AdS$A$9/CFT$B$0, maximizing $B$1 over configurations of the auxiliary subsystem $B$2 produces a rich phase structure governed by the cross ratio $B$3: it vanishes below a critical threshold and, beyond a second phase transition point, saturates the entanglement-of-assistance bound (Ju et al., 25 Dec 2025). This suggests that EWCS-based constructions now extend naturally into assisted and multipartite settings rather than remaining confined to two-party mixed states.

5. Dynamics, quenches, and excited geometries

EWCS has been studied dynamically both from the bulk and directly from two-dimensional holographic CFTs. In local operator quenches, the reflected entropy calculation shows that the dynamics of EWCS can be derived directly in the CFT. The comparison with mutual information and with RCFTs leads to the conclusion that classical correlation plays an important role in subregion/subregion duality even in dynamical setups; in holographic CFTs there are time ranges with $B$4 but $B$5, whereas in RCFTs $B$6 tracks the quasi-particle picture much more closely (Kusuki et al., 2019).

In a thermofield double state perturbed by a heavy operator insertion, the dual localized shock wave geometry yields an EWCS that matches the reflected entropy calculation for sufficiently late times. The time dependence exhibits a plateau before going to zero, closely resembling the behavior found in global quantum quenches, and at high temperatures this is captured by a line-tension picture (Boruch, 2020). This is one of the clearest dynamical examples in which EWCS remains nontrivial while the mixed-state geometry is undergoing a sharp reorganization.

Perturbative analyses around asymptotically AdS excited states reveal a different set of universal tendencies. For purely gravitational excitations, the leading correction to EWCS is negative, so the correlation between boundary subregions decreases. Current condensates give positive corrections, while scalar condensates can either decrease or increase EWCS depending on the operator dimension (Sahraei et al., 2021). One technical lesson of that analysis is that, unlike entanglement entropy, the leading correction to EWCS is sensitive not only to the metric perturbation itself but also to the motion of the RT turning points that bound the entanglement wedge (Sahraei et al., 2021).

Thermal quenches with higher-derivative corrections add another layer. In AdS–Vaidya–Gauss–Bonnet geometries, the static EWCS increases with the Gauss–Bonnet coupling $B$7, and in the dynamical case this monotonic relation with $B$8 persists at fixed time, but the disentanglement structure as a function of separation and time is no longer monotonic in the same way as in the static setup (Li et al., 2021). A plausible implication is that higher-derivative corrections reshape not only the magnitude of mixed-state correlations but also the order in which different correlation channels switch off during thermalization.

6. Phase transitions, critical scaling, and specialized holographic backgrounds

One major use of EWCS is as a probe of thermal and quantum-critical structure. In holographic massive gravity, where both first- and second-order Hawking–Page–type phase transitions occur, EWCS displays a van der Waals–like multivalued structure at first order and a singular behavior at the second-order critical point. Mutual information and holographic entanglement entropy diagnose the same transitions, but EWCS and mutual information show exactly opposite behavior in the critical region, while all three quantities share the same critical scaling

$B$9

The paper interprets this as evidence that EWCS captures degrees of freedom distinct from those captured by mutual information (Liu et al., 2021).

In Einstein–Aether holography, where Lorentz symmetry is broken by a timelike aether vector and the Lorentz-violation parameter enters through a $\rho_{AB}$0 deformation of the blackening factor, EWCS behaves qualitatively differently from both HEE and MI. Subject to the allowed positive-temperature parameter region, HEE increases monotonically with $\rho_{AB}$1, $\rho_{AB}$2, and $\rho_{AB}$3, and MI decreases monotonically with the same parameters, whereas EWCS is generically non-monotonic in $\rho_{AB}$4 and $\rho_{AB}$5, decreasing first and then increasing, while still decreasing monotonically with temperature (Chen et al., 2021). The paper explicitly attributes this to the competing influence of horizon motion and bulk geometric deformation on the shape of the entanglement wedge.

In holographic p-wave superconductors, EWCS cleanly detects both second- and first-order thermal transitions. At a second-order transition it is continuous while $\rho_{AB}$6 is discontinuous; at a first-order transition it jumps. Near the critical temperature, the deviations of both HEE and EWCS from their normal-phase values scale as

$\rho_{AB}$7

while the condensate scales with exponent $\rho_{AB}$8, so the geometric entanglement measures have exponents twice that of the order parameter (Yang et al., 2023). The same work emphasizes that EWCS is more sensitive than HEE in the sense that its quasi-critical exponent tracks the theoretical value over a broader temperature range, and it identifies a growth-rate inequality

$\rho_{AB}$9

near criticality in p-wave and s-wave superconducting transitions (Yang et al., 2023).

Taken together, these results support a broad but non-universal picture. EWCS reliably detects entanglement wedge connectivity changes, first- and second-order phase transitions, and dynamical disentangling transitions across a wide range of holographic models, but its monotonicity, critical exponents, and relation to MI or HEE depend on the mechanism by which the bulk geometry changes. This suggests that EWCS is best understood not as a single universal field-theoretic monotone, but as a geometric mixed-state observable whose detailed behavior tracks the causal, thermodynamic, and multipartite organization of holographic states.