Internal quantum measurements and controlled Born probabilities in closed gravity

Problem statement

Can the partially observed gravitational path integral of a closed universe produce a consistent internal measurement theory, with a quantitative bound on deviations from semiclassical Born probabilities? The target is one operational reconstruction theorem, supported by an explicit dynamical gravitational model, or a no-go theorem for the admissible class below. The problem grows out of Nomura and Ugajin's Physical Predictions in Closed Quantum Gravity [1], especially its conditioning, observational algebra, and fluctuation claims. A resolution would establish when a one-dimensional exact physical state can support the quantum experiments described by an internal observer. The admissible class consists of regulated families of generally covariant gravity theories in at least two spacetime dimensions with compact spatial slices. A two-dimensional dilaton-gravity model with matter on a spatial circle is an eligible starting point. The clock, apparatus, measured system, and environment must be degrees of freedom of that theory; apparatus couplings and their gravitational backreaction must be specified. A dual or matrix description is admissible if it reproduces the specified closed-universe kernels and constraints. For each microscopic sector, the exact closed-universe Gram matrix must have rank one. The same path-integral prescription must include nonzero no-boundary and cylindrical contributions and determine their relative weights. Their values cannot be selected independently to fit a desired experiment. An effective observational algebra may be constructed in an enlarged kinematical space. It is not required to act nontrivially within the one-dimensional exact physical Hilbert space. The question is whether that effective algebra and its measurement probabilities follow consistently from internal conditioning, the constraint measure, and tracing unobserved degrees of freedom. Any observer-dependent restriction on replica connections must be derived from those operations. Merely imposing such a restriction as an additional rule does not establish the requested result. For a finite internal experiment with setup records $C$ and outcomes $i$, construct nonnegative weights $W_\lambda(i,C)$ and probabilities $$p_\lambda(i\mid C)=\frac{W_\lambda(i,C)}{\sum_j W_\lambda(j,C)},$$ where $\lambda$ labels a microscopic realization. Identify an associative observable algebra, a positive normalized state functional, and completely positive measurement instruments. The sum of the outcome maps must preserve trace on the specified effective state space. Include any memory needed for successive measurements, rather than silently assuming a Markovian reduced dynamics. An unresolved outcome must depend on its effect or projector, not on a chosen orthonormal basis inside its subspace. In the ordinary quantum-mechanical limit, a compulsory calibration is $$\sigma=\begin{pmatrix}3/10&1/10&0\\1/10&3/10&0\\0&0&2/5\end{pmatrix},\qquad \Pi_X=\operatorname{diag}(1,1,0),\qquad \Pi_Y=\operatorname{diag}(0,0,1).$$ The result must be $p_X=3/5$, $p_Y=2/5$ under every change of basis inside $X$. This test concerns an unresolved outcome, not postselection onto a particular coherent superposition. It is only a calibration; solving this finite-dimensional exercise does not solve the challenge. The same relational event must have the same probability in two valid bulk gauges and under a monotone reparameterization of its clock coordinate. State the domain in which the clock is a good gauge fixing and account for its Jacobian and any residual gauge volume. The demand is equivalence of descriptions of the same experiment, not equality between different physical clock experiments. The quantitative target concerns normalized probabilities and finite measurement histories. Let $q_{\rm sc}$ be the distribution independently obtained from the model's semiclassical matter-and-apparatus dynamics, and let $\mu_C$ be the microscopic averaging law, with any update by the setup records explicitly derived. For a declared family $\mathcal E$ of experiments with at most $n_0$ outcomes and $m_0$ measurement times, and setup probabilities bounded below by a fixed $p_{\min}>0$, establish a bound of the form $$\sup_{\mathcal E}\left(\mathbb E_{\lambda\sim\mu_C} \bigl[\operatorname{TV}(p_\lambda(\cdot\mid C),q_{\rm sc}(\cdot\mid C))^2\bigr]\right)^{1/2} \leq \frac{A(n_0,m_0,p_{\min})}{\sqrt{D_{\rm eff}}} +\epsilon_{\rm sc}+\epsilon_{\rm reg},$$ where $\operatorname{TV}(p,q)=\frac12\sum_h|p(h)-q(h)|$, and all terms are defined and controlled in an explicit joint limit. The experiment family must include coherent interference, a recorded measurement, unresolved outcomes of unequal ranks, and at least two successive measurements. It cannot be chosen after inspecting a particular microscopic realization. Derive $D_{\rm eff}$ from the model's environment and replica covariance. In a Gaussian limit it must reduce to the appropriate effective dimension $(\operatorname{Tr}K)^2/\operatorname{Tr}(K^2)$ for the positive covariance operator $K$ identified by the calculation. State the allowed spectral and correlation assumptions, and justify any bounds on small normalization denominators. If separate topological sectors require separate effective dimensions, a bound with explicitly derived sector terms is acceptable. A count of available basis states alone is insufficient. The construction must exhibit a nonempty regime in which the bound vanishes as the environment's effective dimension grows and the stated approximation errors vanish. Success would turn the proposed recovery of semiclassical predictability into a controlled operational result. An obstruction would identify a common assumption that must change.

Current state

Nomura and Ugajin [2] proposed suppression of microscopic fluctuations through partial observability. Their explicit microscopic model uses simplifying assumptions about independence and factorization. The linked paper [1] extends that program to no-boundary terms, an observational GNS construction, and histories. The elementary contraction scaling is not the open part of this challenge: independent equal-weight real Gaussian channels already give fractional standard deviation $\sqrt{2/d}$. The operational definitions need an additional consistency check. For unresolved orthonormal alternatives $f_i^a$, the independent double sums in [1, Eqs. (4.6) and (5.4)] insert $|\sum_a f_i^a\rangle\langle\sum_a f_i^a|$. This differs from the unresolved-outcome projector $\sum_a|f_i^a\rangle\langle f_i^a|$. For the calibration state in the problem, the former gives normalized $p_X=2/3$, or $1/2$ after changing the sign of one basis vector, instead of the invariant value $3/5$. This finite-dimensional observation motivates the calibration requirement. Replacing that sum or invoking the abstract GNS theorem supplies no derivation of the dynamical gravitational measurement map or its error bound. There are substantial existing observer constructions. Harlow, Usatyuk, and Zhao [3] develop observer decoherence and controlled toy-model calculations. Abdalla and collaborators [4] obtain a larger observer Hilbert space in JT gravity using a prescription for connecting the observer's worldline between replicas. Harlow [5] includes Hartle-Hawking processes and relates an observer rule to averaging. These are important starting points; deriving the relevant restrictions from the internal instrument and the same gravitational measure is a further requirement here. Zhao [6] explicitly reconstructs bulk quantum mechanics from a one-dimensional topological de Sitter model without mixing microscopic sectors. Thus the unqualified question of whether internal quantum mechanics can coexist with the one-state property already has concrete toy-model answers. Wei [7] develops relational observers and timekeepers in a gravitational path-integral model. The present challenge requires a dynamical model in at least two spacetime dimensions, internal apparatus couplings, compatible instruments, and a bound for normalized probabilities of successive measurements. Held and Maxfield [8] analyze how the constraint inner product, gauge fixing, and its measure fit together, including limits of globally valid clocks. Ivo, Li, and Maldacena [9] compute no-boundary density matrices by tracing an unobserved region. These results address essential parts of the proposed construction. The remaining joint target is the derivation of all these operational and quantitative properties from one controlled gravitational model. In the literature checked for this proposal, I did not identify a result satisfying the full resolution criteria below. This is a claim about that combined target, not an assertion that individual ingredients such as POVMs, GNS reconstruction, relational time, or Gaussian concentration are new or unsolved.

Resolution criteria

A positive resolution must meet all five conditions in one compatible construction. 1. Give the action or explicit dual definition, regulator, integration contour, constraint measure, internal apparatus couplings, and microscopic averaging law of an admissible model. Establish the rank-one exact Gram matrix in each microscopic sector. Compute no-boundary and cylindrical terms, their relative normalization, and the connected replica terms needed for the error estimate. Bound omitted contributions and exhibit a parameter regime with a controlled approximation or regulator limit. 2. Derive a positive normalized observational state and a compatible associative algebra with explicit multiplication and identity. Derive completely positive outcome maps whose sum preserves trace on the stated effective state space, with any required memory included. Obtain normalized outcome and finite-history probabilities from this instrument and the same path integral. An observer-specific replica selection rule must follow from the derivation rather than be added independently. 3. Prove invariance under basis changes within unresolved outcomes and under changes of valid bulk gauge describing the same relational event, including a nontrivial clock-coordinate reparameterization and its measure. Recover $p_X=3/5$ for the stated three-dimensional calibration. Show explicitly how unequal outcome ranks are treated. The algebra representation must preserve its declared product and identity. 4. Exhibit internal experiments with coherent interference and recorded outcomes, including two successive noncommuting measurements. Compare their gravitational probabilities with the independently computed semiclassical Born and measurement-update probabilities. Give controlled errors for the joint distributions and their marginals. For any history probabilities claimed without explicit measurement records, specify and verify the necessary decoherence condition. 5. Prove the normalized-probability bound stated in the problem, or a stronger bound implying it in a stated nonempty regime. Specify the experiment family before the microscopic realization, identify all constants and spectral/correlation assumptions, derive the microscopic weighting after conditioning, and control the random denominator. Derive $D_{\rm eff}$ or explicit sector-dependent substitutes from the model; recover the Gaussian effective-rank expression in that limit. Demonstrate that the bound vanishes in an explicit joint limit and provide checkable derivations and reproducible calculations. A negative resolution must prove that no member of the admissible class can meet these five requirements. It may establish a contradiction from a subset of assumptions shared by the whole class, but it must state that subset and identify the operational condition that fails. The proof must still allow conditioning in an enlarged kinematical space; the observation that operators on a one-dimensional Hilbert space are scalars is not sufficient. A failure of one selected model is a partial result. Correcting a microscopic double sum, quoting GNS or a standard instrument theorem, repeating an existing one-dimensional topological model, or numerically checking independent Gaussian channels alone is a partial result. A claimed completion must map its evidence to every applicable criterion.

Citations

1. Y. Nomura and T. Ugajin, Physical Predictions in Closed Quantum Gravity, arXiv:2602.13387v3 (2026).
2. Y. Nomura and T. Ugajin, Nonperturbative Quantum Gravity in a Closed Lorentzian Universe, arXiv:2505.20390 (2025).
3. D. Harlow, M. Usatyuk and Y. Zhao, Quantum mechanics and observers for gravity in a closed universe, arXiv:2501.02359 (2025).
4. A. I. Abdalla, S. Antonini, L. V. Iliesiu and A. Levine, The gravitational path integral from an observer's point of view, arXiv:2501.02632 (2025).
5. D. Harlow, Observers, alpha-parameters, and the Hartle-Hawking state, arXiv:2602.03835 (2026).
6. Y. Zhao, "It from Bit": The Hartle-Hawking state and quantum mechanics for de Sitter observers, arXiv:2602.05939 (2026).
7. Z. Wei, Observers and Timekeepers: From the Page-Wootters Mechanism to the Gravitational Path Integral, arXiv:2506.21489 (2025).
8. J. Held and H. Maxfield, Gravitational Hilbert spaces: invariant and co-invariant states, inner products, gauge-fixing, and BRST, arXiv:2509.05412 (2025).
9. V. Ivo, Y.-Z. Li and J. Maldacena, The no boundary density matrix, arXiv:2409.14218 (2024).

Linked papers

• Physical Predictions in Closed Quantum Gravityrelated7 Sept 2026
• Basis covariance and fluctuation diagnostics for measurements in closed quantum gravitydeclared at submission7 Sept 2026