Homology and Region Anchoring

An extremal surface does not identify an entropy until it is paired with the correct region. Anchoring fixes how the surface meets the boundary of the spacetime; homology demands an intervening bulk region whose boundary contains both the requested nongravitating region and the candidate surface. These global data determine which bulk algebra enters $S_{\mathrm{bulk}}$ and eliminate extrema belonging to a different entropy problem.

Required background. Quantum Information supplies subsystem complements, purification, and the algebra whose reduced entropy is being computed.

Helpful background. Semiclassical Gravity supplies generalized entropy; Quantum Fields in Curved Spacetime supplies regulated regional algebras; Black-Hole Geometry and Mechanics supplies causal regions, horizons, and codimension-two surfaces.

Anchoring and relative homology

Consider first a boundary region $A$ in a holographic spacetime. A candidate codimension-two surface $X$ is anchored on the entangling surface,

$$\partial X=\partial A,$$

with any end-of-the-world brane or other allowed boundary component stated separately. Anchoring is local boundary data. It does not say which side of $X$ belongs with $A$.

The homology condition supplies that side: there must exist an admissible bulk codimension-one region $r_X$ such that, as an unoriented relative boundary,

$$\partial r_X=A\cup X$$

up to explicitly permitted brane or asymptotic components. In a time-dependent spacetime, $r_X$ lies on an achronal slice and its domain of dependence defines the corresponding bulk region. The generalized entropy then uses the algebra in $r_X$, not an unnamed “inside.” The covariant extremal-surface construction and its causal setting are developed in Hubeny, Rangamani, and Takayanagi 2007, §§2–3; the quantum replacement by generalized entropy is stated in Engelhardt and Wall 2015, §§2–4.

Here the union is a boundary relation with $\mathbb Z_2$ coefficients: each component occurs once, with no orientation sign. With oriented chains one instead writes $\partial r_X=A-X$ after choosing the induced orientations. Charges, edge modes, or allowed brane endpoints can require additional data; changing those data changes the admissible region problem.

Radiation regions and islands

For an evaporating model coupled to a nongravitating bath, let $R$ be the collected radiation region. An island candidate $I$ is a region in the gravitating spacetime, usually with closed boundary $X=\partial I$. The candidate entropy is

$$S_{\mathrm{gen}}(R\cup I) = \frac{\operatorname{Area}(\partial I)}{4G} +S_{\mathrm{bulk}}(R\cup I) +S_{\mathrm{ct}}(\partial I).$$

The union $R\cup I$ is essential. Computing $S_{\mathrm{bulk}}(I)$ alone answers a different subsystem question and misses correlations between the island and radiation. The empty island $I=\varnothing$ is normally an admissible disconnected candidate, with no new area boundary, and competes with nonempty islands. This regional form and its use in radiation problems are given in Almheiri et al. 2021, §8, Eq. (8.2).

Disconnected components are allowed only when the boundary conditions and homology problem allow them. If $I=I_1\cup I_2$, every component contributes to the geometric term and the matter entropy is that of the joint algebra $\mathcal A(R\cup I_1\cup I_2)$. Additivity cannot be assumed: quantum mutual information couples the components.

Why extremization is not enough

Local stationarity cannot detect global topology. Two surfaces can satisfy the same quantum extremality equation while only one bounds a region with $A$. Conversely, several homologous extrema can coexist, and the entropy prescription still requires minimization. The logical order is

$$\text{boundary data} \longrightarrow \text{admissible homology classes} \longrightarrow \text{QESs in each class} \longrightarrow \min S_{\mathrm{gen}}.$$

This order prevents two common errors. One is to minimize over every geometric extremum, including a surface anchored to the wrong boundary component. The other is to define the bulk entropy before choosing $r_X$, which makes its algebra ambiguous.

For a globally pure regulated state, complementary boundary regions provide a useful check. If $X$ is admissible for $A$ and the same surface with opposite orientation is admissible for $A^c$, their paired bulk regions are complementary. Then $S_{\mathrm{bulk}}(r_X)=S_{\mathrm{bulk}}(r_X^c)$, and the geometric terms agree, giving $S(A)=S(A^c)$. If a proposed homology rule violates that equality in a pure state, either a boundary component, center sector, or candidate class has been omitted.

Boundary cases and limitations

End-of-the-world branes, islands ending on interfaces, gauge-theory centers, and fully gravitating asymptotic regions require relative boundary conditions beyond the schematic equation above. Homology is also not a dynamical mechanism: it selects candidate regions but does not explain how radiation degrees of freedom microscopically encode an island.

The prescription assumes an achronal region and a suitable domain of dependence. In spacetimes without the necessary global hyperbolicity, or when singularities obstruct all admissible slices, existence can fail. Generic QFT factorization and detailed black-hole geometry remain separate prerequisites rather than conclusions of the homology rule.

Exercises

Suppose the empty island has $S_{\mathrm{gen}}(R)=s$, while a nonempty island $I$ has area term $A_I/(4G)$ and regulated matter entropies $S(R)$, $S(I)$, and mutual information $I(R:I)$. Express the island candidate in these quantities and explain why adding $S(R)+S(I)$ would give the wrong dominance test.

Solution

By definition, $S(R\cup I)=S(R)+S(I)-I(R:I)$. The island candidate is therefore $A_I/(4G)+S(R)+S(I)-I(R:I)+S_{\mathrm{ct}}$. Omitting the mutual information treats the two regional algebras as uncorrelated and overestimates the matter entropy whenever $I(R:I)\gt0$. Since those correlations are precisely what can make an island competitive, the additive replacement can reverse which homologous QES minimizes the generalized entropy.

References

• Almheiri et al. 2021, Ahmed Almheiri, Thomas Hartman, Juan Maldacena, Edgar Shaghoulian, and Amirhossein Tajdini, “The Entropy of Hawking Radiation,” Reviews of Modern Physics 93 (2021) 035002.
• Engelhardt and Wall 2015, Netta Engelhardt and Aron C. Wall, “Quantum Extremal Surfaces: Holographic Entanglement Entropy beyond the Classical Regime,” Journal of High Energy Physics 01 (2015) 073.
• Hubeny, Rangamani, and Takayanagi 2007, Veronika E. Hubeny, Mukund Rangamani, and Tadashi Takayanagi, “A Covariant Holographic Entanglement Entropy Proposal,” Journal of High Energy Physics 07 (2007) 062.