Break the Bisognano–Wichmann/CHM circularity: geometry from non-symmetric modular flow? (OP-46)

Status: ATTEMPT executed — the deepest reorganization-vs-generation test (OP-46 clause (iii)) is carried out and the obstruction is SHARPENED, per-route, from "no surveyed construction satisfies it" (iteration 3) to a conditional near-no-go: under the standard stress-tensor-flux readout, clause (iii) is unsatisfiable-by-construction. Net: PROMOTED toward a no-go for clause (iii), short of a full theorem; the single live escape (a non-CKV readout on a non-symmetric state) is precisely located. Last updated: 2026-06-08 Iteration: 4

This note executes the iteration-3 open subquestion #3 of 2026-06-08-iter3-encode-vs-generate-criterion.md ("does any $(\mathcal{A},\omega)$ pair with non-symmetric modular flow yield emergent geometry, satisfying clause (iii)?"). It targets OP-46 clause (iii) — the reorganization-vs-generation falsifier — and asks the sharpest possible question: can a non-geometric (non-BW / non-Hadamard / non-CHM) Tomita–Takesaki modular flow output a semiclassical Killing/boost generator and a horizon, as the only thing that would upgrade the modular / crossed-product program from "encode" to "generate"? It sharpens the Borchers–Wiesbrock route (half-sided modular inclusion $\Rightarrow$ positive-energy translation) and asks whether that route genuinely generates causal / translation structure from neutral algebraic data, or still presupposes a positivity / vacuum input.

Scope honesty (folded from referee): the near-no-go below is conditional on the standard geometric readout $K = 2\pi\int_\Sigma \xi^\mu T_{\mu\nu}\,d\Sigma^\nu$. It is not a no-go over all conceivable readouts. The Cao–Carroll mutual-information/MDS readout and a relative-entropy / Connes-spectral-distance readout are untouched by the Sorce CKV theorem and remain the live escape. Two of the underlying facts are theorem-grade and web-verified (Borchers 1992 / Wiesbrock 1993; Sorce 2024); the chaining inference between them is supplied here and is tagged [INFERENCE].

---

Goal

Decide whether ANY pair $(\mathcal{A}, \omega)$ — a von Neumann algebra with cyclic-separating $\Omega$ — whose modular flow $\sigma_t$ is not a known geometric (Bisognano–Wichmann / Hislop–Longo / CHM / Hadamard) flow can nonetheless have a semiclassical boost/Killing structure and a horizon recovered as output. This is the precise content of OP-46 clause (iii) (the circularity-falsifier of 2026-06-08-iter3-encode-vs-generate-criterion.md, line 102): geometry must vary nontrivially over non-symmetric states, or the whole modular program is reconstruction, not generation.

Concretely:

1. Build the cleanest "generation candidate" — discard the entropy/observer machinery and demand only causal/translation output — and find the exact step at which geometry re-enters. [goal]
2. Determine whether Borchers–Wiesbrock generates or merely relocates the positivity/geometric input. [goal]
3. Decide the status of clause (iii) under the standard readout, and name the escape hatch if there is one. [goal]

---

Setup (real math)

Standard data $\mathcal{D} = (\mathcal{A}, \Omega)$: a von Neumann algebra $\mathcal{A}$ with cyclic-separating vector $\Omega$. The Tomita operator $S = J\Delta^{1/2}$ (closure of $a\Omega \mapsto a^\dagger\Omega$) supplies the modular conjugation $J$ and the modular operator $\Delta$. The modular Hamiltonian is $K = -\log\Delta$, the modular flow is

$$\sigma_t(a) = \Delta^{it}\, a\, \Delta^{-it},$$

a one-parameter automorphism group satisfying the KMS condition at inverse temperature $\beta = 1$ w.r.t. $\omega = \langle\Omega,\,\cdot\,\Omega\rangle$. [ESTABLISHED] (Tomita–Takesaki; see ../domains/mathematics.md.)

"Geometric" modular flow means $\sigma_t$ acts by a point diffeomorphism $\phi_t$ of an underlying region — equivalently

$$K \;=\; 2\pi \int_\Sigma \xi^\mu\, T_{\mu\nu}\, d\Sigma^\nu$$

for some vector field $\xi$. This is the standard geometric readout: the map "modular Hamiltonian $\to$ geometry" through identification with the stress-tensor flux of $\xi$.

Two load-bearing theorems frame clause (iii).

Theorem 1 (GEOMETRIC $\Rightarrow$ CKV)

If $\sigma_t$ is geometric on a causally complete region $\mathcal{V}$, the generating field $\xi$ is necessarily a future-directed conformal Killing vector of the background:

$$\nabla_\mu \xi_\nu + \nabla_\nu \xi_\mu \;=\; \frac{2}{d}\,(\nabla\!\cdot\!\xi)\, g_{\mu\nu}, \qquad \xi^2 < 0 .$$

When $\xi$ is not a true isometry, the trace condition forces $T^\mu{}_\mu = 0$, i.e. a genuine CFT. [ESTABLISHED] (Sorce 2024, arXiv:2403.18937 — sole author; explicit 2d constructions Caminiti–Capeccia–Ciambelli–Myers, arXiv:2502.02633, JHEP 08 (2025) 166, eq. (8); the older anisotropic-Unruh input is Arias–Casini–Huerta–Pontello, arXiv:1707.05375.)

Citation correction (referee, severity minor): the iteration-3 verdict table and earlier drafts attributed this to an "Arias–Sorce CKV theorem." No Arias–Sorce coauthored paper exists. The geometric-modular-flow $\Rightarrow$ CKV result is Sorce alone (arXiv:2403.18937); "Arias" appears in the different Blanco–Casini–Huerta-adjacent local-modular / local-temperature line (Arias et al., arXiv:1707.05375) and was conflated. Attribute to Sorce 2024, with Caminiti et al. 2502.02633 for the explicit 2d realization. [corrected]

Theorem 2 (POSITIVITY $\Leftrightarrow$ MODULAR-GEOMETRY — the Borchers–Wiesbrock characterization)

For $\mathcal{A}$ a wedge algebra with $\Omega$ invariant under a one-parameter unitary group $U(a)$, the spectrum condition $P \ge 0$ implies the commutation relations

$$\Delta^{it}\, U(a)\, \Delta^{-it} = U(e^{-2\pi t}\, a), \qquad J\, U(a)\, J = U(-a).$$

This forward direction is Borchers 1992 (CMP 143, 315). The converse — and the statement that a $\pm$half-sided modular inclusion is equivalent to such a positive-generator translation — is Wiesbrock 1993 (CMP 157, 83; gap filled Araki–Zsidó 2004, arXiv:math/0412061). Together they form the biconditional $P\ge0 \Leftrightarrow$ modular commutation relation. [ESTABLISHED]

Attribution precision (referee): naming the equivalence "the Borchers biconditional" is defensible shorthand but conflates Borchers' forward theorem with the Wiesbrock converse that actually supplies "one premise, not two." This note writes Borchers–Wiesbrock characterization throughout. The three modular groups of a modular inclusion/intersection generate $PSL(2,\mathbb{R})$ (Lashkari–Leung–Moosa–Ouseph 2024, arXiv:2412.19882, JHEP 10 (2025) 153). [corrected]

---

The attempt

I attempt to build the generation candidate and find the exact step at which it imports geometry.

Step A — set up the cleanest candidate

Discard the entropy/observer machinery; demand only causal/translation output. The strongest engine is a half-sided modular inclusion (HSMI):

$$\mathcal{N} \subset \mathcal{M}, \quad \text{shared cyclic-separating } \Omega, \quad \Delta_{\mathcal{M}}^{it}\, \mathcal{N}\, \Delta_{\mathcal{M}}^{-it} \subset \mathcal{N} \ \text{ for } t \le 0 .$$

Borchers–Wiesbrock then outputs a one-parameter unitary $U(a)$, $a \ge 0$, with generator

$$P \;=\; \frac{1}{2\pi}\big(\log\Delta_{\mathcal{N}} - \log\Delta_{\mathcal{M}}\big) \;\ge\; 0, \qquad \Delta_{\mathcal{M}}^{it}\, U(a)\, \Delta_{\mathcal{M}}^{-it} = U(e^{-2\pi t}\, a).$$

The pair $\{\log\Delta_{\mathcal{M}},\, P\}$ closes into the $ax{+}b$ (boost $+$ translation) Lie algebra; a second inclusion gives $PSL(2,\mathbb{R})$ (Lashkari et al. 2412.19882). So from algebraic inclusion data one outputs a positive-energy translation and a chunk of conformal symmetry — genuinely as output. This is the maximal honest "generation." [ESTABLISHED inputs; INFERENCE that this is the maximal candidate, high]

Scope (referee): a single HSMI yields only the $ax{+}b$ pair (one boost $+$ one positive translation), not a full Killing/Möbius algebra; the Möbius/$PSL(2,\mathbb{R})$ and Poincaré groups require $\ge 2$ HSMIs (Wiesbrock; Guido–Longo–Wiesbrock; Lashkari et al.). The "+ horizon" language below is the special Rindler/CHM case, not delivered by one HSMI. [corrected]

Step B — locate the smuggle precisely

The HSMI hypothesis is exactly $\Delta_{\mathcal{M}}^{it}\mathcal{N}\Delta_{\mathcal{M}}^{-it} \subset \mathcal{N}$ for one sign of $t$ only. By the Borchers–Wiesbrock characterization this half-sidedness is logically equivalent to the spectrum condition $P \ge 0$. Hence the input is not "neutral inclusion data": choosing a half-sided (rather than two-sided) inclusion is choosing energy positivity. The generated $U(a)$ inherits its positivity from the premise; nothing is created. Positivity and half-sidedness are one premise, not two. [ESTABLISHED — Wiesbrock biconditional]

Step C — promote to a near-theorem for clause (iii)

Chain the two theorems. Suppose a state $\omega$ has modular flow that is not geometric (the clause-(iii) hypothesis). To "recover a semiclassical boost/Killing structure $+$ horizon as output" under the standard readout, one must produce a vector field $\xi$ with $K = 2\pi\int \xi^\mu T_{\mu\nu}$ generating the recovered flow. But Theorem 1 says any such geometric $\xi$ is a CKV — a (conformal) symmetry of some metric, hence existing only on a (conformally) symmetric background. Therefore the readout step that turns modular data into a Killing/boost generator can succeed only where a CKV exists, i.e. on a (conformally) symmetric background.

Consequence: NON-geometric modular flow $\Rightarrow$ no CKV available $\Rightarrow$ the boost/horizon readout has no target to land on. The would-be "generation from a non-symmetric state" is, on this readout, unsatisfiable-by-construction: the very object it must output (a CKV / horizon) presupposes the symmetric structure whose absence defined the non-geometric case. [INFERENCE, medium–high — chains two established theorems; the chaining step is supplied here]

Word choice (referee): "VACUOUS" was the iteration-internal label; the precise statement is "self-undercutting / unsatisfiable-by-construction for the geometric readout" — the clause has content, it is just unmeetable by that one readout. Adopted below.

Step D — test the recovery candidates against this

• (i) Kabat–Lifschytz, arXiv:1812.02915 ("Emergence of spacetime from the algebra of total modular Hamiltonians"). A bulk point $=$ a weakly-maximal Lie subalgebra of total modular Hamiltonians; the metric is recovered up to a conformal factor, the conformal factor fixed only from the (presupposed) extremal-surface equation. Input $=$ the CFT vacuum $+$ the conformal algebra. This is recovery of a presupposed AdS dual, satisfying $g_3$ (symmetric vacuum). Recovery, not generation. [ESTABLISHED]
• (ii) Lashkari et al., arXiv:2412.19882. Modular inclusion $\Rightarrow PSL(2,\mathbb{R})$, but on conformal GFF in the vacuum — positivity implicit; the "emergence" of AdS$_2$ is a geometric reinterpretation of a non-local Hilbert transform, not generation from non-geometric data. [ESTABLISHED]
• (iii) Leutheusser–Liu, arXiv:2110.05497 / 2112.12156 (Phys. Rev. D 108, 086020 (2023)). Emergent interior infalling time $+$ a sharp horizon at strict $N=\infty$ from an emergent Type III$_1$ algebra and the half-sided modular translation; positivity inherited from the boundedness-below of the boundary CFT Hamiltonian on a fixed AdS/TFD background. [ESTABLISHED]

All three confirm Step C: the symmetric/positive input re-enters at the readout. [INFERENCE]

---

Where it breaks / what moved

didItMove = sharpened-obstruction.

The construction breaks at Step C's readout map, and the break is now SHARP (per-route) rather than merely suspected. Two independent gates close on the same point.

Gate 1 (HSMI route). The Borchers–Wiesbrock characterization collapses "purely algebraic inclusion data" into "energy positivity": a half-sided (vs. two-sided) modular inclusion is equivalent to $P\ge0$ (Borchers 1992; Wiesbrock 1993). So Borchers–Wiesbrock RELOCATES rather than REMOVES the positivity input — it does not generate a causal structure from neutral data; it converts an assumed spectrum condition into an $ax{+}b$ / $PSL(2,\mathbb{R})$ group. The generated translation is positive because the premise was. A single HSMI yields only the $ax{+}b$ pair, not a Killing/horizon algebra. [ESTABLISHED]

Gate 2 (geometric readout). The Sorce CKV theorem closes the other gate: the readout "modular data $\to$ boost/Killing generator $+$ horizon" can only produce a conformal Killing vector, which exists only on a (conformally) symmetric background. A genuinely non-geometric modular flow (the clause-(iii) hypothesis) furnishes no CKV, so there is nothing for the horizon/boost readout to recover. [INFERENCE on the consequence; ESTABLISHED lemma]

Net. Relative to the stress-tensor-flux readout, the encode$\to$generate frontier coincides with the presence of a positivity/CKV input, and that input is provably equivalent (per route) to the geometric structure one hoped to derive. Clause (iii) is therefore unsatisfiable-by-construction for this readout — not merely unmet.

What did NOT move (referee scope check): this is not a no-go over all readouts. The CKV theorem only constrains the readout "modular $=$ $T$-flux of a vector field." It leaves untouched non-geometric readouts — the Cao–Carroll mutual-information/MDS spatial metric, or a relative-entropy / algebraic-invariant readout — which iteration 3 already classified as closest-to-generate without a CKV (see 2026-06-08-iter3-encode-vs-generate-criterion.md, construction (d)). The "equivalent" claim is precise per route, not globally for the program. [corrected scope]

---

Furthest defensible claim

[INFERENCE] Under the stress-tensor-flux geometric readout ($K = 2\pi\int_\Sigma \xi^\mu T_{\mu\nu}\,d\Sigma^\nu$), any recovery of a semiclassical boost/Killing structure and horizon from modular data requires a conformal Killing vector (Sorce 2024; Caminiti et al. 2502.02633); and via the half-sided-modular-inclusion route the requisite positive translation is, by the Borchers–Wiesbrock characterization, logically equivalent to the energy-positivity that the half-sidedness assumption already encodes (Borchers 1992; Wiesbrock 1993). Therefore Borchers–Wiesbrock relocates rather than removes the geometric input, and OP-46 clause (iii) — geometry as output from a genuinely non-geometric (non-CKV) modular flow — is unsatisfiable-by-construction for this readout: the demanded output (a CKV/horizon) presupposes the symmetric structure whose absence defines the hypothesis. No $(\mathcal{A},\omega)$ pair is known, and under this readout none can in principle, turn "encode" into "generate."

Scope, stated explicitly: the universal "in principle" is supported only for the named (T-flux) readout. As a claim over all readouts it is unsupported by the cited theorems and conflicts with the iteration-3 Cao–Carroll closest-to-generate verdict — so it is deliberately not asserted globally. Confidence: high that no current construction generates Lorentzian spacetime from non-geometric modular flow; medium that the unsatisfiability is a true no-go rather than a limitation of the standard readout (it is conditional on that readout, and the two underlying theorems, though web-verified, are chained by an inference supplied here).

---

If it fails / weaker statement

• If a proponent rejects the standard readout — e.g. defines "geometry recovered" via an information metric / modular Berry curvature (Czech–Lamprou–McCandlish–Sully, arXiv:1712.07123; Czech et al. arXiv:2305.16384) or a Connes spectral distance, not via a stress-tensor CKV — then the Sorce gate (Gate 2) is bypassed and clause (iii) is no longer unsatisfiable. The burden then becomes sharply isolated: exhibit a non-symmetric $\omega$ whose modular-Berry / spectral data yields a Lorentzian metric. [OPEN — this is the genuinely live research target]
• Fallback weaker claim that still holds: even then, every existing such construction (Kabat–Lifschytz; modular-Berry; Connes reconstruction) presupposes either the vacuum/conformal algebra or recovers only a Riemannian spin$^c$ manifold. Connes' reconstruction theorem (arXiv:0810.2088, J. Noncommut. Geom. 7 (2013) 1–82) yields a closed Riemannian — not Lorentzian — manifold, so the Lorentzian-signature $+$ causal-order output (N4 of iteration 3) remains unachieved. The near-theorem then degrades gracefully to the iteration-2/3 survey verdict: positivity/symmetry input is empirically universal across all recovery routes, pending a non-CKV readout. [INFERENCE]
• Hardness, not forbiddenness (referee): Connes 0810.2088 closes the Riemannian case; a Lorentzian output is OPEN, not forbidden. Lorentzian / pseudo-Riemannian spectral-triple programs (Krein-space Dirac operators; Paschke–Verch; van den Dungen; Besnard–Bizi causality) make the second-stage burden unsolved-but-consistent, not a no-go. [OPEN]

---

Proposed registry items (with referee verdicts)

GAP-LL-ENERGY-POSITIVITY-R2 — (gap-refinement)

Statement. Refinement of GAP-LL-ENERGY-POSITIVITY (iter-3). The Leutheusser–Liu / Borchers–Wiesbrock positivity premise is not an input independent of the half-sidedness of the modular inclusion: by Wiesbrock's characterization (CMP 157 (1993) 83; converse to Borchers 1992, CMP 143, 315), a half-sided modular inclusion is equivalent to a one-parameter translation with positive generator obeying $\Delta^{it}U(a)\Delta^{-it}=U(e^{-2\pi t}a)$. Hence positivity and half-sidedness are one premise [ESTABLISHED], and a single HSMI yields only the $ax{+}b$ pair, not a full Killing/Möbius algebra (which needs $\ge2$ HSMIs) [ESTABLISHED]. Consequently the HSMI route does not supply causal structure from premise-free neutral algebraic data; it converts an assumed spectrum condition into the $ax{+}b$ / $PSL(2,\mathbb{R})$ group (Lashkari et al. 2412.19882).

Referee verdict: KEEP — severity MINOR — tag [INFERENCE] for the interpretive core (with a [CONTESTED] sub-element); the bare algebraic facts — Wiesbrock biconditional, single-HSMI $\to$ $ax{+}b$ — are [ESTABLISHED]. NOT [ESTABLISHED] as a whole. Corrections folded in: (1) attribute the equivalence to the Wiesbrock converse, not "Borchers' biconditional" — Borchers 1992 proves only the forward direction. (2) The phrase "does not GENERATE / converts not generates" over-reads: it flattens the iter-3 PARTIAL verdict (LL genuinely outputs interior time $+$ horizon location) into pure ENCODE; the iter-3 PARTIAL verdict is not overturned. (3) Whether $P\ge0$ is a "relocated geometric (boost/Killing) input" versus an independent positivity/stability axiom is [CONTESTED] (the iter-3 referee took the latter view). All three citations verified real.

HYP-CKV-VACUITY — (new-hypothesis)

Statement. Clause-(iii) self-undercutting (near-theorem). Under the standard geometric readout $K=2\pi\int_\Sigma\xi^\mu T_{\mu\nu}\,d\Sigma^\nu$, recovering a semiclassical boost/Killing structure $+$ horizon from modular data requires a future-directed conformal Killing vector (Sorce 2024, arXiv:2403.18937; Caminiti et al. 2502.02633; older input Arias et al. 1707.05375). A genuinely non-geometric modular flow furnishes no CKV. Therefore OP-46 clause (iii) — geometry as OUTPUT from a non-symmetric (non-CKV) state — is unsatisfiable-by-construction for this readout: the demanded output presupposes the symmetric structure whose absence defines the hypothesis. To escape, one must adopt a non-CKV readout (modular-Berry information metric, Czech et al.; or a Connes spectral distance), at which point the open burden is exhibiting a non-symmetric state whose such data yields a Lorentzian metric (Connes reconstruction, 0810.2088, yields only Riemannian).

Referee verdict: KEEP — severity MINOR — tag [INFERENCE] (resting on an [ESTABLISHED] lemma). Corrections folded in: (1) citation — "Arias 2016 / Sorce 2020" garbled; the CKV theorem is Sorce alone (arXiv:2403.18937); Arias et al. is arXiv:1707.05375 (2017, not 2016); Caminiti et al. = 2502.02633 (4 Feb 2025), real and already in the iter-4 dS note. (2) "VACUOUS" $\to$ "self-undercutting / unsatisfiable-by-construction for the geometric readout" — the clause has content, it is just unmeetable by that one readout. (3) "near-theorem" overstates status — the core (geometric readout $\Rightarrow$ CKV $\Rightarrow$ symmetric $\Rightarrow$ clause (iii) unmeetable) is a clean [INFERENCE] on an [ESTABLISHED] lemma, but the escape-route half (non-CKV readouts could yield geometry) is asserted, not shown; the item is inference $+$ open burden. (4) Margin note: Caminiti et al. show some non-vacuum states still have geometric (CKV) modular flow w.r.t. suitable subregions, so "non-symmetric $\Leftrightarrow$ non-geometric" is not exact.

GAP-NCG-LORENTZIAN-SIGNATURE — (gap-refinement)

Statement. Connes' reconstruction theorem (A. Connes, "On the spectral characterization of manifolds," arXiv:0810.2088; J. Noncommut. Geom. 7 (2013) 1–82) reconstructs, from a commutative spectral triple satisfying the five axioms, a closed oriented smooth Riemannian manifold — positive-definite/Euclidean signature (spin$^c$ in general; spin once the real-structure axioms / charge conjugation $J$ with the KO-dimension sign table are imposed). It does not produce Lorentzian signature or a causal order. Hence the spectral-triple / NCG route, like Cao–Carroll's MDS embedding, fails OP-46 necessary-condition N4 (Lorentzian signature $+$ causal order) at the signature step. Obtaining Lorentzian/pseudo-Riemannian geometry from spectral data (Krein-space Dirac operators, Lorentzian spectral triples) is a separate, unfinished program with no established analogue of the reconstruction theorem.

Referee verdict: KEEP — severity MINOR — tag [ESTABLISHED] for the Euclidean-output and citation claims; [OPEN] for the Lorentzian program. Only fix: replace "spin manifold" with "spin$^c$ (spin, if the real structure $J$ is included) manifold" — the bare commutative theorem delivers spin$^c$; a genuine spin structure additionally needs the real/KO-dimension axioms. The signature conclusion is unaffected: Riemannian either way, so N4 fails at the signature step regardless. Citation (arXiv:0810.2088, 2013) correctly cites the journal year.

---

Verdict

PROMOTED toward a no-go for clause (iii), short of a full theorem. The deepest reorganization-vs-generation test sharpens decisively per route: there is NO known $(\mathcal{A},\omega)$ pair with non-geometric modular flow that outputs a semiclassical Killing/boost $+$ horizon, and — stronger than iteration 3 — under the standard geometric (stress-tensor-flux) readout this is unsatisfiable-by-construction, because (a) the Sorce / Caminiti theorem forces any recovered boost generator to be a conformal Killing vector (hence to presuppose a symmetric background), and (b) the Borchers–Wiesbrock characterization makes the Borchers–Wiesbrock "generation" of a positive translation logically equivalent to the energy-positivity that the half-sided inclusion already assumes.

Borchers–Wiesbrock and the 2024 Lashkari et al. $PSL(2,\mathbb{R})$ result are genuine generation of a symmetry GROUP from positivity — but positivity is the relocated geometric premise, so they do not turn "encode" into "generate" for spacetime. Kabat–Lifschytz recovers the bulk metric only up to a conformal factor from the symmetric CFT vacuum (recovery, $g_3$ true). Connes' reconstruction yields only a Riemannian spin$^c$ manifold (fails N4).

Program A status (the modular / crossed-product program, 2026-06-08-algebraic-background-independence.md): it remains a rigorous reorganization that ENCODES geometry. The single escape hatch is a non-CKV readout (modular Berry / spectral distance) applied to a genuinely non-symmetric state — currently unrealized, and even if realized must still cross the Lorentzian-signature barrier (Connes gives only Riemannian).

Confidence: HIGH that no current construction generates Lorentzian spacetime from non-geometric modular flow; MEDIUM that the unsatisfiability is a true no-go rather than a limitation of the standard readout (it is conditional on that readout, and the two underlying theorems, though web-verified, are chained by an inference supplied here).

This upgrades iteration-3 open subquestion #3 from [OPEN] to a conditional near-theorem (unsatisfiable under the standard readout) with a precisely-named escape hatch (non-CKV readout) that defines the genuinely live research target. It is the modular-flow counterpart of the iteration-4 dS finding (2026-06-08-iter4-dS-firstlaw-SO-family-attempt.md): geometric modular content exists "only where geometry is pre-installed via a conformal/Killing symmetry."

---

Open subquestions

1. Is there a non-CKV readout (modular Berry curvature / quantum information metric, Czech et al. 2305.16384; or a Lorentzian Connes spectral distance) under which a genuinely non-symmetric state's modular data yields a Lorentzian metric — escaping the Sorce-gate? This is the precisely-located live target for clause (iii). [OPEN]
2. Can the Borchers–Wiesbrock characterization be turned into a sharp theorem: "any algebraic derivation of a positive-energy translation from modular-inclusion data is equivalent to assuming the spectrum condition," closing the relocation loophole formally rather than by inference? [OPEN]
3. Does a Lorentzian / pseudo-Riemannian spectral triple (Krein-space Dirac operator) admit a reconstruction theorem analogous to Connes' Riemannian one, and if so does it require a presupposed causal/time-orientation input (an analog of $g_4$)? [OPEN]
4. For the emergent Type III$_1$ at large $N$ (Leutheusser–Liu): is the $N=\infty$ positivity an independent stability axiom or fully inherited from boundary-Hamiltonian boundedness — i.e. is the horizon an emergent OUTPUT or an imprint of the assumed bounded-below boundary spectrum? [OPEN]
5. Does Connes–Rovelli thermal time (Class. Quantum Grav. 11, 2899 (1994), arXiv:gr-qc/9406019) plus the state-independent Connes-cocycle (Radon–Nikodym) flow ever upgrade from a posited time direction to a metric/causal order without adding a factorization or symmetry input — i.e. can it pass N4 unaided? [OPEN]

---

See also

• 2026-06-08-iter3-encode-vs-generate-criterion.md — iteration-3 parent; OP-46 criterion, clause (iii), the six-construction classification, GAP-LL-ENERGY-POSITIVITY this refines.
• 2026-06-08-iter4-dS-firstlaw-SO-family-attempt.md — sibling iteration-4 attempt; "geometric flow only where a conformal/Killing symmetry is pre-installed" (2502.02633) in dS dress.
• 2026-06-08-algebraic-background-independence.md — Program A; "rigorous reorganization, not derivation"; CLPW crossed product; relocating-not-removing input.
• 2026-06-08-iter3-synthesis.md — cross-program synthesis; encode-vs-generate meta-verdict.
• ../OPEN_PROBLEMS.md — OP-46 (this track); clause (iii) status update.
• ../GAPS_AND_CONTRADICTIONS.md — GAP-LL-ENERGY-POSITIVITY (iter-3) and -R2 (here); GAP-NCG-LORENTZIAN-SIGNATURE.
• ../HYPOTHESES.md — H4 (geometry/time from algebra); HYP-CKV-VACUITY proposed here.
• ../domains/mathematics.md — Tomita–Takesaki, Type III$_1$, Borchers–Wiesbrock, Connes reconstruction, spectral triples.
• ../domains/quantum-field-theory.md — local algebras, Bisognano–Wichmann, Reeh–Schlieder, CHM modular Hamiltonian.
• ../domains/general-relativity.md — conformal Killing vectors, causal order $\to[g]$ (Malament/HKM), horizons.
• ../EPISTEMICS.md — epistemic tag definitions used throughout.

Key results cited

• Sorce, "Analyticity and the Unruh effect: a study of local modular flow," arXiv:2403.18937 (2024) — sole author; geometric modular flow $\Rightarrow$ future-directed conformal Killing vector; non-isometry $\Rightarrow T^\mu{}_\mu=0$ (CFT). [VERIFIED — corrects the garbled "Arias–Sorce" attribution]
• Caminiti, Capeccia, Ciambelli, Myers, "Geometric modular flows in 2d CFT and beyond," arXiv:2502.02633, JHEP 08 (2025) 166, eq. (8) — explicit 2d realization; geometric modular flow implements a conformal symmetry of the background. [VERIFIED]
• Arias, Casini, Huerta, Pontello, "Anisotropic Unruh temperatures," arXiv:1707.05375 (2017; Inspire key Arias:2016nip) — local-modular input; not the CKV theorem itself. [VERIFIED — year is 2017, not 2016]
• Borchers, CMP 143, 315 (1992); review "On revolutionizing quantum field theory with Tomita's modular theory," J. Math. Phys. 41, 3604 (2000) — spectrum condition $\Rightarrow$ modular commutation relations (forward direction). [VERIFIED]
• Wiesbrock, "Half-sided modular inclusions of von-Neumann-algebras," CMP 157, 83 (1993); gap filled Araki–Zsidó (2004), arXiv:math/0412061 — HSMI $\Leftrightarrow$ positive-generator translation (the converse closing the biconditional). [VERIFIED]
• Lashkari, Leung, Moosa, Ouseph, "Modular Intersections, Time Interval Algebras and Emergent AdS$_2$," arXiv:2412.19882, JHEP 10 (2025) 153 — (twisted) modular inclusion/intersection $\Rightarrow$ universal cover of $PSL(2,\mathbb{R})$; (0+1)d conformal GFF in the vacuum (positivity implicit). [VERIFIED]
• Kabat, Lifschytz, "Emergence of spacetime from the algebra of total modular Hamiltonians," arXiv:1812.02915 (2019) — bulk point $=$ weakly-maximal Lie subalgebra; metric up to a conformal factor; input $=$ CFT vacuum $+$ conformal algebra. [VERIFIED — recovery, presupposes symmetric vacuum]
• Leutheusser, Liu, "Causal connectability...," arXiv:2110.05497; "Emergent times in holographic duality," arXiv:2112.12156, Phys. Rev. D 108, 086020 (2023) — emergent interior time $+$ sharp horizon at $N=\infty$; positivity from boundedness-below of the boundary CFT Hamiltonian. [VERIFIED]
• Connes, "On the spectral characterization of manifolds," arXiv:0810.2088 (2008), J. Noncommut. Geom. 7 (2013) 1–82 — commutative spectral triple $\Rightarrow$ closed Riemannian spin$^c$ manifold (signature Euclidean, not Lorentzian). [VERIFIED — relevant to N4 failure of the NCG route]
• Connes, Rovelli, "Von Neumann algebra automorphisms and time–thermodynamics relation," Class. Quantum Grav. 11, 2899 (1994), arXiv:gr-qc/9406019 — thermal-time hypothesis; a posited time direction only, no metric/dimension/causal order (fails N4). [VERIFIED]
• Czech, Lamprou, McCandlish, Sully, arXiv:1712.07123; Czech et al., arXiv:2305.16384 — modular Berry connection / curvature; candidate non-CKV readout. [VERIFIED — escape-hatch input]