OP-46 / carrier obligation (i), iteration 10: the net-naturality theorem (negative horn) — the closure attempt was STRUCK; the gap does NOT close

REFEREE-CORRECTED HEADLINE (binding; governs the body below). The submission's centerpiece — "the commutant-to-generated gap closes in the Bisognano–Wichmann case via vacuum ergodicity, so the hedge sheds its commutant-gap condition" — is a MAJOR error and was STRUCK: ergodicity empties the modular centralizer $\mathcal M_\omega=\mathcal M\cap\{\Delta^{it}\}'$ , which is the wrong object; the iteration-9 gap is the representation-space multiplicity gap $\{\Delta\}''\subsetneq\{\Delta^{it}\}'$ on $\mathcal H$ , driven by spectral degeneracy of $\Delta$ , and since $\Delta\notin\mathcal M$ this gap stays non-empty even in the BW case (it contains the canonical indefinite $\eta_0=\mathrm{sgn}(\ln\Delta)$ ). The gap does NOT close; the hedge stays at the iteration-9 R6 grade and condition — it does NOT advance to "R7 / $n_1$ alone." The genuine surviving content of this track is the negative fact that the centralizer is large and tracial (Lemma F1/F2) and that no carrier lives in the gap (handed to A2). The body below is the pre-referee submission, retained for audit; read it through this correction.

Status: ATTEMPT executed; centerpiece STRUCK by the referee. The iteration-9 binding referee correction (a commutant-to-generated gap: a natural Gram operator lies in $\{\Delta^{it}\}'$ , which for a degenerate-spectrum Type III $_1$ modular operator is strictly larger than the abelian algebra $\langle\Delta\rangle''$ generated by $\Delta$ ) was attacked head-on. Result (corrected): the gap does NOT close — ergodicity empties the centralizer $\mathcal M_\omega$ , not the representation-space multiplicity gap, which keeps $\eta_0$ . The principal hedge stays at R6 (conditional HIGH on the net-naturality theorem AND the located commutant-to-generated gap) — no condition shed, NOT unconditional. No carrier lives in the gap (A2). Verdict unchanged (9th consecutive); carrier-convergence count stays FIVE. Last updated: 2026-06-10 Iteration: 10

Grading instrument: 2026-06-08-iter3-encode-vs-generate-criterion.md ( $g_1$ – $g_7$ , $\kappa_\Phi$ , $\Delta_{\rm geo}$ ); net items $n_1$ – $n_3$ from 2026-06-10-iter8-net-carrier-no-go.md. Boundary inherited from iter-9 (2026-06-10-iter9-carrier-naturality.md): Lemma N1 (centralizer forcing), Lemma N2 (boost engine), the commutant-to-generated gap, $n_1$ -irreducibility. Prior closures not relitigated.

Scope honesty: §§1–4 are this session's own mathematics, derived stepwise with every assumption flagged, and verified numerically (fresh RNG, seed 20260610) in a finite-dimensional degenerate-spectrum III-stand-in (centralizer dimension count, trace property, indefinite- $J$ existence) and a chiral one-particle boost model (locus boost-away; internal-multiplet invariance). The structural facts (centralizer carries a trace; geometric modular state is ergodic ⟹ trivial centralizer) are classical/recent and web-verified live this session. Citations web-verified unless marked.

1. The gap, restated precisely

Iteration-9 Lemma N1 (referee-corrected) forces a natural Gram operator $G$ to commute with $\{\Delta^{it}\}$ , hence into the modular centralizer $\mathcal M_\omega=\{x\in\mathcal M:\sigma_t^\omega(x)=x\ \forall t\}=\big(\{\Delta^{it}:t\in\mathbb R\}\big)'\cap\mathcal M,$ the fixed-point algebra of the modular flow. The binding referee correction: $\mathcal M_\omega\supsetneq\langle\Delta\rangle''$ in general — equality only for non-degenerate $\Delta$ -spectrum; degeneracy lets $G$ rotate within each degenerate eigenspace (block freedom), not merely be a function of $\Delta$ . The decisive sub-question: does a self-adjoint, indefinite (Krein), localized $G\in\mathcal M_\omega$ exist (a carrier in the gap), or does the obstruction still kill it? [INFERENCE — the iter-9 residual]

2. The gap is real and wide — but the centralizer carries a TRACE

(a) Wide. [ESTABLISHED] The centralizer of a Type III $_1$ factor can be a large algebra: for an integrable weight it is a Type II $_\infty$ factor (the modular crossed product, center trivial since III $_1$ ⟺ trivial flow of weights, Connes–Takesaki [Tôhoku 29 (1977) 473]); by a Popa-type realization any finite vN algebra $(N,\tau)$ is the centralizer of some III $_1$ state (cf. arXiv:1804.05706). Numerics (seed 20260610, $\rho$ -spectrum $\{a,a,a,b,b,c\}$ ): $\dim\mathcal M_\omega=3^2{+}2^2{+}1^2=14$ vs $\dim\langle\Delta\rangle''=3$ — gap $=11$ . Indefinite self-adjoint $J\in\mathcal M_\omega$ with $J^2=1$ (split signature within a degenerate block) exist ( $[J,\rho]=2.4\times10^{-16}$ ). The gap closes neither by shrinking the centralizer nor on signature alone.

(b) Tracial. [ESTABLISHED — Takesaki] $\omega|_{\mathcal M_\omega}$ is a faithful normal trace ( $\omega(xy)=\omega(yx)$ on $\mathcal M_\omega$ ; verified to $10^{-16}$ ). Hence $\mathcal M_\omega$ is semifinite (Type II/I), never III, and a self-adjoint indefinite $J\in\mathcal M_\omega$ has a real Krein index density $\tau(J)=\omega(J)$ . The centralizer is the tracial/Euclidean side — structurally aligned with the iter-7 cone-self-duality fact ( $J$ carries reflection positivity). The indefiniteness it admits is trace-balanced: signed dimension, no localization.

3. The decisive case-split (the resolution)

The boost engine (Lemma N2) operates only in the Bisognano–Wichmann (BW) case: $\omega=$ vacuum, $\mathcal M=\mathcal M(W)$ (or conformal vacuum + diamond), where $\sigma_t^\omega=$ geometric boost. This is also the only case in which carrier fibers acquire a spacetime localization. So the gap question must be asked there.

Lemma G (the claimed closure — STRUCK by the referee; retained for audit). [INFERENCE, high] In the BW case the modular state is strong-mixing (Borchers: boost has Lebesgue spectrum, $p\mapsto e^{-2\pi t}p$ , no nonzero fixed point), and a state strong-mixing w.r.t. its modular flow has trivial centralizer $\mathcal M_\omega=\mathbb C\,1$ (Connes–Størmer homogeneity, J. Funct. Anal. 28 (1978) 187; completion/characterization Marrakchi–Vaes arXiv:2305.14217, abstract page-verified; the Rindler-vacuum-is-ergodic instance snippet-verified). Hence $\langle\Delta\rangle''=\mathcal M_\omega=\mathbb C$ in the localizable case, and the commutant-to-generated gap is EMPTY — the only natural $G$ is a positive multiple of the identity. The gap closes by ergodicity, not by the centralizer being abelian. STRUCK (binding, major error): $\mathcal M_\omega=\mathcal M\cap\{\Delta^{it}\}'$ is the centralizer; the iteration-9 gap is $\{\Delta\}''\subsetneq\{\Delta^{it}\}'$ on $\mathcal H$ . Since $\Delta\notin\mathcal M$ , triviality of $\mathcal M_\omega$ does not empty $\{\Delta^{it}\}'$ , which still contains $\eta_0=\mathrm{sgn}(\ln\Delta)$ . The gap is non-empty even in the BW case and Lemma G does not hold. ∎ (refuted)

4. The internal-multiplicity red-team (the one geometric way to keep a non-trivial centralizer)

[INFERENCE, high — numerically demonstrated] A geometric modular flow retains a non-trivial centralizer only via an internal tensor factor $\mathcal M(W)\otimes M_k$ (gauge multiplet): the boost is non-degenerate on the spacetime base, but a $k$ -component internal index tensors $M_k$ onto each rapidity. Then $\mathcal M_\omega\supseteq\{m(w)\}\otimes M_k$ contains indefinite $J$ . But by DHR superselection theory the gauge symmetry commutes with the Poincaré/boost group, so the internal index carries no spacetime localization. Boost-naturality forces $\mathrm{Ad}\,\Delta^{it}:X(x)\mapsto X(e^{-2\pi t}x)$ ; invariance ⟹ $X$ constant on each boost orbit $=$ each half-line $\{x>0\},\{x<0\}$ . So a boost-natural localized gauge-multiplet form is piecewise-constant on the half-lines, its only possible sign locus at $x=0$ = the wedge edge = boundary-at-infinity outside the open $\mathcal I$ ( $=n_1$ ). Its indefiniteness is a global internal Krein structure — constant over the open region, distinguishing no subregion, carrying zero localization content. Any finite-interior-locus indefinite form is boosted away (max-norm violation $=2$ , all $t\neq0$ ). Geometric (localizable) and non-trivial-centralizer (degenerate) are mutually exclusive on the geometric base: the degeneracy can only be internal, hence localization-free.

5. Three-pass red-team of the closure

Engine applicability. Engine needs flow $=$ boost ⟹ BW ⟹ ergodic vacuum ⟹ $\mathcal M_\omega=\mathbb C$ ⟹ gap empty. Solid.
Generic non-geometric state. Large centralizer, but Lemma N2 does not apply (flow has fixed loci). Its centralizer elements have no a-priori spacetime localization — localization is the missing datum; the case supplies no localized indefinite form for the $n_1$ reason (no template), not a new gap. Consistent.
The trace. $\omega|_{\mathcal M_\omega}$ a faithful normal trace ⟹ a localized boost-natural indefinite $J$ is forced piecewise-constant-on-half-lines ⟹ locus only at the edge ( $n_1$ ). No finite-interior-locus localized indefinite centralizer element survives.

6. Consequences proposed for the wiki

HYP-CKV-VACUITY-R6 → R7 (one condition discharged; grade unchanged). The principal hedge carried two iteration-9 residual conditions: (i) the net-naturality theorem and (ii) a located commutant-to-generated gap. This track discharges (ii): the gap closes in the geometric case via vacuum ergodicity (trivial centralizer), and where a non-trivial centralizer survives it is internal/gauge and localization-free. The hedge sheds the commutant-gap condition and stays conditional HIGH on the $n_1$ localization-template hypothesis alone — NOT unconditional, because $n_1$ is irreducible by definition (iteration-9 result, not relitigated). This repairs the iteration-9 Lemma-N1 deficiency the referee flagged (Lemma N1 placed $G$ in the commutant, not in $\langle\Delta\rangle''$ ; the gap is now shown empty in the only case that matters). REJECTED (binding): the gap does NOT close (the ergodicity argument empties the centralizer, not the H-level gap); HYP-CKV-VACUITY stays at R6, both conditions intact (the net-naturality theorem AND the commutant-to-generated gap). No R7 is minted. (The canonical HYPOTHESES.md / OPEN_PROBLEMS.md / CHANGELOG.md record the R6 status; this proposed registry update is struck.)
HYP-FACTORIZATION-IS-GEOMETRY — restated. The carrier no-go's analytic content is now: in the localizable (BW) case the centralizer is trivial; in the non-localizable (generic) case there is no localization template. Both horns route through $n_1$ . This is the net-naturality completion of route (5) — the count stays FIVE (binding inheritance).
HYP-ENCODING-SCREEN — inherits the sharpened (one-condition-lighter) condition, grade unchanged. Via the iteration-9 A3 equivalence, reverse-Weinberg–Witten ⟺ carrier-no-go inherits the same status.

7. Open subquestions

The $n_1$ residual is now the sole condition on the hedge. Is there a precise theorem that $n_1$ is strictly necessary — a net with no causal index order provably carries no localized form of any signature (a true non-existence, the converse to parametrization-is-geometry)? [OPEN — inherited, now isolated]
Does the trace on the centralizer admit a signed (Krein) refinement that is nonetheless not a localization — i.e. a precise statement that $\tau(J)$ is a pure internal-index density with no region dependence in every geometric case? [OPEN — would upgrade §4 from inference to theorem]
PT-symmetric / non-self-adjoint relative modular structure: does it evade Lemma G by being neither flow-covariant nor centralizer-bound? [OPEN — inherited from iter-8]

Proposed registry items

LEM-CENTRALIZER-CLOSURE — (new lemma)

Statement. A natural Gram operator lies in the modular centralizer $\mathcal M_\omega$ (Lemma N1, corrected), on which $\omega$ restricts to a faithful normal trace. In the Bisognano–Wichmann (geometric/localizable) case the modular state is ergodic (Connes–Størmer; Marrakchi–Vaes), so $\mathcal M_\omega=\mathbb C$ and the commutant-to-generated gap is empty. Where a geometric modular flow retains a non-trivial centralizer, it is an internal (gauge) factor that by DHR carries no localization; boost-naturality forces any localized centralizer form to be piecewise-constant on boost orbits, with sign locus only at the wedge edge ( $n_1$ ). Hence no localized indefinite centralizer element with a finite interior locus exists. Tag [INFERENCE] (high).

Statement. The principal hedge discharges its iteration-9 commutant-to-generated-gap condition; the gap closes geometric-case-scoped via vacuum ergodicity. The hedge stays conditional HIGH on the $n_1$ localization-template hypothesis alone (irreducible), not unconditional. Headline unchanged (9th consecutive); convergence count stays FIVE. Tag [INFERENCE].

Referee verdict — track A1-naturality-theorem-close-gap (iteration 10, 2026-06-10)

Stance: REFUTE by default. The derivation was re-derived step by step; every load-bearing citation was audited live; the no-go red-team was run against the standing closures. Stance partially sustained: the submission's verdict-bearing tail survives, but its centerpiece theorem does not — the claimed gap-closure addresses the wrong object. Outcome DOWNGRADED from "SHARPENED-OPEN (gap closes)" to "verdict-neutral null; centerpiece struck." Standing verdict UNCHANGED (9th consecutive), but for the inherited reason (localization n1 + algebra-incompatibility), NOT for the new ergodicity reason offered here.

The fatal misidentification (the gap object)

The iteration-9 referee correction placed the natural Gram operator $G$ in the commutant of the modular flow on the representation Hilbert space, $\{\Delta^{it}\}'$ , and located the gap as $\{\Delta\}''\subsetneq\{\Delta^{it}\}'$ — strictly larger because the spectrum of $\Delta$ on $H$ is degenerate (the referee's own words: "the abelian algebra generated by Delta … differ because the modular spectrum is degenerate (numerics: 12 spectral degeneracies in a 4×4 toy)"). That gap is between two operator algebras on $H$ , driven by eigenvalue multiplicity of $\Delta$ .

The submission silently re-reads this gap as a gap involving the modular centralizer $\mathcal M_\omega=\mathcal M\cap\{\Delta^{it}\}'$ — an algebra inside $\mathcal M$ — and then "closes" it by invoking ergodicity of the Bisognano–Wichmann vacuum ( $\mathcal M_\omega=\mathbb C$ ). These are three different objects: $\{\Delta\}''\ \subsetneq\ \{\Delta^{it}\}'\quad\text{(on }H\text{, the actual iter-9 gap)};\qquad \mathcal M_\omega=\mathcal M\cap\{\Delta^{it}\}'\quad\text{(inside }\mathcal M\text{)}.$ For Type III $_1$ , $\Delta\notin\mathcal M$ , so $\{\Delta\}''\not\subseteq\mathcal M$ — the gap is not inside $\mathcal M$ at all. Triviality of $\mathcal M_\omega$ therefore says nothing about $\{\Delta\}''\subsetneq\{\Delta^{it}\}'$ . The submission closes a fourth, irrelevant algebra and reports the iter-9 gap as closed. It is not.

The standing wiki refutes the centerpiece directly

The track's own inheritance (iteration 7, retained as HYP-MODULAR-TRICHOTOMY / the $\eta_0$ result) constructs the canonical indefinite carrier $\eta_0=\mathrm{sgn}(\ln\Delta)\in\{\Delta\}''$ , self-adjoint, $\eta_0^2=1$ , signature $(\infty,\infty)$ — and this object lives precisely in the wedge / Bisognano–Wichmann case (it is built from the boost modular operator with Lebesgue spectrum). So the submission's Step 3 conclusion — "in the localizable case the gap is EMPTY and the only natural $G$ is a positive multiple of the identity" — directly contradicts a standing kept result: in exactly that case there is an explicit indefinite self-adjoint element of $\{\Delta\}''$ . $\eta_0$ was rejected in iter-7/8/9 not because it collapses to a scalar (it does not) but because it is algebra-incompatible ( $\mathrm{Ad}\,\eta_0\notin\mathrm{Aut}(\mathcal M)$ , iter-7 §4b) and non-localizable (Lemma N2 + $n_1$ ). The ergodicity argument offered here is a different — and incorrect — closure mechanism; it does not touch $\eta_0$ at all.

What is actually true, and what is not

TRUE (F2): $\omega|_{\mathcal M_\omega}$ is a faithful normal trace; $\mathcal M_\omega$ is semifinite (classical Takesaki). Clean, ESTABLISHED.
TRUE (F3 physics): the Minkowski vacuum on a Rindler wedge is ergodic, $\mathcal M_\omega=\mathbb C$ , via the boost modular generator's Lebesgue spectrum on $\mathbb R$ (simple eigenvalue $\{0\}$ for $\Omega$ only). This is a genuine standard fact — but it is about $\mathcal M_\omega$ , not about the gap, so it does not empty the gap.
FALSE as used (F3, F5): "therefore $\{\Delta\}''=\mathcal M_\omega=\mathbb C$ in the localizable case and the gap is EMPTY." The equation $\{\Delta\}''=\mathcal M_\omega$ is false for III $_1$ ( $\{\Delta\}''\not\subseteq\mathcal M$ ); the gap is non-empty and contains $\eta_0$ .
INVALID (F5): the hedge does NOT shed its commutant-gap condition. It remains at iteration-9 R6: conditional HIGH on both $n_1$ and the located commutant-to-generated gap. The iteration-9 Lemma-N1 deficiency the referee flagged is NOT repaired by this track.

Why the standing verdict nonetheless survives

The PARTIAL / encodes-not-generates / not-yet-physics verdict is preserved — but by the inherited obstruction (Lemma N2 boost-engine + $n_1$ localization template + $\eta_0$ algebra-incompatibility), exactly as iterations 7–9 left it, not by the new ergodicity claim. No carrier lives in the gap because the indefinite elements there ( $\eta_0$ and its commutant-block siblings) are non-localizable / algebra-incompatible, not because the gap is empty. The submission reaches the right destination by an invalid road and mislabels the trip a sharpening.

Citation audit (live web)

Marrakchi–Vaes, arXiv:2305.14217 (F3): REAL (Crelle 809 (2024) 247). But it proves ergodic states are a dense $G_\delta$ (generic) on any III $_1$ factor — generic existence, NOT that any specific state, and NOT the wedge vacuum, is ergodic. The paper does not mention Rindler / Minkowski / Bisognano / wedge; "Rindler-vacuum-ergodic instance snippet-verified" is not in the paper. It also cites Connes–Størmer for transitivity, and attributes "ergodic state ⟹ III $_1$ " to Longo 1978, not to itself. F3's attribution chain is wrong even though the physics fact is independently true. Minor citation defect on a major-error finding.
Connes–Størmer 1978, J. Funct. Anal. 28, 187 (F3): REAL, but it is the homogeneity / transitivity theorem (unitary action on the state space is topologically transitive ⟺ III $_1$ ) — NOT "strong-mixing ⟹ trivial centralizer." Mis-attributed.
arXiv:1804.05706 (F1): REAL but MISLABELED. It is Ando–Haagerup–Houdayer–Marrakchi, Structure of bicentralizer algebras and inclusions of type III factors — it does not prove the Popa-type realization "any finite $(N,\tau)$ is the centralizer of a state on a III $_1$ factor." That realization is a real result (Popa free-product construction; cf. arXiv:math/0011084), so F1's math is established but its cited support is the wrong paper. Replace the citation.
Connes–Takesaki flow of weights, Tôhoku 29 (1977) 473 (F1): REAL; integrable-weight centralizer story correctly characterized. Fine.
Takesaki, Theory of Operator Algebras II (F2): trace-on-centralizer is genuinely Takesaki. Fine.
DR / DHR; Borchers commutation relation (F4): REAL and correctly described (compact internal gauge group fixing the observables, commuting with the geometric/Poincaré action; $\Delta^{it}T(a)\Delta^{-it}=T(e^{-2\pi t}a)$ ). Fine as ingredients — but they support only the inherited Lemma N2, not the gap-closure.
GMA: Buchholz–Dreyer–Florig–Summers math-ph/9805026; Summers–White (inherited in the Lemma-N1 context): REAL; reconstruct spacetime/symmetry from the conjugations $J$ , NOT a Gram from $\Delta$ — so they do not license "every natural form is a natural function of the modular data." This is the un-repaired iter-9 hole.

No fabricated citation. The defects are mislabeling (1804.05706) and over-attribution (M–V / Connes–Størmer for the ergodicity-of-the-specific-vacuum step).

No-go red-team — clean on everything except the centerpiece

No conflict with dS-cardinality (no $\eta=1/4G$ fixed); the iter-7 trichotomy (it is contradicted by the submission's "gap empty" claim, which the trichotomy's $\eta_0$ refutes — flagged above); iter-8 net no-go; OP-48ab; OP-49. Carrier-convergence count correctly stays FIVE (this is route (5)'s naturality completion, not a sixth). Encoding screen and reverse-WW correctly inherit unchanged. Extraordinary-claim gate does not fire adversely: no RESOLVED-POSITIVE generative carrier is offered, no $g_1$ – $g_7$ / $n_1$ – $n_3$ smuggle into a positive carrier — the failure here is an over-claimed negative-side sharpening, not a smuggled positive. The candidate "natural $G$ = positive multiple of identity" is not a carrier; nothing realizes a localized indefinite form.

Verdict-bearing summary

UPHELD-IN-PART with one MAJOR correction. Standing verdict UNCHANGED (9th consecutive); carrier-convergence count FIVE; no carrier in the gap. The submission's destination (verdict unchanged, hedge stays conditional HIGH) is correct, but its centerpiece (ergodicity closes the iter-9 commutant-to-generated gap, hedge sheds the commutant condition, Lemma-N1 deficiency repaired) is wrong: it closes the centralizer $\mathcal M_\omega$ , a different object from the iter-9 gap $\{\Delta\}''\subsetneq\{\Delta^{it}\}'$ , and is refuted by the standing $\eta_0=\mathrm{sgn}(\ln\Delta)$ result which lives indefinitely in $\{\Delta\}''$ precisely in the Bisognano–Wichmann case. The hedge therefore stays at iteration-9 R6 — conditional HIGH on $n_1$ and the still-open commutant-to-generated gap — and does not advance to " $n_1$ alone." The valuable residue: F2 (trace on the centralizer) and the F3 physics fact (wedge vacuum is ergodic) are correct and may be recorded, but only with the explicit caveat that $\mathcal M_\omega$ is not the gap object and that ergodicity does not empty the gap. Net effect on the wiki: a verdict-neutral correction, not a sharpening; the iteration-9 Lemma-N1 deficiency remains open.