Covariant Holographic Entropy Cone

The holographic entropy cone classifies the possible entanglement structures of quantum states with a classical gravity dual. For static geometries, Bao et al. established that this cone is polyhedral by constructing a graph model from Ryu-Takayanagi (RT) surfaces on a time-symmetric slice. Extending this framework to general, time-dependent states governed by the Hubeny-Rangamani-Takayanagi (HRT) formula has remained an open problem, as the relevant extremal surfaces do not lie on a common spatial slice. We resolve this by constructing a graph model directly from the causal structure of entanglement wedges. By proving a key “no-short-cut” theorem, we show that minimization over graph cuts reduces to a consideration of cuts corresponding to unions of complete HRT surfaces, establishing the equivalence of the covariant and static holographic entropy cones. Consequently, all foundational results, including polyhedrality and the finite nature of entropy inequalities, extend to general holographic states.

💡 Research Summary

The paper tackles a long‑standing open problem in holographic quantum information: extending the classification of entanglement entropy vectors—the holographic entropy cone—from static, time‑symmetric bulk geometries to fully dynamical spacetimes. In static settings, Bao et al. showed that the cone is polyhedral by constructing a graph model from Ryu‑Takayanagi (RT) minimal surfaces on a single Cauchy slice and applying max‑flow/min‑cut theorems. However, for time‑dependent states the relevant extremal surfaces are given by the Hubeny‑Rangamani‑Takayanagi (HRT) prescription, which generally do not lie on a common spatial slice, preventing a straightforward graph construction.

The authors resolve this by building a graph directly from the causal structure of entanglement wedges. They partition the bulk using the boundaries of all entanglement wedges—called entanglement horizons. Each bulk region bounded by portions of these horizons becomes a vertex; an edge connects two vertices when their corresponding regions share a horizon segment, and the edge weight is the area of that segment (divided by 4 G_N). Boundary regions are encoded as colored vertices, and a “cut” of the graph corresponds to a hypersurface assembled from pieces of entanglement horizons. The discrete entropy S* of a collection of boundary regions is defined as the minimal total weight over all cuts that separate exactly those colored vertices.

Two technical lemmas are proved first. Lemma 2.2 (the “no‑multiple‑crossing” lemma) shows that a single HRT surface cannot enter, exit, and re‑enter the same entanglement wedge through the same connected component of its horizon; this follows from entanglement‑wedge nesting, the null energy condition, and a strong comparison principle for minimal‑surface operators. Lemma 2.4 addresses the case of two intersecting entanglement horizons: the sum of the “outer” portions of the two HRT surfaces is always at least as large as the area of the HRT surface for the union of the two boundary regions. The proof uses projection of the outer pieces along null generators onto a common Cauchy slice, which never increases area.

Armed with these lemmas, the central “No‑Short‑Cut” theorem (Theorem 2.3) is established. It states that any graph cut built from arbitrary partial horizon segments can be replaced by a cut built solely from complete, non‑intersecting HRT surfaces without increasing the total weight. The proof proceeds iteratively: whenever a cut contains pieces from two intersecting horizons, Lemma 2.4 is applied to replace those pieces by the outer parts of the corresponding full HRT surfaces, thereby reducing (or leaving unchanged) the total weight. Repeating this process eliminates all partial pieces, leaving a cut that is precisely a union of full HRT surfaces. Consequently, the discrete entropy S* coincides exactly with the holographic entropy computed via the HRT prescription.

Since the discrete entropy obeys the same linear inequalities as the graph‑theoretic min‑cut, the holographic entropy cone for dynamical spacetimes inherits the polyhedral structure proved for static cases. All previously known entropy inequalities (e.g., strong subadditivity, monogamy of mutual information, etc.) remain valid, and no new independent inequalities arise from time dependence. Thus the covariant holographic entropy cone is identical to the static one, confirming that the set of universal holographic entropy constraints is finite and fully captured by the polyhedral cone.

The paper concludes with a discussion of implications: the graph model provides a concrete combinatorial representation of covariant entanglement structures, opening avenues for numerical studies of dynamical holographic states, extensions to higher‑derivative gravity, and explorations of quantum corrections. It also suggests that the entanglement‑wedge nesting and causal structure alone are sufficient to enforce the full set of holographic entropy inequalities, highlighting a deep interplay between geometry, causality, and quantum information in AdS/CFT.