Generalized Ryu–Takayanagi Formula

• Generalized RT is a framework that extends the classical entropy-area relation by incorporating quantum corrections and geometric modifications through an extremization procedure.
• It unifies diverse prescriptions such as covariant extremal surfaces, semiclassical entropy corrections, higher-derivative functionals, and defect-localized contributions.
• This formulation bridges holography with algebraic methods and tensor-network analogues, significantly impacting our understanding of entanglement in quantum gravity.

The generalized Ryu–Takayanagi formula denotes a family of extensions, refinements, and reformulations of the original relation between boundary entanglement entropy and a bulk codimension-2 surface. In its classical Einstein-gravity form, the Ryu–Takayanagi prescription assigns to a boundary subregion $A$ the entropy $S(A)=\mathrm{Area}(\gamma_A)/(4G_N)$, with $\gamma_A$ the appropriate minimal or extremal homologous surface. In subsequent literature, “generalized” has acquired several distinct meanings: a covariant extremal-surface prescription, a semiclassical extremization of a geometric term plus $S_{\rm bulk}$, higher-derivative entropy functionals, algebraic and quantum-error-correcting formulations, finite-cutoff prescriptions, defect-localized generalizations, and more abstract phase-space or tensor-network analogues (Averin, 20 Aug 2025).

1. Classical baseline: RT, HRT, and the geometric structure of holographic entropy

In the static setting, the Ryu–Takayanagi formula can be written as a minimization over bulk regions $r\subseteq\Sigma$ with prescribed boundary footprint $A\subseteq\dot\Sigma$,

$S(A)=\min_{r\subseteq \Sigma:\dot r=A}\area(\partial r),$

or, equivalently, as

$S_B=\frac{A(\tilde B)}{4G_N^{(d+1)}},$

with $\tilde B$ the least-area codimension-2 bulk surface homologous to the boundary region $B$ (Headrick, 2013, Jaksland, 2017). In Lorentzian spacetimes, the natural covariant generalization is the Hubeny–Rangamani–Takayanagi prescription,

$S(A)=\mathrm{Area}(\gamma_A)/(4G_N)$0

which replaces slice-wise minimality by spacetime extremality (Callan et al., 2012).

At the purely classical level, the RT formula already implies a substantial algebraic structure for entropy. In a fixed static asymptotically AdS spacetime, it gives continuity of $S(A)=\mathrm{Area}(\gamma_A)/(4G_N)$1 under continuous deformations of $S(A)=\mathrm{Area}(\gamma_A)/(4G_N)$2, positivity, subadditivity, Araki–Lieb, strong subadditivity, and monogamy of mutual information, together with structural properties of the map $S(A)=\mathrm{Area}(\gamma_A)/(4G_N)$3, such as monotonicity and disjointness for disjoint boundary regions (Headrick, 2013). A key point is that monogamy of mutual information,

$S(A)=\mathrm{Area}(\gamma_A)/(4G_N)$4

is a specifically holographic consequence of the classical area term rather than a generic quantum-information inequality.

The covariant formula is more delicate. In $S(A)=\mathrm{Area}(\gamma_A)/(4G_N)$5-Vaidya spacetimes, strong subadditivity is obeyed when the bulk matter satisfies the null energy condition, $S(A)=\mathrm{Area}(\gamma_A)/(4G_N)$6, and violations of the null energy condition correlate with violations of one form of strong subadditivity (Callan et al., 2012). This sharpened the view that the classical extremal-surface prescription behaves as a genuine entropy formula only under geometric conditions that control the causal and focusing properties of the bulk.

2. Semiclassical generalization: from additive bulk entropy to quantum extremality

A historically intermediate step beyond classical RT is the additive correction

$S(A)=\mathrm{Area}(\gamma_A)/(4G_N)$7

where $S(A)=\mathrm{Area}(\gamma_A)/(4G_N)$8 is the bulk entanglement entropy across the classical extremal surface $S(A)=\mathrm{Area}(\gamma_A)/(4G_N)$9. In the review of entanglement and linearized gravity, this is presented as the leading $\gamma_A$0 correction due to bulk matter fields, but without the later quantum-extremal-surface language (Jaksland, 2017). That stage is conceptually close to generalized entropy, but it still treats the correction as an additive modification to the classical surface.

A more systematic generalization is provided by the phase-space proof for arbitrary diffeomorphism-invariant field theories. There the entropy is first expressed exactly as a phase-space functional-integral expectation value,

$\gamma_A$1

with

$\gamma_A$2

At leading order in $\gamma_A$3, this becomes the extremization prescription

$\gamma_A$4

and beyond leading order it becomes

$\gamma_A$5

with

$\gamma_A$6

For Einstein gravity, $\gamma_A$7, so the leading term reproduces RT and HRT, while the corrected formula reproduces the quantum extremal surface prescription (Averin, 20 Aug 2025).

In this sense, the modern generalized RT formula is not merely “area plus a correction.” It is an extremization of a generalized entropy functional whose geometric part is theory-dependent and whose quantum part is $\gamma_A$8. The phase-space treatment is noteworthy because it does not require a holographic dual: holography appears as a special case of a more general diffeomorphism-invariant construction (Averin, 20 Aug 2025).

3. Higher-derivative and defect-localized entropy functionals

In higher-derivative gravity, the classical area functional is generally not the correct entropy functional. For curvature-squared theories, the replica derivation yields a codimension-2 functional consisting of Wald entropy plus a correction quadratic in the extrinsic curvature,

$\gamma_A$9

The first term is Wald entropy,

$S_{\rm bulk}$0

but for generic holographic entangling surfaces this is incomplete because nonzero extrinsic curvature produces additional contributions. In Gauss–Bonnet gravity the resulting functional reduces to the Jacobson–Myers expression (Camps, 2013).

This higher-derivative generalization is important for terminology. In this literature, a “generalized RT formula” may mean that the surface is still found by extremization, but the object being extremized is no longer area and not even Wald entropy alone. The correction is local on the surface and is fixed by the second derivative of the Lagrangian with respect to the Riemann tensor (Camps, 2013).

A different generalization arises when the bulk contains a defect brane or string carrying its own quantum degrees of freedom. The defect extremal surface prescription is

$S_{\rm bulk}$1

Here the usual RT/HRT area term is corrected by the entropy of the defect theory across the intersection $S_{\rm bulk}$2. For reflected entropy, the analogous proposal extremizes

$S_{\rm bulk}$3

In $S_{\rm bulk}$4, this defect extremal cross section formula matches the island formula for reflected entropy in both static and evaporating-black-hole examples (Li et al., 2021).

These developments show that generalized RT need not only mean adding $S_{\rm bulk}$5. It can also mean replacing the area law by a different local geometric functional, or augmenting it with defect-localized quantum information while preserving an extremal-surface logic.

4. Finite cutoffs, renormalization, and nonstandard holographic settings

One important clarification concerns holography away from the asymptotic AdS boundary. In holographic duals of $S_{\rm bulk}$6-deformed CFTs, the argument is not that a new entanglement prescription is required, but that if the duality obeys a GKPW-like dictionary with Dirichlet conditions at a finite radial cutoff, then the Lewkowycz–Maldacena derivation goes through essentially unchanged. The classical entropy of the bare $S_{\rm bulk}$7-deformed theory is therefore still

$S_{\rm bulk}$8

with $S_{\rm bulk}$9 now anchored on the cutoff surface rather than the asymptotic boundary (Murdia et al., 2019).

The crucial distinction in that setting is between bare and renormalized entropy. Boundary counterterms shift the entropy by local geometric terms on the entangling surface,

$r\subseteq\Sigma$0

but they do not alter the location of the RT surface. In this usage, “generalized RT” does not denote a new dynamical extremization rule; it denotes the usual area term together with optional local entangling-surface counterterm contributions (Murdia et al., 2019).

A different nonstandard setting is provided by multi-boundary AdS$r\subseteq\Sigma$1/CFT$r\subseteq\Sigma$2 states prepared by path integrals on bordered Riemann surfaces. For Hartle–Hawking states of multi-boundary black holes, the classical RT formula and its phase structure can be derived directly from boundary CFT data by approximating heavy OPE coefficients by Gaussian moments in the large-$r\subseteq\Sigma$3 ensemble. In the three-boundary pair-of-pants state, the entropy of one boundary obeys

$r\subseteq\Sigma$4

and each RT phase arises from a distinct leading-order Gaussian contraction pattern in the replica computation (Bao et al., 16 Apr 2025). This is still a classical RT derivation, not a derivation of $r\subseteq\Sigma$5, but it broadens the meaning of “generalized” to include multipartite and multi-phase entanglement structures reconstructed entirely from boundary data.

5. Algebraic and quantum-error-correcting formulations

A major line of generalization replaces geometric surfaces by operator-algebraic structures. In quantum error correction, a code subspace with complementary recovery obeys a quantum-corrected RT formula of the form

$r\subseteq\Sigma$6

where $r\subseteq\Sigma$7 is a von Neumann algebra on the code subspace, $r\subseteq\Sigma$8 its commutant, and $r\subseteq\Sigma$9 is a central “area operator” (Harlow, 2016). In this formulation, generalized RT is equivalent to complementary recovery and to equality of bulk and boundary relative entropies.

The algebraic extension to non-factorizing boundary theories replaces the boundary tensor factor by a boundary von Neumann algebra $A\subseteq\dot\Sigma$0, possibly with nontrivial center. In that setting the natural entropy is the algebraic entropy

$A\subseteq\dot\Sigma$1

and complementary recovery implies

$A\subseteq\dot\Sigma$2

with $A\subseteq\dot\Sigma$3 a central operator. By contrast, the distillable boundary entropy does not obey such an RT formula, while an entropy modified by a $A\subseteq\dot\Sigma$4 term does obey one, but with a different area operator that is interpreted as a cutoff-sensitive regularization effect (Kamal et al., 2019).

A further algebraic generalization appears in the type I/II factor setting. There the algebraic reconstruction theorem can be strengthened so that entanglement wedge reconstruction, JLMS relative-entropy matching, and an algebraic RT statement become equivalent. The entropy equality is

$A\subseteq\dot\Sigma$5

for cyclic and separating states and type I/II factors (Xu et al., 2024). This theorem does not contain an explicit area operator; the physical interpretation offered there is that the algebraic entropy on the physical algebra corresponds to a renormalized boundary entropy after subtraction of the area term, not to the bare geometric entropy itself.

These algebraic developments clarify a common misconception. In one strand of the literature, the generalized RT formula is not primarily a formula for a spacetime area plus bulk entropy; it is an exact statement about algebras, centers, and recoverability, with the geometric area term represented by a central operator or by an implicit subtraction.

6. Broader extensions: tensor networks, many-body systems, null flows, and phase-space microphysics

Outside semiclassical AdS/CFT, the phrase “generalized RT formula” is also used for RT-inspired entanglement–geometry relations in discrete, algebraic, or non-gravitational settings. In flow equation holography, the proposal is a perturbative formula for generic quantum many-body systems in which the subsystem’s min-entropy equals, to leading order, the integrated disentangling density along a unitary RG-like flow,

$A\subseteq\dot\Sigma$6

and in one dimension this can be rewritten as a minimal-curve problem in an emergent metric (Kehrein, 2017).

In tensor-network and group-field-theoretic settings, RT-type laws are obtained from minimal cuts in large bond dimension. For generalized GFT/random-tensor networks one finds

$A\subseteq\dot\Sigma$7

while for fixed-spin spin-network states one obtains

$A\subseteq\dot\Sigma$8

For symmetric random tensor networks with a local gauge-like invariance, the leading second Rényi entropy is

$A\subseteq\dot\Sigma$9

and for simple interacting group field theories the leading law

$S(A)=\min_{r\subseteq \Sigma:\dot r=A}\area(\partial r),$0

survives at linear order in the interaction for a broad class of graphs (Chirco et al., 2017, Chirco et al., 2017, Chirco et al., 2019). In this family of works, “generalized RT” means that the minimal-cut law survives the passage from unconstrained random tensors to gauge-invariant, spin-network-compatible, and weakly interacting pre-geometric states.

A geometric reformulation appears in the null-flow approach. In static, spherically symmetric, asymptotically AdS vacuum spacetimes, wave fronts of null rays emitted from a point on the AdS boundary become extremal surfaces because their mean curvature is proportional to the null expansion, which vanishes as the affine parameter goes to infinity. In BTZ, the wave front coincides with the RT surface, and the area can be rewritten as a flux of the Klein–Gordon current (Tsujimura et al., 2020). This is not a full HRT theorem, but it is a concrete flow-based reinterpretation of RT.

A more radical phase-space generalization identifies the microscopic degrees of freedom responsible for black-hole entropy. For stationary black holes with bifurcate Killing horizons, the generalized RT prescription gives, to leading order,

$S(A)=\min_{r\subseteq \Sigma:\dot r=A}\area(\partial r),$1

and the relevant microstates are phase-space states distinguished by Hamiltonian surface charges on the bifurcation surface (Averin, 31 Mar 2026). This proposal treats generalized RT as a phase-space statement rather than only as a relation between boundary entropy and bulk geometry.

Across these settings, a single caution is essential. “Generalized Ryu–Takayanagi formula” does not denote one universally fixed formula. In some works it means the semiclassical generalized entropy extremized over quantum extremal surfaces; in others it means higher-derivative entropy functionals, counterterm-renormalized area laws, algebraic entropy equalities, defect-corrected prescriptions, minimal-cut laws in tensor networks, or phase-space formulas with no assumed holographic dual. What unifies these constructions is not a single expression, but the persistence of an extremal or minimal codimension-2 structure that organizes entanglement in a geometrically or algebraically controlled way.