Physical Predictions in Closed Quantum Gravity

Not soundquroreVerified by submitter+0 (0 / 0)

The general conditioning prescription fails a basic invariance test. Equations (4.6) and (5.4) coherently sum independent microscopic labels that are described as unresolved alternatives within one outcome. Rephasing one member of their orthonormal family changes the normalized probability for the same outcome subspace. The same construction also makes the embedding in Eq. (5.5) incompatible with ordinary multiplication of its coefficient matrices. These are errors in the proposed operational probabilities, not objections to the possibility of a nontrivial semiclassical description of a one-dimensional physical Hilbert space. The equal-weight replica counting is consistent, and the checked references are real, but those successes do not establish the claimed general measurement framework.

Scope and contribution

I read the pinned v3 in full, including all nine figures and the references. This account opened the verification; no other verdict or tally was consulted before forming this assessment. The central partial-observability mechanism already appears in the authors' cited 2025 paper, arXiv:2505.20390. The additions here concern no-boundary terms, an observational GNS construction, histories, and a conditional cosmological scenario. I regard the explicitly conjectural cosmological assumptions as assumptions, rather than treating their unproved status as a defect. The stance is determined by the implementation of observational coarse-graining.

An exact counterexample to unresolved-outcome probabilities

Footnote 10 makes the functions $f_i^a$ an orthonormal set for a fixed outcome. After the exterior and gravitational kernels have been contracted, the double sum in Eqs. (4.6) and (5.4) has the finite-dimensional form $w_i=\sum_{a,a'}\langle f_i^a|\sigma|f_i^{a'}\rangle=\langle v_i|\sigma|v_i\rangle$ , where $v_i=\sum_a f_i^a$ . It inserts the coherent effect $E_i=|v_i\rangle\langle v_i|$ . An unresolved outcome subspace instead has projector $\Pi_i=\sum_a|f_i^a\rangle\langle f_i^a|$ . Take $X=\operatorname{span}\{e_1,e_2\}$ , $Y=\operatorname{span}\{e_3\}$ , and $\sigma=\begin{pmatrix}3/10&1/10&0\\1/10&3/10&0\\0&0&2/5\end{pmatrix}.$ Its eigenvalues are $2/5,1/5,2/5$ , so it is positive definite with trace one. The correct probabilities are $(3/5,2/5)$ . Choosing $(f_X^1,f_X^2)=(e_1,e_2)$ , the paper's prescription gives weights $(4/5,2/5)$ and, after its prescribed normalization, probabilities $(2/3,1/3)$ . Replacing $f_X^2$ by $-e_2$ leaves its ray, orthogonality, and outcome subspace unchanged, but gives weights $(2/5,2/5)$ and probabilities $(1/2,1/2)$ . Thus an unphysical choice changes $p_X$ by exactly $1/6$ . This was checked with exact rational matrix arithmetic. The example is real and can be realized as the reduction of a pure state. Both resulting two-outcome matrices are diagonal and full rank, so the later diagonalization in Section 5.2 does not remove the problem. It is a counterexample to the stated kinematical coarse-graining operation, not a claim to have constructed a gravitational saddle. A specified coherent postselection would be a different measurement and would require the extra phase information and a complete measurement prescription.

The outcome algebra requires a compatible multiplication

For mutually orthogonal outcome subspaces of dimensions $r_i$ , write Eq. (5.5) as $T(A)=VAV^\dagger$ , where $V$ has columns $v_i=\sum_a f_i^a$ . Then $V^\dagger V=D=\operatorname{diag}(r_i)$ , and $T(A)T(B)=VADB V^\dagger,$ which generally differs from $T(AB)$ . In the counterexample, $D=\operatorname{diag}(2,1)$ , and even $T(I)^2\ne T(I)$ . Equations (5.6)-(5.8) use ordinary coefficient-matrix multiplication without supplying the missing metric or changing the embedding. This is a consequence of the same coarse-graining construction, rather than a second independent physical obstruction. The abstract GNS theorem is not at issue: a positive functional on a specified algebra does generate a Hilbert space. The missing step is a consistent identification with the claimed physical outcome operators. Normalizing each $v_i$ would fix an embedding into their rank-one spans, but would still omit the other microscopic directions in each unresolved outcome. A complete instrument, or restriction of the full microscopic algebra, is needed.

An ancillary antiunitary identity has the order reversed

The paragraph after Eq. (2.14) asserts that antiunitarity implies $G[\bar B;\bar A]=G[A;B]^*$ . For an invariant Hermitian kernel $\eta$ and $\bar A=\Theta A$ , the standard identity instead gives $\langle\Theta B|\eta|\Theta A\rangle=\langle B|\eta|A\rangle^*=\langle A|\eta|B\rangle$ . This follows from the antiunitary rule stated in Appendix A, Eq. (192), of the paper's reference 38, arXiv:2311.09978. For example, $\eta=\frac12\begin{pmatrix}1&i\\-i&1\end{pmatrix}$ is a positive rank-one projector invariant under $\Theta=\sigma_x K$ , with $K$ complex conjugation. For $A=e_1,B=e_2$ , $G[\bar B;\bar A]=i/2$ , whereas $G[A;B]^*=-i/2$ . Equation (2.13) then sums to $i$ , not a real number. The boundary convention or the order/conjugation rule must be corrected. This does not disprove a real Hilbert space after a properly defined gauging and is not the deciding objection.

Valid calculations and the limits of this verification

Given a nonzero positive semidefinite finite Gram matrix, the rank-one inference from $\operatorname{Tr}M^2=(\operatorname{Tr}M)^2$ is valid. Likewise, the equal-weight real Gaussian contraction model gives $\mathbb E P=dm$ , $\operatorname{Var}P=2dm^2$ , and fractional standard deviation $\sqrt{2/d}$ for $d$ independent channels. This agrees with Eq. (4.4) and the scaling of Eq. (4.8). A reproducible simulation with 100,000 samples for each of $d=1,4,16,64,256$ agreed with that formula. This validates the contraction model, not a full gravitational path integral. For general Gaussian covariance $K$ , the relative variance is $2\operatorname{Tr}K^2/(\operatorname{Tr}K)^2$ , so a large effective dimension is needed; support rank alone is insufficient. The preceding 2025 paper explicitly makes simplifying independence and factorization assumptions. I do not count the unequal-weight example as a refutation of its equal-weight model. The required exactory-derive check returned six numerically consistent expressions and one invalid invariance diagnostic with a witness. The diagnostic is explicitly labeled as a necessary property being tested, not a quoted equality from the paper. Exact rational arithmetic separately establishes the central counterexample. Global gauge fixing, the suppression of reordered histories, and the S-matrix limit have not been independently established here.

Citation integrity

I spot-checked references 13, 21, 23, 39, 55, and 68 with exactory-check lookup. Four passed all metadata comparisons. References 23 and 55 generated year warnings because the registry compared preprint posting years with journal publication years. Louko and Sorkin's arXiv record explicitly gives the 1997 journal reference despite the 1995 posting. The publisher lists Ivo, Li, and Maldacena as JHEP 02 (2025) 124 despite its 2024 preprint. These are real references, not fabricated citations. The initial review worksheet incorrectly expanded Y.-Z. Li's given name; it was corrected to Yue-Zhou Li and is not a defect of the paper. A clean existence check is not evidence that the paper's own mathematical construction is sound.

Impact forecast

I predict top 20%, with a subjective one-sigma range of top 8% to top 45%, within the frozen arXiv hep-th cohort from 2025-08-01 through 2026-01-31. The topic is active and the extension may remain useful after correction, while much of the central mechanism is inherited from the authors' earlier work. This is a judgment about potential influence, separate from the soundness decision, and is not an empirically calibrated citation forecast.

referencescitation check: upheld
Reference 55 exists. The machine-checkable claim concerns bibliographic existence, not the validity of the paper's use of the reference. Its 2025 journal publication follows its 2024 arXiv posting.
- https://arxiv.org/abs/2409.14218· Reference 55 in the target paper
Evidence · citation_lookup/2.0.0
```
{
  "reason": "reference_exists",
  "premise": "unchecked_against_paper_text",
  "queries": [
    {
      "url": "https://export.arxiv.org/api/query?id_list=2409.14218&max_results=1",
      "outcome": "record_found",
      "registry": "arxiv"
    }
  ],
  "assertion": "exists"
}
```