Holographic Entanglement-Wedge Reconstruction

Holographic entanglement-wedge reconstruction is the geometric realization of an operator-algebra recovery theorem. Quantum extremal surfaces assign a bulk wedge to a boundary region; the JLMS relation matches relative entropy across that assignment; equality in data processing then yields boundary representatives of the wedge algebra. Every step is restricted to a declared semiclassical code subspace and perturbative accuracy.

Required background. Quantum Information supplies relative entropy, recovery channels, and operator-algebra codes, while Gauge Theory and String-Theory Preparation supplies the holographic encoding map and large-N expansion. Helpful background. Quantum Gravity supplies semiclassical constraints, and Black-Hole Information supplies the general reconstruction problem and its information-theoretic theorem.

Boundary region and holographic wedge algebra

Fix a finite-dimensional regulator and an isometry

$$V:{\cal H}_{\rm code}\longrightarrow {\cal H}_A\otimes{\cal H}_{\bar A}$$

that embeds a controlled family of semiclassical bulk states into the boundary theory. Select one boundary region $A$ and a state domain for which the QES and its entanglement wedge $a$ are fixed, or decompose the domain into center blocks where that statement is true. Let ${\cal M}_a$ be the algebra of relationally dressed bulk operators assigned to $a$. The area operator and other gravitational charges may lie in its center; pretending that the algebra is always a tensor factor loses this data.

For code states $\rho$ and $\sigma$, let $\rho_A$ and $\sigma_A$ be their boundary reductions and let $\rho_a$, $\sigma_a$ denote their restrictions to ${\cal M}_a$. Relative entropy is

$$D(\rho\Vert\sigma) =\operatorname{Tr}\rho(\log\rho-\log\sigma),$$

when $\operatorname{supp}\rho\subseteq\operatorname{supp}\sigma$; it is $+\infty$ otherwise. Logarithms and recovery inverses are taken on their supports. The general recovery theorem belongs to Entanglement-Wedge Reconstruction; the present page explains why holographic geometry supplies its hypotheses.

Statement and hypotheses

Statement (holographic reconstruction implication). Assume a regulated code isometry, fixed boundary and bulk algebras, a specified state family, consistent treatment of the area or charge center, and equality

$$D(\rho_A\Vert\sigma_A)=D(\rho_a\Vert\sigma_a)$$

for every pair of density matrices on the declared code, including mixtures and superpositions of its basis states. Equality on an arbitrary smaller collection of prepared states is insufficient for the following full-code conclusion. Then the boundary channel to $A$ is sufficient for ${\cal M}_a$: every $O_a\in{\cal M}_a$ has a boundary representative $O_A$ satisfying

$$O_A V=V O_a$$

on ${\cal H}_{\rm code}$. For approximate recovery one must specify the theorem and its error norm: statewise trace-norm, correlator, and channel diamond-norm errors are different. A channel-norm claim needs the corresponding control with an auxiliary reference system, or an explicit dimension-dependent conversion. A small scalar relative-entropy mismatch is not itself a numerical decoder-error bound. This statement does not apply to arbitrary states in the full CFT Hilbert space.

The hypotheses include more than “the point lies inside a drawn wedge.” They fix the encoding, bulk and boundary algebras, state family, regulator, boundary region, gravitational order, dressing, and error notion. If the QES changes across the family, its center or block structure must be retained. If errors are only known for a few correlators rather than uniformly, no channel-norm reconstruction bound follows.

Holographic proof architecture

The argument has four stages.

1. Modular-energy identity. Within a perturbative code subspace, the JLMS relation takes the schematic form

$$P K_A^{\rm CFT}P =P\left(\frac{\widehat{\operatorname{Area}}(X_A)}{4G_N} +K_a^{\rm bulk}\right)P+O(1/N),$$

where $P$ projects onto the code and each modular Hamiltonian is defined relative to a reference state. Jafferis, Lewkowycz, Maldacena, and Suh derive the bulk/boundary relative-entropy relation in JLMS 2016, §§ 2–4.

2. Entropy subtraction. Relative entropy equals modular-energy change minus entropy change. Combining JLMS with the generalized-entropy formula cancels the state-dependent area-center contribution consistently and gives equality of boundary and wedge relative entropy to the retained order. Matching only one entropy would be insufficient: reconstruction requires preservation of distinguishability for the state family.

3. Sufficiency. Data processing says a quantum channel cannot increase relative entropy. Equality for every relevant pair is its saturation condition. In each fixed center block, a Petz-type recovery map reverses the restriction channel on the code. Dong, Harlow, and Wall give the operator-algebra argument establishing entanglement-wedge reconstruction from this equality Dong, Harlow, and Wall 2016, complete Letter, pp. 1–5.

4. Heisenberg reconstruction. Taking the recovery channel’s adjoint maps a wedge operator to $O_A$. The intertwining relation above ensures equal matrix elements for all code states. If another boundary region $B$ also contains the operator in its wedge, it can have a different representative $O_B$. The two need agree only on the code; their distinction outside it is how overlapping reconstruction avoids a full-Hilbert-space cloning paradox.

A finite-code check

The three-qutrit erasure code makes the geometry-independent mechanism visible. With arithmetic modulo three, define

$$\begin{aligned} |\widetilde0\rangle&=(|000\rangle+|111\rangle+|222\rangle)/\sqrt3,\\ |\widetilde1\rangle&=(|012\rangle+|120\rangle+|201\rangle)/\sqrt3,\\ |\widetilde2\rangle&=(|021\rangle+|102\rangle+|210\rangle)/\sqrt3. \end{aligned}$$

Tracing out any one qutrit leaves enough information in the other two to reconstruct the logical qutrit; the erased qutrit’s reduced state is independent of the logical input. Thus the same logical algebra has representatives on overlapping pairs. Tensor-network models use this erasure structure to illustrate how boundary region size controls a discrete wedge Almheiri, Dong, and Harlow 2015, §§ 2–4. The code is a check of the recovery logic, not a derivation of continuum AdS/CFT or the QES formula.

Failure modes and scope

Enlarged-code counterexample. Suppose two code states differ in the wedge algebra but have identical reductions on $A$. Their boundary relative entropy vanishes while their wedge distinguishability is nonzero. Equality fails, so no $A$-supported operator can distinguish them. Enlarging a code across an erasure threshold or across a QES transition can create precisely this obstruction.

Nonuniform perturbative error. An $O(1/N)$ error for each fixed simple operator need not stay small for an exponentially large code or for complexity growing with $N$. Existence also does not imply an efficient decoder.

Dressing and state domain. Gravitational operators need relational dressing, which affects algebra assignment. A state-dependent wedge can be handled blockwise only when the center and transition are controlled. The theorem does not license one state-independent interior algebra on the entire Hilbert space and does not by itself settle black-hole interior reconstruction; that frontier is treated on Black-Hole Interior Reconstruction.

Exercises

Show why one erased qutrit carries no logical information. For the three codewords above, trace out the last two qutrits and evaluate both diagonal and off-diagonal logical matrix elements on the first qutrit.

Solution

Each codeword contains the first-qutrit labels zero, one, and two exactly once, paired with three mutually orthogonal states of the other two qutrits. Its first-qutrit reduction is therefore the maximally mixed state $I_3/3$. For two distinct logical codewords, every pair attached to a given first label is orthogonal, so the reduced off-diagonal operator vanishes. The erased qutrit consequently has no dependence on any logical density matrix. By the erasure-correction condition, the complementary pair contains a complete reconstruction, while one qutrit alone cannot represent the full logical algebra.

References

• Almheiri, Ahmed, Xi Dong, and Daniel Harlow. “Bulk Locality and Quantum Error Correction in AdS/CFT.” Journal of High Energy Physics 04 (2015): 163. DOI. Open PDF.
• Dong, Xi, Daniel Harlow, and Aron C. Wall. “Reconstruction of Bulk Operators within the Entanglement Wedge in Gauge-Gravity Duality.” Physical Review Letters 117 (2016): 021601. DOI. Open PDF.
• Jafferis, Daniel L., Aitor Lewkowycz, Juan Maldacena, and S. Josephine Suh. “Relative Entropy Equals Bulk Relative Entropy.” Journal of High Energy Physics 06 (2016): 004. DOI. Open PDF.