Baby-Universe Hilbert Space in Quantum Gravity

• Baby-Universe Hilbert Space is a quantum gravity construct that encodes states of closed universes through topology-changing, abelian operator algebras.
• It demonstrates a range of dimensional behaviors, including a one-state property in high dimensions and richer structures in lower-dimensional or holographic models.
• Models like DSSYK, JT gravity, and matrix integrals highlight its pivotal role in addressing factorization, ensemble averaging, and holographic duality in quantum gravity.

The baby-universe Hilbert space is a structural entity in quantum gravity that encodes the quantum states associated with closed universes and topological fluctuations, such as those produced by Euclidean wormholes. Its precise definition and physical interpretation are highly theory-dependent, particularly with respect to spacetime dimension, holographic dual descriptions, and the role of gauge redundancies, ensemble averaging, and superselection sectors. The construction of the baby-universe Hilbert space is central to debates on quantum information in gravity, the swampland conjectures, and holographic duality.

1. Definition and Operator Structure

The baby-universe Hilbert space $\mathcal{H}_{\rm BU}$ is the vector space (often equipped with an inner product from the gravitational path integral) built to capture quantum states arising from disconnected, spatially closed universes in gravitational path integrals. In many constructions, it is generated as the Fock space resulting from creation operators that add closed topologies—so-called "baby-universe creation operators"—acting on an appropriate vacuum, commonly the Hartle–Hawking (no-boundary) state (McNamara et al., 2020, Okuyama, 2024).

In concrete models, such as the double-scaled SYK (DSSYK) matrix model, the baby-universe sector is populated by a family of commuting creation operators $\{\mathcal{B}_a\}_{a\in\mathbb{N}}$: $$\mathcal{B}_a = \int_0^\pi \frac{d\theta}{2\pi} 2\cos(a\theta)|\theta\rangle\langle\theta|$$ These act on the "chord Hilbert space," and their commutative algebra is explicitly abelian. All bulk operators (e.g., disk, trumpet, double-trumpet amplitudes) admit nontrivial expansions in the $\mathcal{B}_a$ basis: $$1 = \sum_{r\in\mathbb{Z}} (-1)^r q^{r(r+1)/2} \mathcal{B}_{2r}$$ In this context, each $\mathcal{B}_a$ creates a "baby universe" of size $a$, and the operator algebra reflects the combinatorics of topological transitions in the model (Okuyama, 2024). The abelian property is a universal feature of baby-universe operator algebras in gravitational settings, in contrast to the non-abelian field operator algebra of conventional QFT (Casali et al., 2021, Gesteau et al., 2020).

2. Dimensionality and the One-State Conjecture

A central theme is the dimensionality of $\mathcal{H}_{\rm BU}$, which is deeply tied to swampland constraints, global symmetries, and holography. In theories of quantum gravity with $d>3$, it is conjectured—under strong swampland and holographic constraints—that $\mathcal{H}_{\rm BU}$ is one-dimensional (McNamara et al., 2020): $\dim \mathcal{H}_{\rm baby} = 1$ This "one-state property" emerges because all baby-universe creation/annihilation operators act trivially on physical states due to large gauge redundancies between topologically distinct universes, enforced via a quotient by null gauge-equivalent combinations in the path-integral inner product. The only gauge-invariant state is the Hartle–Hawking wavefunction. This result forbids nontrivial, strictly local bulk operators, reducing every compactly supported bulk observable to a multiple of the identity and removing any independent information content from the closed-universe sector (McNamara et al., 2020, Zhao, 5 Feb 2026).

In contrast, in $d\leq 3$ and especially in models such as JT gravity or double-scaled SYK, the Hilbert space may be infinite-dimensional or possess a graded structure labeled by discrete or continuous parameters (e.g., representations of a gauge group, $\alpha$-parameters, or quantum numbers arising from boundary insertions) (Gardiner et al., 2020, Gardiner et al., 2020, Betzios et al., 2022, Okuyama, 2024). In such cases, summing over wormhole configurations leads to nontrivial superselection sectors.

3. Construction via Path Integral, GNS, and Operator Algebras

The baby-universe Hilbert space is rigorously constructed using operator-algebraic methods. Starting from the algebra of boundary observables $\mathcal{A}$, the full quantum gravity algebra is enlarged to $$\mathcal{A}_{\rm QG} = \mathcal{A}_{\rm res} \otimes \mathcal{A}_{\rm baby}$$, where $\mathcal{A}_{\rm baby}$ is abelian and encodes topology change (Gesteau et al., 2020). The path integral defines a positive linear functional (state) $\omega$ on $\mathcal{A}_{\rm QG}$, and the Gelfand–Naimark–Segal (GNS) construction yields the Hilbert space and cyclic representation: $\langle A|B\rangle = \omega(A^\dagger B)$ Null states—linear combinations with vanishing norm—are quotiented out, ensuring the elimination of negative-norm sectors (Marolf et al., 2020, Gardiner et al., 2020).

Commutativity of $\mathcal{A}_{\rm baby}$ ensures the operators are simultaneously diagonalizable; the superselection sectors (often called "alpha-sectors") correspond to the spectrum of the abelian algebra and can be described as probability measures or characters (multiplicative linear functionals) on this algebra (Gesteau et al., 2020).

Miraculous cancellations invoked in the gravitational path integral, necessary for the restoration of factorization and unitarity, are interpreted mathematically as results of the path integral state being a pure character on $\mathcal{A}_{\rm baby}$ (Gesteau et al., 2020).

4. Models, Topological Examples, and Matrix Integrals

Concrete realizations of the baby-universe Hilbert space are provided by models such as:

• Double-Scaled SYK (DSSYK): The Hilbert space is built from the chord Fock space, with the $\mathcal{B}_a$ operators generating the full operator algebra, and all amplitudes (disk, trumpet, TFD) expand in the $\mathcal{B}_a$ basis using $q$-Hermite polynomials and combinatorial identities (Okuyama, 2024).
• 2d Topological Quantum Field Theory and Dijkgraaf–Witten Gravity: The baby-universe Hilbert space decomposes into superselection sectors labeled by irreducible representations, and within each sector the number of universes obeys an independent Poisson distribution. Positive-norm projectors and defect mechanisms are invoked to ensure unitarity and restore local descriptions (Gardiner et al., 2020).
• Matrix Integral Models: In JT gravity or related matrix models, geodesic-boundary insertions correspond to operator insertions (such as $\operatorname{Tr}\cos(a\sqrt{H})$), and summing over topologies realizes the nonperturbative completion of amplitudes. The amplitude for a basis state $|\ell,a\rangle$ includes both discrete spectral data and operator insertions, with inner product structure inherited from the matrix eigenvalue distribution (Zolfi, 29 Jan 2026).

These models admit explicit calculations for bulk amplitudes, phase transitions (e.g., black-hole to white-hole tunneling), and the distinction between smooth and firewall geometries.

5. Superselection, Ensemble Averaging, and Complementarity

The concept of superselection is ubiquitous: the Hilbert space is often decomposed into sectors, each labeled by conserved quantities such as representations ($R$), $\alpha$-parameters, or eigenvalues of certain insertion operators. In many toy models and in 2d gravity, the baby-universe sector can carry substantial entropy and exhibit ensemble-averaged behavior (Gardiner et al., 2020, Mori et al., 25 Nov 2025). Notably, in certain AdS/CFT constructions, the baby-universe Hilbert space appears as a logical subspace (code subspace) that is inaccessible from any single boundary factor, exhibiting a breakdown of complementary recovery and allowing the baby-universe sector to carry entropy of order $O(1/G_N)$ (Mori et al., 25 Nov 2025).

However, in $d>3$, such ensemble averaging is forbidden by quantum gravity constraints, and all baby-universe sectors collapse to a single state (McNamara et al., 2020, Gesteau et al., 2020).

6. Exceptional Cases and the Bulk/Baby-Universe Distinction

A recurring theme is the sharp distinction between the baby-universe Hilbert space—encoding the ensemble of closed universes and their superselection data—and the bulk Hilbert space of local observers. In models such as de Sitter gravity, the "one-state" property applies strictly to the baby-universe Hilbert space: $\mathcal{H}_{\rm BU}$ only encodes the statistics of an ensemble of classical universes with commuting observables and does not possess the structure of a true quantum Hilbert space. Genuine quantum mechanics for internal observers is constructed via the GNS procedure from patch (region) operator data, leading to a finite-dimensional bulk Hilbert space, e.g., of dimension $Z_L$ in a toy 1D de Sitter model (Zhao, 5 Feb 2026). All genuine quantum structure (noncommuting observables, quantum amplitudes) emerges from the statistical data of $\mathcal{H}_{\rm BU}$.

Table: Summary of Baby-Universe Hilbert Space Structure in Key Models

Model/Class	Operator Algebra	Dimensionality
$d>3$ gravity (Swampland)	Abelian; baby-universe ops trivial	$1$ (Hartle–Hawking state)
DSSYK / JT gravity	Abelian; $\mathcal{B}_a$ nontrivial	infinite-dimensional
2d TQFT / Dijkgraaf–Witten	Abelian; sector-labeled insertions	Direct sum of Poisson-sectors
Matrix-integral models	Abelian; operator insertions, spectra	Fock-space or code subspace
AdS/CFT random encoding (Mori et al., 25 Nov 2025)	Abelian (code subspace inaccessible to single boundary)	$e^{S_{\rm bu}}$ (logical qubits)

7. Physical Implications and Open Problems

• Gauss's Law for Entropy: The one-dimensionality of $\mathcal{H}_{\rm BU}$ in $d>3$ enforces a global entropy constraint—total entropy is carried by boundaries, and a truly closed universe has zero entropy (McNamara et al., 2020).
• Breakdown in Low Dimensions: In $d=2,3$, and in topological and holographic toy models, $\mathcal{H}_{\rm BU}$ can be macroscopic, with boundary theory duals exhibiting ensemble averaging and nontrivial factorization structure (Gardiner et al., 2020, Okuyama, 2024, Liu, 17 Sep 2025).
• Emergent Quantum Structure: Quantum mechanics for bulk observers arises from classical statistics in $\mathcal{H}_{\rm BU}$; attempts to ascribe quantum structure directly to the baby-universe space conflate classical and quantum data (Zhao, 5 Feb 2026).
• Factorization and Unitarity: Path integral constructions and GNS representations ensure the proper elimination of negative-norm/null states and restore consistency with unitarity and factorization (Marolf et al., 2020, Gardiner et al., 2020, Gesteau et al., 2020).
• Holographic Encoding: In AdS/CFT, encoding of closed universes in the boundary CFT depends critically on bulk entanglement; with sufficient entanglement, the CFT encodes a large Hilbert space, otherwise the encoding reflects only a single state (Antonini et al., 14 Jul 2025).

The structure, physical content, and relevance of the baby-universe Hilbert space remain central to the foundations of quantum gravity, holography, and the swampland program, with ongoing research elaborating its role in black hole physics, information recovery, and topology change.