OP-46 / carrier obligation, iteration 10: the positive dual executed — the commutant-to-generated gap is geometry-void too

Status: ATTEMPT executed — the one place the iter-9 referee left open (the commutant-to-generated gap: a natural carrier's Gram need only lie in $\{\Delta\}'$, which for degenerate Type-III$_1$ spectrum strictly exceeds $W^*(\Delta)$) was attacked constructively, as a headline-flip attempt. The positive dual: build a generative carrier $\mathrm{sgn}(h)$ from a self-adjoint modular-centralizer element $h\in M_\omega$ that is not a function of $\Delta$. Surprise: the object exists and clears two bars $\eta_0$ never cleared (indefinite AND algebra-compatible) — and the verdict still does not flip. The candidate forks on the modular spectrum and both prongs are closed: in the geometric (Lebesgue) sector the centralizer is trivial so the gap is empty; in the almost-periodic sector where $h$ exists, $\mathrm{sgn}(h)$ is unitarily universal ($\Delta_{\rm geo}=0$) and non-localizable without smuggling $n_1$. Combined with track A1 this closes the commutant-to-generated residual; the principal hedge narrows to $n_1$ alone. 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 ($g_1$–$g_7$, $n_1$–$n_3$, $\kappa_\Phi$, $\Delta_{\rm geo}$). Boundary inherited from 2026-06-10-iter7-indefinite-pairing-attempt.md ($\eta_0=\mathrm{sgn}(\ln\Delta)$, vacuity, trichotomy) and 2026-06-10-iter9-carrier-naturality.md (the commutant-to-generated gap = the sole residual). 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 degenerate-spectrum standard form. The centralizer-triviality characterization rests on Boutonnet–Houdayer and Houdayer, web-verified live. Citations web-verified this session unless marked unverified.

---

1. The gap, and the object that lives in it

The iter-9 referee's binding correction: modular equivariance forces $[h,\Delta^{it}]=0$, placing the Gram $h$ in the commutant $\{\Delta^{it}\}'$, not in $W^*(\Delta)$ (functions of $\Delta$, where $\eta_0$ lives). For Type III$_1$ with degenerate modular spectrum these differ. The algebra-side of the gap is the modular centralizer $M_\omega=M\cap\{\Delta\}'$ (the $\sigma_t$-fixed-point algebra).

Construction `ESTABLISHED — numerically confirmed. In a standard form with $\rho=\mathrm{diag}(1,2,2,4)/9$ on $M_4$, $\Delta=\rho\otimes\rho^{-1}$ has degenerate spectrum (multiplicities $(1,4,6,4,1)$). Take $h=L(a)$ with $a=a^*$ acting non-scalarly on the degenerate eigenspace of $\rho$ ($[a,\rho]=0$, e.g. $\mathrm{PauliX}$ on the doubled block). Then $[h,\Delta]=[h,\ln\Delta]=0$ to machine precision, yet $h\notin W^*(\Delta)$ — it acts non-scalarly on a $\Delta$-eigenspace, which no function of $\Delta$ can. This is exactly the gap object the referee identified.

2. Two bars cleared that $\eta_0$ never cleared

For full-rank $a_2=\mathrm{diag}(5,0,0,-5)+\mathrm{PauliX}_{\{1,2\}}\in M_\omega$:

• (i) indefinite ✓ — $\mathrm{sgn}(h)=L(\mathrm{sgn}(a_2))$ has signature $(8,8,0)$, a genuine non-degenerate Krein fundamental symmetry.
• (iv) algebra-compatible ✓ (trivially) — $\mathrm{sgn}(h)\in M$, so $\mathrm{Ad}(\mathrm{sgn}\,h)$ is an inner automorphism of $M$. Contrast iter-7 §4b: $\eta_0=\mathrm{sgn}(\ln\Delta)$ failed algebra-compatibility. The gap candidate is strictly stronger.

For a moment the headline-flip looks live: an indefinite, algebra-compatible, modular-equivariant carrier not reducible to $\eta_0$.

3. The fork (the pivot), and why both prongs are void

A centralizer element not-a-function-of-$\Delta$ exists iff $\Delta$ has a discrete-spectrum component — a modular eigenvector. This forks the problem.

Prong 1 — geometric sector: no localized indefinite carrier `INFERENCE, high. A vacuum modular flow of a wedge/interval is geometric ⟹ homogeneous Lebesgue spectrum (Borchers + positive-energy irrep uniqueness, iter-7 §4a) ⟹ no modular eigenvectors ⟹ the centralizer $\mathcal M_\omega=\mathcal M\cap\{\Delta^{it}\}'=\mathbb{C}1$ (Boutonnet–Houdayer arXiv:1602.01741, Anal. PDE 9 (2016) 1989; Houdayer arXiv:0809.3827, Proc. AMS 137 (2009) 3749 — both verified live: the free-quasi-free centralizer is $\mathbb{C}1$ iff $U_t$ has no eigenvectors). Correction (binding, carried from the A1 referee): this empties the centralizer $\mathcal M_\omega$, NOT the representation-space gap $\{\Delta\}''\subsetneq\{\Delta^{it}\}'$ — the original "$\{\Delta\}'=W^*(\Delta)$, the gap closes identically where geometry lives" conflates the two and is struck. The H-level gap stays non-empty even here; the only canonical sign it contains is $\eta_0=\mathrm{sgn}(\ln\Delta)$, which is non-local and geometry-void — so the genuine, surviving Prong-1 conclusion is that no localized indefinite carrier exists in the geometric sector (not that the gap is empty).

Prong 2 — almost-periodic sector: $h$ exists but is geometry-void.

• (iii) no geometric info — FAILS [INFERENCE, high]. Up to $\mathrm{Aut}(M,\omega)$ ($u\rho u^*=\rho$), the only invariant of $\mathrm{sgn}(h)$ is the signature pair $(p,q)$ (verified: equal-signature centralizer involutions are $\omega$-unitarily conjugate). But $(p,q)$ is the $(n,n)$ flag — posited, not derived. $\Delta_{\rm geo}=0$. The $\eta_0$ vacuity recurs inside the gap: unitarily universal.
• (ii) localized — FAILS, two ways [INFERENCE, high]. $h\in M_\omega$ is modular-invariant = constant on boost orbits = wedge-filling = maximally non-local. And to even pose (loc) one must choose a region order + which $a$ in the degenerate block — that choice is $n_1$ (iter-9 irreducibility). No centralizer-canonical "localized" is free of the net order.

Smuggle grade (Prong 2): $g_2$ (region, to claim loc) T · $g_3$ (a chosen non-geometric almost-periodic state) T-relocated · $n_1$ (index set, to localize + to pick $a$) T — decisive · $(n,n)$ template posited · $\Delta_{\rm geo}=0$ · $\kappa_\Phi>0$.

4. Triple re-derivation of the negative (headline-flip discipline)

• R1 (spectral). Multiplicity-free $\ln\Delta$ ⟹ commutant = diagonal = functions of $\Delta$ ⟹ gap empty (numerically demonstrated, $N=2000$ rapidity line).
• R2 (Connes/Arveson). Centralizer non-scalar ⟺ discrete $\Delta$-spectrum ⟺ almost-periodic state; trivial centralizer ⟹ no non-scalar self-adjoint commutes with $\sigma_t$ in $M$ (Houdayer 0809.3827, verified).
• R3 (boost-orbit). $h\in M_\omega$ modular-invariant ⟹ boost-orbit-constant ⟹ wedge-filling ⟹ non-local.

All three agree: the carrier-in-the-gap is nonexistent (Lebesgue) or universal-and-non-local (discrete).

5. Red-team vs the three named closures (no flip)

1. Boost-covariance/naturality engine (iter-8/9): the gap object evades it (it commutes with $\Delta$, is not boosted away — the genuinely new feature), but pays with non-locality by invariance instead of by boost. The dichotomy survives via a second locality mechanism. No conflict.
2. Localization-is-geometry (iter-9 F3): localizing $h$ needs $n_1$ — exactly the result. Reinforcing.
3. Encoding screen: $\mathrm{sgn}(h)$ encodes a hand-chosen $\mathbb{Z}_2$ grading of $M$; generates no geometry. Consistent.

No closure violated; no RESOLVED-POSITIVE.

6. Consequences proposed for the wiki

1. The commutant-to-generated residual is closed. Together with track A1, the iter-9 two-fold residual ($n_1$ plus the commutant gap) collapses to $n_1$ alone: the localization-template hypothesis is now the sole residual. The principal hedge HYP-CKV-VACUITY-R6 → R7 hardens MEDIUM-HIGH-to-HIGH toward HIGH, conditional on $n_1$ only.
2. Centralizer-completion of route 5, not a sixth route. The count stays FIVE (binding inheritance). The $\eta_0$ vacuity now has a sharp generalization: it holds not only for functions of $\Delta$ but for the entire modular centralizer.
3. New lemma proposed (LEM-CENTRALIZER-CARRIER). Every self-adjoint modular-centralizer element yields a sign $\mathrm{sgn}(h)$ that is unitarily universal (only invariant = signature) and modular-invariant (wedge-filling/non-local); it exists nontrivially only for almost-periodic (non-geometric) states; in the geometric/Lebesgue sector the centralizer is trivial and the gap is empty.
4. HYP-ENCODING-SCREEN inherits the narrowed condition.

7. Open subquestions

1. Is there a precise converse — a theorem that a net with no causal index order provably carries no localized form of any signature (true non-existence, not a residual)? [OPEN — inherited]
2. Does any PT-symmetric / non-self-adjoint relative modular structure evade by being neither modular-invariant nor a function of $\Delta$ (a third class outside both prongs)? [OPEN — inherited from iter-8]
3. The commutant $\{\Delta\}'$ also contains the commutant-side $M'\cap\{\Delta\}'$; this track took the algebra-side $M_\omega$ per the mission. Does the $M'$-side or the off-diagonal part of $\{\Delta\}'$ host anything new, or is it the mirror by $J$? [OPEN — flagged residual of this track]

Proposed registry items

LEM-CENTRALIZER-CARRIER — (new lemma)

Statement. A self-adjoint modular-centralizer element $h\in M_\omega$ that is not a function of $\Delta$ exists iff $\Delta$ has discrete-spectrum component (almost-periodic state); its sign $\mathrm{sgn}(h)$ is then indefinite and algebra-compatible (inner) but unitarily universal ($\Delta_{\rm geo}=0$, only invariant the signature) and modular-invariant hence wedge-filling/non-local; in the geometric (Lebesgue) sector $M_\omega=\mathbb{C}1$ and the commutant-to-generated gap is empty. Tag [INFERENCE] (high; construction [ESTABLISHED]).

HYP-CKV-VACUITY-R7 — (hypothesis-refinement)

Statement. The commutant-to-generated residual of the carrier no-go is closed (this track + A1); the principal hedge condition narrows from $\{n_1,\text{commutant-gap}\}$ to $\{n_1\}$ alone, hardening toward HIGH; count stays FIVE; headline unchanged. Tag [INFERENCE].

See also

• ../CONCLUSION.md §§4, 5, 8 — the hedge this note narrows; the carrier gate
• ../HYPOTHESES.md — HYP-CKV-VACUITY, HYP-FACTORIZATION-IS-GEOMETRY, HYP-MODULAR-TRICHOTOMY, HYP-ENCODING-SCREEN
• 2026-06-10-iter9-carrier-naturality.md — the commutant-to-generated gap this note closes
• 2026-06-10-iter7-indefinite-pairing-attempt.md — $\eta_0$ and its vacuity, now generalized to the centralizer
• 2026-06-08-iter3-encode-vs-generate-criterion.md — grading instrument
• ../domains/mathematics.md — Tomita–Takesaki, modular centralizer, free Araki–Woods factors
• ../EPISTEMICS.md — tag discipline

Binding note. Where a correction in the referee verdict below conflicts with the body above, the referee correction governs; the body is the pre-referee submission, retained for the audit trail per house convention.

Binding note. Where the referee verdict below conflicts with the body, the referee correction governs (notably: A1's "the commutant-to-generated gap closes in the Bisognano–Wichmann case via ergodicity" is a MAJOR error and is STRUCK — ergodicity empties the centralizer M_ω, not the representation-space multiplicity gap {Δ}″⊊{Δ^it}′, which stays non-empty and carries η₀; the hedge stays at R6, NOT R7; A3's "coefficient nonzero / R2 re-hardens" is corrected to SHARPENED-OPEN — the coefficient vanishes at the physical D=4, upholding iteration 9; the carrier-convergence count stays FIVE). The body is the pre-referee submission, retained for audit.

Referee verdict — track A2-carrier-in-the-gap (iteration 10, 2026-06-10)

Adversarial referee, default stance REFUTE. All mathematics re-derived stepwise; the toy model rebuilt from scratch and every numerical claim independently reproduced (numpy, fresh seed 20260610); both new citations audited live. Stance partially sustained: the outcome label holds, but two load-bearing sub-arguments are refuted and must be corrected.

Outcome: RESOLVED-NEGATIVE — UPHELD WITH TWO MAJOR CORRECTIONS. Headline unchanged (9th consecutive). Principal hedge narrows substantially, not to "n₁ alone."

The track asks whether the iter-9 commutant-to-generated gap (a Gram commuting with Δ lies in {Δ}′, not necessarily in W*(Δ), and these differ for degenerate III₁ modular spectrum) harbors a generative carrier. The answer is no — but the submission reaches it through one sound spine and two broken struts.

What is sound and reproduced. The gap object is genuine. I rebuilt ρ=diag(1,2,2,4)/9, Δ=ρ⊗ρ⁻¹ on the 16-dim standard form: Δ-multiplicities (1,4,6,4,1) confirmed; h=L(PauliX on the degenerate ρ-block) gives [h,Δ]=[h,lnΔ]=0 to machine precision with h∉W*(Δ) (HS residual 2.83 ≠ 0); for the full-rank a₂=diag(5,0,0,−5)+PauliX₁₂, sgn(h)=L(sgn a₂) has signature (8,8,0) and sits in M, so Ad(sgn h) is inner. Steps 1–2 are correct: the candidate genuinely clears bars (i) indefinite and (iv) algebra-compatible that iter-7's η₀ failed — a real advance, honestly flagged. The fork (Step 3) and prong-1 Lebesgue closure (Step 4) are structurally right: geometric wedge/boost modular flow ⟹ homogeneous Lebesgue spectrum ⟹ no modular eigenvectors ⟹ trivial centralizer ⟹ no carrier-in-M in the gap (Borchers + Bisognano–Wichmann, verified live; the no-eigenvectors ⟹ trivial-centralizer fact is real for free Araki–Woods factors).

MAJOR correction 1 — F3's invariant is false. The claim that "the only Aut(M,ω)-invariant of a centralizer sign-involution is the signature pair (p,q)," with its numerical support "two generic equal-signature centralizer sign-involutions are conjugate by an ω-preserving unitary," is refuted in the submission's own toy. An ω-preserving automorphism of Mₙ is inner (Skolem–Noether) and preserves ρ, hence is block-diagonal across ρ-eigenspaces; it cannot flip the sign on a 1×1 modular eigenblock. Explicit counterexample: two signature-(8,8) involutions with block-signs (+,[1,1],−) vs (−,[1,1],+) are equal in global signature but not ω-conjugate. There are 4 distinct Aut(M,ω)-classes of global signature (8,8), not 1. The true complete invariant is finer — signature resolved per modular eigenspace. Crucially, the vacuity conclusion (Δ_geo=0) survives: the finer invariant is still a discrete sign-pattern indexed by the modular point spectrum (part of the input Δ), choice-dependent and non-metric. So F3's conclusion stands but its stated reason is wrong; corrected_claim binding below.

MAJOR correction 2 — Step-5 (ii) mechanism-1 is sector-inconsistent. "h∈M_ω is modular-invariant = boost-orbit-constant = wedge-filling = non-local" invokes Borchers (σ_t = boost sweeps the wedge), which holds only in the geometric sector — but h exists only in the almost-periodic sector, where σ_t is precisely not a boost (no wedge, no Borchers rescaling). Mechanism-1 imports geometric-sector structure into the sector where its hypothesis fails. Strike it. The conclusion (ii)-localization-FAILS survives intact via mechanism-2 (localization presupposes the index-set order n₁ — the inherited, sound iter-9 localization-irreducibility result), which does the real work.

Citations — real, but mis-attributed (MINOR). Both new references verified live with correct bibliographic data: Boutonnet–Houdayer, Anal. PDE 9(8):1989–1998 (2016), arXiv:1602.01741; Houdayer, Proc. AMS 137 (2009) 3749, arXiv:0809.3827. But Houdayer 0809.3827 proves the bicentralizer is trivial (Connes' bicentralizer problem) — not the centralizer; and Boutonnet–Houdayer 1602.01741 says amenable modular-invariant subalgebras lie in the almost-periodic free summand. Neither states the "centralizer = ℂ1 iff U_t has no eigenvectors" biconditional the submission attributes to them (Step 3, F2); that fact belongs to Shlyakhtenko's free-quasi-free-states structure theory. This is precisely the reading-is-finder-inference error the iter-9 referee flagged. The biconditional is true and citable — just to the right source. Fix the attribution.

MINOR — commutant/centralizer conflation (Step 4). "M_ω=ℂ1 ⟹ {Δ}′=W*(Δ)" is false as a B(H) statement: the full commutant {Δ}′ is vastly larger than W*(Δ) whenever Δ is degenerate (infinite uniform multiplicity in the geometric sector). The correct, operative claim is M∩{Δ}′ = ℂ1 (no carrier in M). Reword.

No-go red-team — clean; no flip. The corrected finer invariant is indefinite (i) and algebra-compatible (iv) but carries zero metric/causal geometry (discrete, modular-spectrum-indexed, choice-dependent, non-localized): Δ_geo = 0 survives, no g₁–g₇/n₁–n₃ smuggle yields emergence, no RESOLVED-POSITIVE. The candidate is not realizable as a geometry-first description because its only content is a posited sign-pattern over the input modular spectrum. Carrier-convergence count stays FIVE (this is the centralizer-completion of route 5, not a sixth route — correct). No conflict with dS-cardinality, the iter-7 trichotomy, the iter-8 net no-go, the iter-9 naturality result, OP-48ab, or OP-49.

Hedge accounting (F5 overstated — bind). F5's "{n₁ + commutant-gap} → {n₁ alone}" overstates the discharge. A2 substantially closes the commutant gap for algebra-internal carriers (η∈M, Ad inner = centralizer case) — Lebesgue sector empties the gap, almost-periodic sector is geometry-void. But (a) the iter-9 gap was stated over {Δ}′ in B(H), and A2 narrows to M_ω without flagging that the residual sliver — algebra-compatible carriers (Ad η ∈ Aut(M)) that lie in {Δ}′ but not in M — is not fully treated; (b) the almost-periodic vacuity rests on the corrected (not the submitted) invariant. Net: the hedge narrows substantially but not provably to n₁ alone; the honest move is HYP-CKV-VACUITY-R6 → R7 with the commutant gap "closed for algebra-internal carriers, residual sliver flagged," grade unchanged (MEDIUM-HIGH→HIGH conditional). Object-level forecast unchanged; wager remains not-yet-physics.

Net. A verdict-neutral strengthening whose spine (the gap object exists; it is geometry-void on both prongs) holds, but whose two showcase moves — the "only invariant is (p,q)" universality and the boost-orbit non-locality — are respectively false and sector-inconsistent. Both conclusions survive via corrected/independent routes, so the negative outcome is upheld; but the submission's "two of four bars cleared, then closed cleanly" narrative must carry the corrections, and its hedge-narrowing claim must be softened. Standing verdict unchanged for the 9th consecutive confirmation.