Edward Witten (JHEP 2024) proposed the operator algebra along an observer's worldline as a background-independent algebra of quantum gravity and hypothesized that the Hartle-Hawking no-boundary state ΨHH\Psi_{\mathrm{HH}} is a universal state of maximum entropy, so that the entropy of any observer state is its relative entropy with respect to ΨHH\Psi_{\mathrm{HH}} and the completed observer algebra A^obs\widehat{\mathcal{A}}_{\mathrm{obs}} is never of von Neumann Type III. The open problem is to give this hypothesis a precise mathematical foundation in generic spacetimes: determine whether ΨHH\Psi_{\mathrm{HH}} is a normalizable state or only a normal weight of A^obs\widehat{\mathcal{A}}_{\mathrm{obs}} and which operators have finite expectation values in it; prove or refute the conjecture that the algebra is always of Type I or Type II (with ΨHH\Psi_{\mathrm{HH}} as the tracial state or weight), for which the only controlled cases so far are the de Sitter static patch (Type II1), JT gravity, and inflationary FLRW cosmologies (Type II-infinity, with no maximum-entropy state); fix the nonperturbative normalization of the universal additive constant −log⁡Z-\log Z in the entropy formula; and control the topology-changing directions in field space expected to render ΨHH\Psi_{\mathrm{HH}} non-normalizable. Later work has sharpened but not settled these questions: the tracial role of the no-boundary state was verified in JT gravity, a no-go result forbids KMS states on asymptotically de Sitter backgrounds other than pure de Sitter, and the perturbative norm of the no-boundary state was shown to vanish to all orders, complicating the definition of ZZ.

Background

In quantum gravity the gauge symmetry of general relativity - diffeomorphism invariance - removes the local observables on which ordinary quantum field theory is built: any measurable operator must be gravitationally "dressed" by reference to some physical structure. A convenient dressing is an observer, an idealized massive particle carrying a clock, modeled by a worldline with clock variable q≥0q \geq 0 and conjugate momentum. Gauge-invariant operators dressed to this worldline capture everything the observer can measure; by the timelike tube theorem of algebraic quantum field theory (the statement that the algebra of fields in the causal envelope of a timelike curve coincides with the algebra on the curve itself), a worldline algebra is a genuine substitute for a local field algebra.

The natural language for such questions is the theory of von Neumann algebras, i.e. algebras of bounded Hilbert-space operators closed in the weak operator topology. A factor is a von Neumann algebra whose center consists only of scalars, and factors are classified into types by their trace structure: a Type I factor is the algebra of all bounded operators on a Hilbert space, with pure states and ordinary density matrices; a Type II factor has no irreducible Hilbert-space representation but admits a trace, so density matrices and a renormalized entropy still exist - a Type II1 factor has a finite normalized trace and hence a maximum-entropy "tracial" state, while a Type II-infinity factor has only an unnormalizable trace; a Type III factor has no trace at all, so neither density matrices nor entropy are defined for its states. Entropies of Type II factors are fixed only up to an additive constant, which must be supplied by additional input. A standard mechanism producing Type II factors is the crossed product, which adjoins to an algebra the modular flow of a state; this is how equilibrium (KMS, for Kubo-Martin-Schwinger - the algebraic characterization of thermal equilibrium) states of quantum field theory in a thermodynamic limit, or gravity in a horizon-bounded region, yield Type II algebras.

Applying this to the de Sitter static patch (the region causally accessible to a single observer in de Sitter space, bounded by a cosmological horizon), Chandrasekaran, Longo, Penington, and Witten showed that the algebra of observer-dressed observables is a Type II1 factor with a maximum-entropy state corresponding to empty de Sitter space, whose entropy differences agree with the generalized entropy Sgen=A/4GN+SoutS_{\mathrm{gen}} = A/4G_N + S_{\mathrm{out}} (horizon area in Planck units plus the entropy of matter). Penington and Witten showed that the analogous boundary algebra in Jackiw-Teitelboim (JT) gravity - a solvable two-dimensional dilaton gravity model - is a Type II-infinity factor.

Witten (2024) then asked for a background-independent statement. Abstractly, the observer algebra Aobs\mathcal{A}_{\mathrm{obs}} is characterized as an operator product algebra (OPE) - an algebra whose products are defined by universal short-distance expansions; along the worldline, after coupling to gravity, the OPE in the observer's proper time becomes an expansion in 1/q1/q. A state of Aobs\mathcal{A}_{\mathrm{obs}} is a positive linear functional compatible with these OPE relations; any such state generates a Hilbert-space representation via the GNS construction (the standard Gelfand-Naimark-Segal construction that realizes a positive functional as a vector state), and completing Aobs\mathcal{A}_{\mathrm{obs}} in that representation yields a von Neumann algebra A^obs\widehat{\mathcal{A}}_{\mathrm{obs}} whose type can be asked about.

The paper's central speculative proposal concerns the Hartle-Hawking no-boundary state ΨHH\Psi_{\mathrm{HH}} - the state of a closed universe defined by a Euclidean gravitational path integral over geometries whose only boundary is the surface on which measurements are performed. The hypothesis is that ΨHH\Psi_{\mathrm{HH}} is a universal state of maximum entropy, i.e. a tracial state or weight for A^obs\widehat{\mathcal{A}}_{\mathrm{obs}}:

Tr a=⟨ΨHH∣ a ∣ΨHH⟩,\mathrm{Tr}\, a = \langle \Psi_{\mathrm{HH}} |\, a\, | \Psi_{\mathrm{HH}} \rangle,

so that the entropy of any observer state ρ\rho is its relative entropy (the quantum-information-theoretic divergence between two states) with respect to ΨHH\Psi_{\mathrm{HH}}:

S(ρ)=− Tr ρlog⁡ρ−log⁡Z,S(\rho) = -\,\mathrm{Tr}\,\rho \log \rho - \log Z,

where Z=⟨ΨHH∣ΨHH⟩Z = \langle \Psi_{\mathrm{HH}} | \Psi_{\mathrm{HH}} \rangle is a universal constant. A weight is the unnormalizable analogue of a state - a positive linear functional that may take the value +∞+\infty, called normal if it is compatible with limits of increasing operators; a trace is called semifinite when its finite-trace operators are rich enough to approximate every element of the algebra. If ΨHH\Psi_{\mathrm{HH}} is only a weight, the trace is defined only on operators that cancel its divergence. The proposal passes one nontrivial check - comparing de Sitter vacua with different cosmological constants gives sensible relative entropies - and underlies the conjecture that A^obs\widehat{\mathcal{A}}_{\mathrm{obs}} is always of Type I or Type II, never Type III, so that an observer's experience is always described by a density matrix. Heuristics with Schwarzschild-de Sitter spacetime (a black hole in de Sitter space) suggest, however, that entropy can grow without bound along directions in field space that point toward topology change, along which ΨHH\Psi_{\mathrm{HH}} is expected to be non-normalizable; whether a maximum-entropy vector exists in the perturbative Hilbert space, and what the nonperturbative value of ZZ is, are left open. (The paper also develops a parallel large-NN construction for anti-de Sitter space based on a deformation-quantized algebra of single-trace operators (deformation quantization being the deformation of a classical phase-space Poisson algebra into a noncommutative algebra), whose analytic continuation in NN raises analogous unresolved positivity questions; those directions are neighboring open problems, not part of the stated objective.)

Problem Statement

For a generic spacetime in semiclassical quantum gravity - closed universes, and spacetimes whose causally accessible region is bounded by horizons - settle the mathematical status of the Hartle-Hawking no-boundary state and of the observer algebra, as posed by Witten (2024):

(1) Status of ΨHH\Psi_{\mathrm{HH}}: in the GNS Hilbert space associated with a state of Aobs\mathcal{A}_{\mathrm{obs}} on a given spacetime, is ΨHH\Psi_{\mathrm{HH}} a normalizable vector state, or only a normal weight of the completed algebra A^obs\widehat{\mathcal{A}}_{\mathrm{obs}}? If it is only a weight, characterize precisely the operators in A^obs\widehat{\mathcal{A}}_{\mathrm{obs}} with finite ΨHH\Psi_{\mathrm{HH}}-expectation values, and determine whether they carry a normal semifinite trace extending Tr a=⟨ΨHH∣a∣ΨHH⟩\mathrm{Tr}\,a = \langle\Psi_{\mathrm{HH}}|a|\Psi_{\mathrm{HH}}\rangle.

(2) Type of A^obs\widehat{\mathcal{A}}_{\mathrm{obs}}: the two parts below are independent and both required.

(2a) Prove, or refute by an explicit controlled counterexample, the conjecture that the completed observer algebra is never of Type III - Type I when the observer's causal region contains a complete Cauchy hypersurface, and otherwise Type II (finite or infinite), with ΨHH\Psi_{\mathrm{HH}} as the tracial state or weight. An explicit controlled spacetime whose completed observer algebra is of Type III refutes (2a) outright; a proof must establish the full stated alternative.

(2b) Separately, give precise conditions on spacetime topology, horizons, or the behavior of the no-boundary amplitude under which the algebra is of Type I, Type II1, Type II-infinity, or Type III. A refutation or proof in (2a) does not by itself discharge (2b).

(3) Nonperturbative normalization of −log⁡Z-\log Z: define Z=⟨ΨHH∣ΨHH⟩Z = \langle\Psi_{\mathrm{HH}}|\Psi_{\mathrm{HH}}\rangle unambiguously beyond the semiclassical approximation - resolving how topology change and nonperturbative saddle contributions enter it - so that the entropies of observer states belonging to different universes can be compared.

(4) Topology-changing directions: identify and control the directions in field space, including topology-changing transitions as in the Schwarzschild-de Sitter example, along which ΨHH\Psi_{\mathrm{HH}} is expected to grow without bound, and determine whether they genuinely preclude a maximum-entropy vector in the perturbative Hilbert space, or whether a more refined nonperturbative framework restores normalizability.

Scope: this problem comprises the four open directions of Witten (2024) that concern the maximum-entropy status of ΨHH\Psi_{\mathrm{HH}} and the type of A^obs\widehat{\mathcal{A}}_{\mathrm{obs}}. Two further directions identified in the same paper - the persistence of the 1/q1/q operator-product characterization of Aobs\mathcal{A}_{\mathrm{obs}} beyond perturbation theory in GG, and the analytic continuation and positivity of the anti-de Sitter large-NN algebra away from integer NN - concern the construction of the algebra itself and are treated here as neighboring open problems, not part of the stated objective.

References

  1. E. Witten, "A background-independent algebra in quantum gravity," J. High Energ. Phys. 2024, 77 (2024). DOI: 10.1007/JHEP03(2024)077; arXiv:2308.03663. https://arxiv.org/abs/2308.03663
  2. E. Witten, "A Background Independent Algebra in Quantum Gravity," Gen. Rel. Grav. 57, 17 (2025). DOI: 10.1007/s10714-025-03360-x. https://link.springer.com/article/10.1007/s10714-025-03360-x
  3. V. Chandrasekaran, R. Longo, G. Penington, E. Witten, "An Algebra of Observables for de Sitter Space," J. High Energ. Phys. 2023, 82. DOI: 10.1007/JHEP02(2023)082; arXiv:2206.10780. https://arxiv.org/abs/2206.10780
  4. G. Penington, E. Witten, "Algebras and States in JT Gravity," arXiv:2301.07257. https://arxiv.org/abs/2301.07257
  5. J. Kudler-Flam, S. Leutheusser, G. Satishchandran, "Algebraic Observational Cosmology," Phys. Rev. D 114, 045007 (2026). DOI: 10.1103/gbv2-n513; arXiv:2406.01669. https://arxiv.org/abs/2406.01669
  6. A. Blommaert, J. Kudler-Flam, E. Y. Urbach, "Absolute entropy and the observer's no-boundary state," J. High Energ. Phys. 2025, 113. DOI: 10.1007/JHEP11(2025)113; arXiv:2505.14771. https://arxiv.org/abs/2505.14771
  7. J. Cotler, K. Jensen, "Norm of the no-boundary state," J. High Energ. Phys. 2026, 180. DOI: 10.1007/JHEP03(2026)180; arXiv:2506.20547. https://arxiv.org/abs/2506.20547
  8. J. Cotler, "Higher-loop norm of the no-boundary state," arXiv:2601.20993. https://arxiv.org/abs/2601.20993
  9. J. Cotler, K. Jensen, "Inner products and closed universes," arXiv:2608.26253. https://arxiv.org/abs/2608.26253
  10. A. Blommaert, A. Levine, "Sphere amplitudes and observing the universe's size," J. High Energ. Phys. 2026, 136. DOI: 10.1007/JHEP07(2026)136; arXiv:2505.24633. https://arxiv.org/abs/2505.24633
  11. D. Bak, C. Kim, S.-H. Yi, "Janus Deformation of de Sitter Space and Transitions in Gravitational Algebras," J. High Energ. Phys. 2024, 94. DOI: 10.1007/JHEP11(2024)094; arXiv:2407.04316. https://arxiv.org/abs/2407.04316
  12. C. Gomez, "On the algebraic meaning of quantum gravity for closed Universes," arXiv:2311.01952. https://arxiv.org/abs/2311.01952
  13. B. Chen, J. Xu, "An algebra for covariant observers in de Sitter space," arXiv:2511.00622. https://arxiv.org/abs/2511.00622
  14. H. Geng, Y. Jiang, J. Xu, "Algebras, Entanglement Islands, and Observers," Class. Quantum Grav. 43, 135012 (2026). DOI: 10.1088/1361-6382/ae811e; arXiv:2506.12127. https://arxiv.org/abs/2506.12127
  15. K. Jensen, S. Raju, A. J. Speranza, "Holographic observers for time-band algebras," J. High Energ. Phys. 2025, 242. DOI: 10.1007/JHEP06(2025)242; arXiv:2412.21185. https://arxiv.org/abs/2412.21185
  16. B. Freivogel, A. Fumagalli, D. M. Hofman, "JT gravity on the worldline," arXiv:2607.14213. https://arxiv.org/abs/2607.14213
  17. H. Liu, "Towards a holographic description of closed universes," arXiv:2509.14327. https://arxiv.org/abs/2509.14327
  18. M. Cirafici, "Observers, local measurements, and topology," arXiv:2607.15007. https://arxiv.org/abs/2607.15007
  19. E. Witten, "Bras and Kets in Euclidean Path Integrals," Beijing J. Pure Appl. Math. 3, 1 (2026). DOI: 10.4310/bpam.260113013520; arXiv:2503.12771. https://arxiv.org/abs/2503.12771
  20. J. Hartle, S. W. Hawking, "Wave Function of the Universe," Phys. Rev. D 28, 2960 (1983). DOI: 10.1103/PhysRevD.28.2960. https://journals.aps.org/prd/abstract/10.1103/PhysRevD.28.2960