RUP-25-23 Baby universe as logical qubits: information recovery in random encoding
Takato Mori, Beni Yoshida · 2025
Read RUP-25-23 Baby universe as logical qubits: information recovery in random encoding by Takato Mori, Beni Yoshida (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-20747.txt.
Introduction
Takato Mori and Beni Yoshida revisit in 2025 a puzzle about baby universes inside holographic models of gravity. A baby universe is a small closed spacetime budding off from a larger one. Path-integral calculations hint its state space might be one-dimensional, leaving no room for entropy or inner life. Mori and Yoshida argue it can carry large entropy instead, using random encoding as their lens.
In their picture, the baby universe holds logical information that neither boundary region can read alone. Haar random encoding scrambles the data so thoroughly that reconstruction from either half is impossible. Two puzzles then dissolve: its states cannot be cloned, and no single observer can witness their fate at the singularity. The heavy operator that creates the universe also serves as its observer, making each microstate observer-dependent.
Catalog record
| CREATOR | Takato Mori, Beni Yoshida |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 60 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-20747
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
RUP-25-23 Baby universe as logical qubits: information recovery in random encoding Takato Mori, Beni Yoshida Abstract We revisit whether a semiclassical baby universe in AdS/CFT necessarily possess a trivial one-dimensional Hilbert space or may instead carry a large entropy. Recent results on Haar random encoding suggest a breakdown of complementary recovery, in which no logical operators can be reconstructed from individual bipartite subsystems. Motivated by this, we propose an interpretation where a baby universe emerges as logical degrees of freedom that cannot be accessed from either boundary alone, assuming pseudorandom dynamics in holographic CFTs. We then analyze two conceptual puzzles: an apparent cloning of baby-universe microstates and its eventual fate at the singularity. Both puzzles are avoided because no single boundary observer can access the baby-universe degrees of freedom, be it classical or quantum, reflecting an emergent form of complementarity due to the structure of random encoding. In this interpretation, observers arise naturally: the same heavy operator that prepares the baby-universe geometry also serves as observer-like degrees of freedom that define an observer-dependent baby-universe microstate. 1 Introduction Does a closed baby universe possess a trivial one-dimensional Hilbert space with nothing to explore, or can it accommodate a large entropy and host rich internal physics? Recent developments in gravitational path integrals suggest the existence of semiclassical gravitational saddles describing joint states of two asymptotically AdS regions entangled with a semiclassical baby universe [ 1 ] . However, when non-perturbative wormhole effects are included, the same path integrals indicate an extremely small, possibly one-dimensional, Hilbert space associated with the baby universe [ 2 ] . While a semiclassical baby universe might in principle be captured by analyzing the boundary CFTs beyond the naive large- N limit, it remains obscure how, or even where, the baby-universe Hilbert space is encoded within the boundary Hilbert spaces. Quantum error correction provides a framework to analyze how quantum information can be encoded non-locally in many-body entangled systems. This perspective has proven invaluable in quantum gravity, where information is scrambled and delocalized by chaotic dynamics [ 3 , 4 ] . Although a complete resolution of the baby-universe puzzle ultimately requires a full theory of quantum gravity, the language of quantum error correction offers a precise yet intuitive framework to clarify underlying assumptions and illuminate the physical mechanisms behind it. In this paper, we construct a simple toy model of quantum error-correcting codes that capture essential features of how a baby universe may emerge in two-sided CFTs. A key property of the semiclassical baby universe from [ 1 ] is that it resides outside the entanglement wedges of both CFTs on the boundary. At first sight, this seems to support the idea that the baby universe is absent, or that its Hilbert space is trivial. However, it also leaves open the possibility that the baby universe may be encoded as logical qubits in a quantum error-correcting code where its codeword states remain inaccessible to either boundary individually, but recoverable through non-local joint operations involving both boundaries. Concretely, we consider a Haar random encoding with an isometry \displaystyle V:C\rightarrow AB (1) where A and B denote the two boundary CFTs, and C represents bulk degrees of freedom to be reconstructed from them. This random quantum error-correcting code exhibits the characteristic behavior expected of a baby universe. When the Hilbert space dimensions satisfy \displaystyle d_{A}<d_{B}d_{C},\qquad d_{B}<d_{A}d_{C}, (2) no quantum information about C can be retrieved from either A or B alone, as neither subsystem supports any non-trivial unitary logical operators of the code. This property follows from recent results on the absence of bipartite entanglement in typical tripartite Haar random states [ 5 ] . From the viewpoint of each boundary, the Hilbert space C appears absent, yet it still exists as logical qubits accessible only through joint, non-local operations acting on A and B together. This property of a random encoding has a somewhat surprising implication: the breakdown of complementary recovery . Roughly speaking, complementary recovery refers to a special class of quantum error correction in which the input Hilbert space C can be bipartitioned as C\simeq C_{A}\otimes C_{B} such that C_{A} and C_{B} are reconstructable from A and B respectively [ 6 ] . Complementary recovery holds in most known examples of AdS/CFT and also in stabilizer codes when formulated through operator-algebraic frameworks. In contrast, a Haar-random encoding exhibits a complete breakdown of complementary recovery where no logical operator can be reconstructed on any bipartite subsystem. 1 1 1 The HaPPY code is constructed with stabilizer tensors, and thus satisfies complementary recovery [ 7 ] . On the contrary, a random tensor network code [ 8 ] is constructed with Haar random tensors, and thus violate complementary recovery in certain cases [ 9 ] . Equipped with this toy model, we revisit the semiclassical baby universe setup by Antonini, Sasieta, and Swingle (AS 2 ) to examine how a pseudorandom encoding may arise [ 1 ] . In the AS 2 construction, a shell of heavy operators O is inserted and evolved in Euclidean time by e^{-\frac{\beta H}{2}} . Here we generalize to a family of randomly-chosen shells of heavy operators \mathcal{M}_{\mathrm{heavy}}=\{O^{(i)}\} , and study the corresponding ensemble of boundary states. Each insertion constructs a two-sided CFT wavefunction expanded in the low-energy basis below the Hawking-Page temperature as \displaystyle|\psi_{AB}^{(i)}\rangle=\frac{1}{\sqrt{Z^{(i)}}}\sum_{m,n}\tilde{O}^{(i)}_{mn}|m\rangle_{A}\otimes|n\rangle_{B} (3) where \tilde{O}^{(i)}_{mn} are matrix elements of the Euclidean-evolved heavy operator. We suggest that this tensor \tilde{O}^{(i)}_{m,n} exhibits pseudorandom features akin to those of a tripartite Haar random state. This expectation rests on an assumption that holographic CFT correlation functions involving mixtures of heavy and light operators display pseudorandom statistics due to underlying chaotic dynamics of holographic CFTs [ 10 , 11 , 12 ] . To make this structure explicit, we introduce a fictitious third boundary system C entangled with the label i of the heavy operator: \displaystyle|\Psi_{ABC}\rangle=\frac{1}{\sqrt{Z}}\sum_{i,m,n}\tilde{O}^{(i)}_{mn}|m\rangle_{A}\otimes|n\rangle_{B}\otimes|i\rangle_{C}. (4) This construction suggests that the baby-universe Hilbert space can be regarded as the code subspace of a random isometry, C\rightarrow AB . Namely, projecting onto a particular state of C prepares a codeword state on AB which may be regarded as a specific baby-universe microstate. If one’s access is limited to local operators on A or B , no information about C can be recovered. Then the boundary observer perceives a mixed state, \displaystyle\rho_{AB}=\sum_{i}|\psi_{AB}^{(i)}\rangle\langle\psi_{AB}^{(i)}| (5) where the semiclassical baby universe may emerge as an effectively isolated region. A mixed state interpretation of a baby universe has also appeared in a recent work [ 13 , 14 ] . If a semiclassical baby universe possesses a large internal Hilbert-space dimension, it appears that its microstates are cloned. Namely, one copy lives as degrees of freedom within the baby universe itself, while the other is encoded in the entanglement patterns between two asymptotic AdS regions. This apparent cloning puzzle can be resolved by appealing to an emergent form of complementarity, in which the baby-universe observer and the boundary observer have no operational means to compare or verify the cloning. The essential distinction we emphasize here is that, in our picture, complementarity is not assumed as a fundamental principle but rather arises dynamically from the structure of random encoding. Any communication between a single boundary observer and the baby-universe observer is