Holographic Incompressibility in Quantum Gravity

• Holographic incompressibility is a principle where holographic states and geometries obey strict entropy constraints, excluding many quantum-allowable patterns.
• It reveals that holographic entanglement structures, entropy cones, and tensor network models enforce multipartite rather than pairwise correlation, marking a clear quantum–holographic separation.
• The concept underpins rigorous bounds in bulk reconstruction, error correction, and probe brane dynamics, challenging naïve UV/IR coarse-graining with precise geometric and entropic limits.

Searching arXiv for the cited papers and closely related work on holographic entropy cones, probe branes, and holographic codes. Holographic incompressibility is an interpretive label for a family of rigidity phenomena in holography. In this usage, it denotes the fact that holographic states, geometries, and codes occupy a highly constrained subset of the larger quantum-information landscape: not every quantum-mechanically allowed entropy pattern is holographically realizable; nontrivial holographic extreme rays are forced into irreducibly multipartite sectors; exact saturation of holographic entropy inequalities can forbid specific erasure-correction patterns; naïve UV/IR coarse-graining fails to capture all bulk states inside a finite region; and probe-brane consistency requires an area–volume inequality that prevents overly efficient enclosure of bulk volume (He et al., 2023, He et al., 2020, Czech et al., 17 Feb 2025, Rosenhaus, 2013, Ferrari et al., 2014, Cheng, 6 Jun 2025). The phrase does not denote a single theorem or a single compression bound. Rather, it organizes several precise statements of holographic rigidity across entropy cones, entanglement wedges, AdS/CFT state selection, Euclidean probe dynamics, and holographic tensor networks.

1. Entropy-space formulation and the subadditivity cone

A central formulation of holographic incompressibility arises in entropy space. For an $(N+1)$-partite pure system with $N$ named parties and one purifier, one considers the entropy vector

$\vec S \in \mathbb{R}^{2^N-1},$

whose components are the von Neumann entropies $S_I$ of all nonempty subsets $I\subseteq [N]$. The weakest outer approximation relevant here is the subadditivity cone (SAC), cut out by

$$S_{\underline I}+S_{\underline J}-S_{\underline I\underline J}\ge 0, \qquad \forall\, \underline I,\underline J\subseteq [N+1],\quad \underline I\cap \underline J=\varnothing .$$

Because

$I(X:Y)=S_X+S_Y-S_{XY},$

each saturated subadditivity inequality corresponds to vanishing mutual information and hence, in quantum mechanics, to factorization or independence of the corresponding subsystems. Faces of the SAC therefore encode patterns of marginal independence, and its one-dimensional faces, the extreme rays, represent extremal such patterns (He et al., 2023).

The distinction relevant to holography is between quantum realizability and holographic realizability. Quantum realizability means that an entropy vector is the von Neumann entropy vector of some quantum state. Holographic realizability means that it is realized by a holographic graph model, equivalently by the holographic entropy cone defined from RT/HRRT-type min-cut models. Since the holographic entropy cone sits inside the quantum entropy cone, every holographic entropy vector is quantum, but not conversely. Holographic incompressibility, in this entropic sense, is the statement that holography excludes some extremal independence structures that quantum mechanics permits even at the level of the SAC, which only tracks vanishing and nonvanishing pairwise mutual informations (He et al., 2023).

This framing is significant because an extreme ray is maximally rigid within the SAC: if a vector lies on an extreme ray, it cannot be decomposed into nonparallel positive combinations of other vectors in the cone. A gap at this level shows that holography is not obtained merely by imposing the universal quantum inequalities and then selecting a convenient geometric realization. Instead, the allowed holographic correlation architecture is more restricted already at the level of extremal independence data.

2. Quantum–holographic separation at six parties

The sharpest entropy-cone manifestation of holographic incompressibility is the explicit separation between quantum and holographic extreme rays of the SAC. A conjecture formulated in (Hernández-Cuenca et al., 2022) proposed that, for any $N$, every extreme ray of the $N$-party SAC realizable by quantum states should also be realizable by graph models. That conjecture was known to hold for $N\le 5$. The separation first appears at six parties: for $N$0, there exist extreme rays of the $N$1-party SAC that are realizable by quantum states but not by holographic states, and an explicit example already exists for $N$2 (He et al., 2023).

For the explicit six-party ray, the saturated subadditivity equalities have rank $N$3, where $N$4, showing that the ray is indeed extreme. The crucial obstruction is not a failure of subadditivity or of strong-subadditivity-compatible criteria, but violation of a genuinely holographic inequality, monogamy of mutual information (MMI). The violated instance is

$N$5

equivalently

$N$6

with

$N$7

Since holographic entropy vectors obey

$N$8

this single violation proves that the ray is not holographically realizable (He et al., 2023).

The proof of quantum realizability uses hypergraph entropy models rather than ordinary graphs. A hypergraph model allows hyperedges joining three or more vertices, and entropy is again the minimum cut cost. The construction begins from the observation that the ray violates exactly one MMI inequality and that a prototypical source of MMI violation is a GHZ-type hyperedge. The model therefore starts from a weight-$N$9 hyperedge joining coarse-grained subsystems

$\vec S \in \mathbb{R}^{2^N-1},$0

with $\vec S \in \mathbb{R}^{2^N-1},$1 the purifier boundary vertex, and augments this with ordinary weighted edges. Quantum realizability then follows from a theorem of Walter–Witteveen: every entropy vector realizable by a hypergraph model is realized by a quantum stabilizer state (He et al., 2023).

This result is the most direct entropic statement of holographic incompressibility. It shows that holographic states admit fewer extremal ways of arranging vanishing pairwise mutual informations together with higher-order correlations. A plausible implication is that geometric min-cut structure forbids certain “GHZ-like” extremal packings of correlation that remain fully admissible in generic quantum theory.

3. Superbalance, Bell-pair exclusion, and irreducible multipartiteness

A second line of work sharpens holographic incompressibility by constraining the extreme rays and inequalities of the holographic entropy cone itself. For an entropy inequality

$\vec S \in \mathbb{R}^{2^N-1},$2

one defines

$\vec S \in \mathbb{R}^{2^N-1},$3

The inequality is balanced iff

$\vec S \in \mathbb{R}^{2^N-1},$4

It is superbalanced iff it and all its purifications are balanced; equivalently, in the $\vec S \in \mathbb{R}^{2^N-1},$5-basis it contains only terms of rank at least $\vec S \in \mathbb{R}^{2^N-1},$6,

$\vec S \in \mathbb{R}^{2^N-1},$7

In the $\vec S \in \mathbb{R}^{2^N-1},$8-basis, superbalance is equivalent to the absence of $\vec S \in \mathbb{R}^{2^N-1},$9 terms, namely Bell-pair directions (He et al., 2020).

The key structural theorem states that if $S_I$0 is any non-Bell-pair extreme ray of the $S_I$1-party holographic entropy cone $S_I$2, written in the $S_I$3-basis as

$S_I$4

then

$S_I$5

From this it follows that all non-redundant holographic entropy inequalities other than subadditivity are superbalanced. The geometric consequence is that the remaining extreme rays correspond to geometries with vanishing mutual information between any two parties,

$S_I$6

and therefore contain no Bell-pair-type entanglement (He et al., 2020).

This establishes a precise irreducibility statement. Outside the subadditivity/Bell-pair sector, the nontrivial extreme structure of the holographic entropy cone is forced into a genuinely multipartite regime. Holographic incompressibility, in this sense, means that the interesting part of holographic entanglement cannot be built from pairwise-entangled atoms. The result does not classify all extreme rays and does not prove that no simpler non-pairwise primitives exist, but it does rigorously exclude Bell-pair decomposition as the microscopic language for nontrivial holographic extreme structure (He et al., 2020).

4. Entropy inequalities, wedge overlaps, and erasure-correction rigidity

A more operational version of holographic incompressibility appears in holographic error correction. Consider a holographic entropy inequality of the form

$S_I$7

with positive coefficients, together with a contraction proof $S_I$8. Define bulk regions

$S_I$9

which partition the bulk time slice according to membership in the entanglement wedges of the $I\subseteq [N]$0. Here $I\subseteq [N]$1 means that there exists a bulk region with a specific pattern of recoverability and erasability across the boundary subsystems (Czech et al., 17 Feb 2025).

The main statement is that if a candidate wedge produced by the contraction map violates entanglement wedge nesting, and the inequality is saturated, then

$I\subseteq [N]$2

Equivalently, non-saturation of the inequality is a necessary condition for the corresponding erasure-correction pattern to exist. The mechanism is geometric. A contraction proof constructs candidate wedges $I\subseteq [N]$3 from the $I\subseteq [N]$4. If some nonempty $I\subseteq [N]$5 would force one of these candidate wedges to violate nesting, then the candidate cannot equal the physical wedge. But if the entropy inequality is saturated, the candidate wedges must realize the true minimal entropies, hence must coincide with the physical wedges. This contradiction forces $I\subseteq [N]$6 to vanish (Czech et al., 17 Feb 2025).

The five-party cyclic inequality furnishes a concrete example. If

$I\subseteq [N]$7

then the bulk region

$I\subseteq [N]$8

must be empty. Thus exact saturation forbids a selective redundancy pattern in which the same bulk information would be recoverable from some three-party boundary regions but not from others (Czech et al., 17 Feb 2025).

This is an incompressibility statement in the language of coding theory. Holographic redundancy is real, because a bulk operator may lie in several entanglement wedges, but it is not arbitrary. Some desired support patterns cannot coexist with saturated entropy geometry. The paper also emphasizes an exception: MMI is the only known inequality whose contraction proof need not invoke nesting-violating candidate wedges, so this no-go mechanism does not apply to MMI in the same way (Czech et al., 17 Feb 2025).

5. UV/IR failure and state-space truncation in AdS/CFT

A different notion of holographic incompressibility concerns whether the physics inside a finite bulk region can be captured by a coarse-grained set of boundary degrees of freedom. In global AdS,

$I\subseteq [N]$9

one may ask whether the bulk region

$$S_{\underline I}+S_{\underline J}-S_{\underline I\underline J}\ge 0, \qquad \forall\, \underline I,\underline J\subseteq [N+1],\quad \underline I\cap \underline J=\varnothing .$$0

is describable by a boundary theory truncated to spatial resolution $$S_{\underline I}+S_{\underline J}-S_{\underline I\underline J}\ge 0, \qquad \forall\, \underline I,\underline J\subseteq [N+1],\quad \underline I\cap \underline J=\varnothing .$$1. The UV/IR proposal attempts this by putting the CFT on a lattice whose spacing is of order $$S_{\underline I}+S_{\underline J}-S_{\underline I\underline J}\ge 0, \qquad \forall\, \underline I,\underline J\subseteq [N+1],\quad \underline I\cap \underline J=\varnothing .$$2, motivated by matching the number of boundary degrees of freedom to the area of the cutoff sphere (Rosenhaus, 2013).

Explicit counterexamples show that this naïve compression scheme fails. For an oscillating relativistic particle in AdS, the turning point obeys

$$S_{\underline I}+S_{\underline J}-S_{\underline I\underline J}\ge 0, \qquad \forall\, \underline I,\underline J\subseteq [N+1],\quad \underline I\cap \underline J=\varnothing .$$3

and the induced boundary stress tensor is concentrated on a shell of thickness

$$S_{\underline I}+S_{\underline J}-S_{\underline I\underline J}\ge 0, \qquad \forall\, \underline I,\underline J\subseteq [N+1],\quad \underline I\cap \underline J=\varnothing .$$4

As the particle passes through the center of AdS, the shell remains thin, so for $$S_{\underline I}+S_{\underline J}-S_{\underline I\underline J}\ge 0, \qquad \forall\, \underline I,\underline J\subseteq [N+1],\quad \underline I\cap \underline J=\varnothing .$$5 the UV/IR-lattice CFT cannot resolve the boundary image even though the particle traverses the supposedly describable bulk region (Rosenhaus, 2013).

A second counterexample uses a relativistic scalar wavepacket in the near-boundary Poincaré approximation

$$S_{\underline I}+S_{\underline J}-S_{\underline I\underline J}\ge 0, \qquad \forall\, \underline I,\underline J\subseteq [N+1],\quad \underline I\cap \underline J=\varnothing .$$6

The packet is constructed so that its boundary imprint is

$$S_{\underline I}+S_{\underline J}-S_{\underline I\underline J}\ge 0, \qquad \forall\, \underline I,\underline J\subseteq [N+1],\quad \underline I\cap \underline J=\varnothing .$$7

with transverse width remaining of order $$S_{\underline I}+S_{\underline J}-S_{\underline I\underline J}\ge 0, \qquad \forall\, \underline I,\underline J\subseteq [N+1],\quad \underline I\cap \underline J=\varnothing .$$8 for

$$S_{\underline I}+S_{\underline J}-S_{\underline I\underline J}\ge 0, \qquad \forall\, \underline I,\underline J\subseteq [N+1],\quad \underline I\cap \underline J=\varnothing .$$9

Choosing

$I(X:Y)=S_X+S_Y-S_{XY},$0

and sufficiently large $I(X:Y)=S_X+S_Y-S_{XY},$1, the packet moves well inside the bulk region while its boundary image remains localized on a scale much smaller than the UV/IR cutoff (Rosenhaus, 2013).

A third counterexample is static. In AdS$I(X:Y)=S_X+S_Y-S_{XY},$2, a heavy scalar mode can satisfy

$I(X:Y)=S_X+S_Y-S_{XY},$3

so the mode is localized near the center by the AdS confining potential, while its boundary expectation value oscillates on angular scales much smaller than $I(X:Y)=S_X+S_Y-S_{XY},$4. More generally, a truncated angular basis with $I(X:Y)=S_X+S_Y-S_{XY},$5 cannot represent a generic bulk field localized inside a finite region (Rosenhaus, 2013).

These examples imply that deep bulk localization does not generally correspond to broad boundary support at the level relevant for faithful reconstruction. The alternative proposed is not a tensor-factor truncation of boundary modes but a restriction to a low-energy subspace of CFT states: $I(X:Y)=S_X+S_Y-S_{XY},$6 where $I(X:Y)=S_X+S_Y-S_{XY},$7 is the mass of a black hole of radius $I(X:Y)=S_X+S_Y-S_{XY},$8. This suggests that any finite holographic description of a bulk region is more naturally a subspace of states than a spatially coarse-grained boundary theory (Rosenhaus, 2013).

6. Probe branes, isoperimetry, and area–volume resistance

A geometric version of holographic incompressibility arises from Euclidean probe branes in Poincaré–Einstein spaces. Boundary large-$I(X:Y)=S_X+S_Y-S_{XY},$9 scaling implies a relation between the minimum Euclidean probe-brane action $N$0 and the Euclidean on-shell gravitational action $N$1,

$N$2

In the bulk derivation, the probe brane action takes the form

$N$3

where $N$4 is a hypersurface homologous to the boundary, $N$5 is its area, and $N$6 is the enclosed bulk volume. After fixing the additive ambiguity by an asymptotic holographic renormalization condition, one obtains

$N$7

The holographic identity is therefore reduced to the geometric problem of minimizing $N$8 (Ferrari et al., 2014).

The decisive inequality is

$N$9

When it holds, the infimum of $N$0 is $N$1, attained by a shrunk brane, and the desired proportionality between $N$2 and $N$3 follows. The inequality is proved in the relevant setting by results of Lee and Wang, and its validity depends on the conformal boundary having nonnegative Yamabe invariant. If the Yamabe constant is negative, the inequality fails, $N$4 can run to $N$5, and the probe-brane action becomes unbounded below, signaling instability (Ferrari et al., 2014).

This is an incompressibility statement in a literal geometric sense. The bulk does not permit a homologous hypersurface to enclose arbitrarily large volume at insufficient area cost. The cosmological term in the brane action favors expansion through the volume contribution, while the DBI term penalizes area. Stable holographic geometry enforces a fixed area–volume ratio, so volume gain can never outstrip area cost enough to destabilize the background (Ferrari et al., 2014).

The Schwarzschild–AdS$N$6 example makes the mechanism concrete. For symmetric probe D3-branes at constant radius, the Euclidean action is monotonically decreasing toward the tip of the cigar. The minimum is therefore obtained by shrinking rather than by stabilizing at finite radius. This illustrates the same resistance to “cheap” enclosure of bulk volume that the general inequality expresses (Ferrari et al., 2014).

7. Hyperinvariant holographic codes and geometry-aware local non-compressibility

Tensor-network models provide a code-theoretic version of holographic incompressibility. Hyperinvariant tensor networks and hyperinvariant codes are built on regular hyperbolic tilings or honeycombs using vertex tensors $N$7 and edge tensors $N$8. The standard requirements are that $N$9 and $N\le 5$0 satisfy single- and multi-tensor isometry constraints, that $N\le 5$1 is invariant under the rotational symmetry group of its vertex figure, and that

$N\le 5$2

Hyperinvariance is defined operationally as the necessity of enforcing multi-tensor isometries in order to ensure a well-defined encoding in a holographic tensor network (Cheng, 6 Jun 2025).

The main new criterion is angular $N\le 5$3-uniformity. After arranging the physical indices of a tensor according to the vertex figure of a $N\le 5$4-dimensional polytope, one says that $N\le 5$5 is angular $N\le 5$6-uniform if for every strongly angularly connected subset $N\le 5$7 of physical indices with $N\le 5$8, the map

$N\le 5$9

is an isometry, and no larger subset $N$00 with $N$01 satisfies the isometry condition. A further extension, multi-angular $N$02-uniformity, allows disjoint unions of angularly disconnected input sectors (Cheng, 6 Jun 2025).

This formalism identifies a tension between local correctability and nontrivial holographic correlation structure. A stated no-go theorem says that for a holographic tensor network or code defined on a non-simplicial regular hyperbolic tiling, with all vertex tensors rotationally invariant, if the tensors are maximally angular $N$03-uniform, meaning $N$04 equals the full size of a $N$05-dimensional facet of the vertex figure, then the network is not hyperinvariant and the boundary correlation functions are not universally non-trivial. In the author’s formulation, when $N$06, the network suppresses bulk correlation (Cheng, 6 Jun 2025).

The implication for holographic incompressibility is local and geometric rather than Shannon-theoretic. If the tensor is too uniform, too many local reconstructions become exact, and the model loses the long-range, geometry-sensitive propagation characteristic of holography. If the tensor is too restrictive, holographic encoding fails. Angular $N$07-uniformity is proposed as the intermediate regime in which local recovery is allowed only for specific angular sectors determined by the vertex figure (Cheng, 6 Jun 2025).

The paper also emphasizes residual regions and approximate complementary recovery. In the $N$08 honeycomb, the residual region depends on angular $N$09-uniformity and on the boundary bipartition. For $N$10, certain boundary conditions yield exact complementary recovery, while bipartite boundary conditions leave a 1D residual area; for $N$11, there is no exact complementary recovery and the residual region is exactly the $N$12 tiling. This suggests that some bulk information cannot be cleanly compressed into either complementary boundary region alone. At the same time, multi-angular $N$13-uniformity permits uberholographic recovery from disconnected regions, indicating that recoverability is constrained more by geometry than by region size alone (Cheng, 6 Jun 2025).

Across these formulations, holographic incompressibility refers to the same general pattern: holographic data are delocalized, redundant, and geometrically organized, but only in highly specific ways. The notion does not currently amount to a single universal lower bound on boundary support, nor to a full classification of holographically realizable entropy patterns. What is established is a set of precise rigidity statements: holography excludes some quantum extreme rays; its nontrivial extreme structure is Bell-pair-free; saturation of entropy inequalities can eliminate candidate coding regions; finite bulk regions are not captured by naïve boundary coarse-graining; stable Euclidean holography obeys an area–volume bound; and hyperinvariant codes require geometry-aware limits on local isometric flexibility. In that cumulative sense, holography is incompressible: bulk and boundary information cannot be rearranged into arbitrary extremal, local, or overly economical patterns without violating the geometric and entropic structure that defines the holographic regime.