Quartz 4

Holographic Quantum Error Correction — When Physics Discovered Coding Theory

7 min read

Holographic Quantum Error Correction — When Physics Discovered Coding Theory

One of the most surprising connections in modern science: the mathematics of how spacetime encodes information about its bulk interior is identical to the mathematics of quantum error-correcting codes. The universe appears to protect bulk information from boundary erasure using the same algorithms that quantum computers use to protect qubits from decoherence.

Confidence: established (mathematical equivalence between holographic RT formula and quantum error correction); established (HaPPY code and tensor network models); emerging (full QEC structure of real AdS/CFT); theoretical (implications for real quantum gravity)

The Surprising Discovery

Ahmed Almheiri, Xi Dong, and Daniel Harlow (2015, “Bulk Locality and Quantum Error Correction in AdS/CFT”) noticed something strange about how bulk operators in AdS/CFT can be reconstructed from the boundary.

In AdS/CFT, a bulk operator (a field value at a point in the interior) can be represented by boundary operators. But here’s the strange part: the same bulk operator can be represented by different boundary operators depending on which boundary region you use. A bulk operator in the center of AdS can be written as:

  • An operator on the left boundary
  • An operator on the right boundary
  • An operator on either the top or bottom of the boundary

This redundancy is exactly what error-correcting codes do. A single logical qubit is encoded across multiple physical qubits — you can reconstruct the logical information from any sufficiently large subset.

The realization: The bulk AdS space is the “logical” level (the message). The boundary CFT is the “physical” level (the encoded codeword). Spacetime is a code.

The Rindler-Wedge Reconstruction

The precise statement: a bulk operator in the Rindler wedge associated with boundary region A can be reconstructed only from operators in that boundary region. You don’t need the rest of the boundary.

This property — called subregion duality or Rindler-wedge reconstruction — is exactly what error correction does: any sufficiently large “backup copy” (boundary region) contains enough information to reconstruct the logical data (bulk operator). Any small piece of the boundary is insufficient — the message is spread redundantly across the whole boundary.

The HaPPY Code (2015)

Fernando Pastawski, Beni Yoshida, Daniel Harlow, and John Preskill (“HaPPY”) constructed the first explicit holographic code — a tensor network that simultaneously:

  1. Satisfies the Ryu-Takayanagi formula exactly (entropy of boundary region A = area of minimal surface in bulk)
  2. Is a quantum error-correcting code (bulk information recoverable from any sufficiently large boundary piece)
  3. Has bulk operators that can be reconstructed from multiple boundary regions

The construction uses perfect tensors — tensors where any partition of the tensor’s indices into two equal halves gives a maximally entangled state. The tensor network tiles the hyperbolic plane (AdS-like geometry) with pentagons, hexagons, or other tilings.

What HaPPY proved: The RT formula and error correction are not separate features of AdS/CFT. They are the same feature, described in two different languages.

The Code Structure

  • Logical qubits (bulk): the information about interior spacetime
  • Physical qubits (boundary): the CFT degrees of freedom
  • Encoding map: the holographic tensor network
  • Code distance: determined by the minimal RT surface — erasure of a boundary region smaller than the RT surface cannot destroy the bulk information

The Approximate QEC Connection (2019-2020)

The HaPPY code is exact but uses perfect tensors (an idealization). Real AdS/CFT is better described by approximate quantum error correction — the encoding is not perfect but works up to exponentially small errors in 1/G_N (the Planck constant).

Harlow (2016) proved that exact quantum error correction is essentially equivalent to the RT formula holding exactly — and that approximate QEC is equivalent to RT holding with quantum corrections. This connects:

  • Exact RT ↔ Exact QEC
  • Quantum-corrected RT (FLM formula) ↔ Approximate QEC

The island formula (which resolves the black hole information paradox) has a natural interpretation in this language: at late times in black hole evaporation, the “code subspace” shifts — the interior spacetime region that can be reconstructed from the radiation changes. The island is the region that “joins” the code.

Hyperinvariant Tensor Networks (2023-2025)

A new class of holographic codes — hyperinvariant tensor networks — extends HaPPY into quantum codes that reproduce expected CFT boundary correlation functions (which the original HaPPY code got wrong). These are a recent (2023-2025) advance that brings the toy model closer to real AdS/CFT:

  • Correct two-point functions in the boundary CFT
  • Satisfy RT formula
  • Full QEC structure
  • Can be extended to “quantum LEGO” formalism (LEGO_HQEC, 2024)

LEGO_HQEC (October 2024): A software package that automates construction and analysis of holographic codes on regular hyperbolic tilings using the quantum LEGO formalism. Demonstrated universal fault-tolerant logic with heterogeneous holographic codes in 2025.

What Holographic Codes Tell Us About Quantum Gravity

Locality Is Not Fundamental

In standard physics, local operators in the bulk (gravity side) are local — they affect only nearby regions. But in the boundary description, a “local” bulk operator corresponds to a highly nonlocal boundary operator. The boundary has no notion of bulk locality. Locality in spacetime is an emergent property, not a primitive one. The code structure creates the appearance of locality.

The Horizon as a Code Boundary

A black hole horizon is a special kind of RT surface — the minimal surface that separates interior and exterior. The error-correction interpretation: information about the black hole interior is encoded in the exterior CFT degrees of freedom, but it’s protected — you can’t reconstruct the interior from a small piece of the exterior. You need access to more than half the boundary to “decode” the black hole interior. This is why Hawking radiation carries information only collectively, never locally — individual Hawking quanta carry no information about the interior, but the full ensemble does.

Quantum Error Correction in the Real Universe?

A profound question: if AdS/CFT is a QEC code, does this mean our universe (which is approximately de Sitter, not AdS) is also a QEC code? Some researchers (Almheiri, Harlow, others) have speculated that the cosmic horizon — the boundary of our observable universe — is encoding all the interior information redundantly, as a code. We would be the “logical” data; the cosmic horizon would be the “physical” codeword.

Cross-Connection: Real Quantum Computing (2024-2025)

Holographic codes are now being actively studied for practical quantum error correction, not just as theoretical curiosities:

  • Holographic codes have tunable encoding rates and distance scaling competitive with other well-studied codes
  • Excellent recovery thresholds for certain noise models
  • The quantum LEGO formalism enables automated construction and optimization
  • 2025 saw demonstration of “universal fault-tolerant logic with heterogeneous holographic codes”

The traffic flows both ways: quantum error correction theory has been illuminated by holography, and holographic physics is now informing real quantum computer designs.

The Deeper Pattern

Holographic QEC reveals something profound: nature is information-theoretically optimal. The way spacetime encodes information about its interior achieves the maximum possible redundancy without redundant storage — exactly what optimal error-correcting codes do. This suggests that information-theoretic constraints might be more fundamental than spacetime geometry — that the laws of physics are, at bottom, laws of information.

This connects to Claude Shannon’s 1948 insight that information has physical substrate (Shannon entropy), John Wheeler’s “It from Bit” (1990), and the holographic entropy bound: the universe stores no more information than the laws of quantum error correction allow.

Key Facts

  • Almheiri-Dong-Harlow (2015): first identification of AdS/CFT as QEC code
  • HaPPY code: holographic quantum error-correcting code on hyperbolic tensor network tiling
  • Perfect tensor: any bipartition of indices gives maximally entangled state — the building block of holographic codes
  • RT formula = QEC property: these are mathematically equivalent statements about the same structure
  • Code distance of holographic code: set by the area of the minimal RT surface (in Planck units)
  • Subregion duality: bulk operator in wedge W(A) ↔ reconstructable from boundary region A alone
  • Harlow 2016: proved RT equivalence to exact QEC; FLM formula equivalent to approximate QEC
  • Hyperinvariant tensor networks (2023): produce correct boundary CFT two-point functions

See Also

Graph View

Backlinks