# Reconstruction status — a thorough review of what is machine-verified (2026-07-23) **Purpose.** A single honest review of what the `csd-lean4` corpus actually proves, at commit `HEAD` (after the Σ-layer (projective-sector, Paper C) and the honest-alignment closeout). It supersedes scattered claims; where it and an older document disagree, this file and [`connectivity-manifest.md`](connectivity-manifest.md) win. Everything below is `sorry`-free, `lake build CsdLeanTests` green, and AxiomAudit-pinned to the foundational triple (`propext`, `Classical.choice`, `Quot.sound`) unless explicitly noted otherwise. > **The engine vs. the thesis — do not conflate them** (frame: [`CSD-CHARTER.md`](CSD-CHARTER.md)). > 1. **The QM calculation engine, on a concrete projective witness** — Ω-region volume ratios → Born, the > T1–T16 inventory inhabited. This is what the Lean proofs deliver, and it is essentially complete. It is > the **consistency floor**, not the thesis. (Reproducing QM ≠ the achievement.) > 2. **Completing the reconstruction** — making QM genuinely *arise from* Σ and Ω-regions on the ontic > surface, not accumulating more QM on the epistemic side. The record layer's **kinematic interface** is > formalized end-to-end (2026-07-25, `record-layer-plan.md §4`): measurement = `context + unknown > microstate → record` on the base×fibre Σ, Born = the law of large numbers over the unknown microstate, > outcome probabilities = the Kähler moment map. ⚠️ *This is NOT the completed reconstruction (corrected > 2026-07-28): the partition is preparation-indexed, so general-`N` A7 is open; no `H_int` generating the > basins is constructed; and the fibred Σ is not shown to be an A1 sector.* See §7. > **UPDATE 2026-07-31 — the first of those three is now addressed, the other two are not.** > `RecordLayer/GlobalBasin.lean` + `GlobalRecordClosure.lean` put the record layer on a > **context-fixed** partition at every `N`: `Bᵢ(c) = {(p,θ) : θ₁ ∈ circleCell (c.rate p) i}` reads > the rate field **at the ontic point**, so the record event is a function of `(context, outcome, > time)` and of no preparation, and `globalRecordClosure_born` still returns `‖⟨eᵢ,ψ⟩‖²`. ⚠️ This > closes the **preparation-indexing** defect only. It is **fibred**, not base-only — whether Paper C > intends `Ωᵢ(M)` to live on the base (where the ⏸ parked `ContextFixedA7` chain governs) is a > question about the axiom. **Still no `H_int`, and `KSigma` is still not shown Kähler.** > > The precise, defensible claim about the Lean today is: > > > The repository proves the finite-QM calculation engine on a concrete projective product witness satisfying > > the exact formalised subset of the Paper C assumptions. > > It does **not** claim to *derive* projective geometry, μ_FS, or Schrödinger evolution from a more primitive > model — those are **built into** the witness. And **Σ is the floor and is everything; deriving Σ is a > non-question**, not an open goal. ## 1. One-paragraph verdict The corpus is a **rigorous, forward, finite-dimensional QM reconstruction on one concrete projective-sector ontic witness.** A single genuine `Φ ≠ id` Kähler sector yields BOTH pillars (Born + Schrödinger) at general `N` with arbitrary Hermitian `H` (`manyToOneSchrodingerSetup_both_pillars`), and ONE ontic model `(Σ = ℂℙ^{M}×T², μL = μFS⊗vol, Φ, π = Prod.fst)` carries isolated Hamiltonian dynamics and de-isolating measurement together. `unified_projectiveSector_capstone` BUNDLES the dynamics + measurement core (six proved properties). Time-indexed record semantics, Born-frequency convergence (for EVERY unit `ψ`), and the conditioning = Lüders correspondence are ASSEMBLED into one tiered record: **`FiniteQMClosure`** / `unifiedFiniteQMClosure` (`SigmaLayer/FiniteQMClosure.lean`) collects all eleven proved-on-the-model facts as fields, each discharged by its source lemma, and states honestly what is a theorem here vs. a projective-sector posit vs. a QM adapter vs. still open (no field is `sorry`). **This is a concrete consistency witness, not a derivation of the Paper C architecture:** the witness has `μL = μFS ⊗ vol` and `Φ_t = (e^{-itH}·[p], θ)` built in, so the pushforward-to-`μFS` and the Schrödinger pillar are *compatibility facts about the witness*, not derivations of `μFS` or of unitary evolution from a fibre-primitive ontology. **This closure is the QM *calculation engine* demonstrated — the consistency floor, not the thesis** (see [`CSD-CHARTER.md`](CSD-CHARTER.md)). **Σ is the floor and is everything; deriving it is a non-question** (Paper C is explicitly "a reconstruction, not a derivation"). Toward the goal — **complete the reconstruction of QM from Σ and Ω-regions** — the **record layer (MD-1)** supplies a formalized **kinematic interface** (2026-07-25, `record-layer-plan.md §4`), on the base×fibre model space Σ, with Born = the law of large numbers over the unknown microstate and the outcome probabilities = the Kähler moment map. ⚠️ **The goal is NOT met** (corrected 2026-07-28, external review; an earlier version of this sentence said it was). Three items are open: general-`N` A7 (the partition is preparation-indexed — `N = 2` is done, `LF4/QubitBorn.lean`), the de-isolation `H_int` generating the basins (`DeIsolationInteraction.basin_rate` is a hypothesis field), and whether the fibred Σ is an A1 sector. ⚠️ **Updated 2026-07-31: the first item is now addressed *on the fibre*** — `GlobalBasin` / `GlobalRecordClosure` give a partition whose events mention no preparation, at every `N`; the **base-only** question stays ⏸ parked. ⚠️ **Updated 2026-08-25: item three is now CLOSED** — `kMuL_unique` (`LF4/LiouvilleUnique.lean`) exhibits the sector's measure as forced by its own symmetry rather than merely Haar, completing the A1 ontic-sector row (all three sub-items done; the residual `dω = 0` / top-power half is a Mathlib wall, `Q8`, not a CSD one — ⚠️ and 2026-09-07 the `dω = 0` half of that wall fell: `fsForm_mextDeriv`, closedness of the Fubini–Study form on `ℂℙⁿ` at manifold level; ⚠️ 2026-09-10 the top-power identity fell too, `fsVolume_eq_smul_fsMeasure` (`ω^{∧n} = (4π)ⁿ μ_FS`), so `Q8` is closed in-corpus). **Item two stands, and 2026-09-05 it is stated more precisely** — "no `H_int` generates the basins" ran together three things that `specs/POSITS.md` now separates: *which* interaction an apparatus realises is a **permanent boundary** (`R-015`, never closes); the chart→arena transport of Hamiltonian generation is **open mathematics** (`R-016`); and deriving the rate field from the de-isolation dynamics is **open foundations** (Posit 1's discharge condition). `ShearDeIsolation.lean` discharges `basin_rate` from a constructed propagator — a witness, not a derivation. The goal is still not met, and it is the third of those that is why. Beyond those, the residue is a mechanical field naming in the pinned closure (no new theorem). (The Ω-regions are epistemic, on `ℂℙⁿ⁻¹`; the record is the ontic selection in Σ.) The earlier "SO-1 = derive the sector" framing that appeared here is a **retired error** (§7). ## 2. The Paper C axiom map (A1–A7) — canonical formalisation status This is the canonical map of *what the Lean corpus formalises against each Paper C axiom.* "Partial", "witness", and "exact ε=0" are used honestly; do not read "represented" as "derived". | Paper C axiom | Lean status | |---|---| | **A1** compact Kähler ontic surface | **Partial.** Compactness and the *pointwise* Kähler compatibility core are represented (`IsFubiniStudyKahler`, `isFubiniStudyKahler`); full manifold exterior calculus (`dω=0`, `ωⁿ/n! = μ_FS`) is formalised (2026-09-08; the Kähler predicate 2026-09-10, `fsForm_isKahler`) and wired to the sectors (`LF4/SectorManifold.lean`, W1). ⚠️ **Separately (2026-07-30): the *fibred* Σ's A1 status is a dimension-parity question, not a tooling one** — `ℂℙⁿ⁻¹ × ℝ` (`FibredSigma`) and `ℂℙⁿ⁻¹ × AddCircle 1` (`CircleFibre`) both have real dimension `2n-1`, **odd**, so neither admits *any* symplectic, hence Kähler, structure. Only `KSigma = ℂℙⁿ⁻¹ × T²` (dimension `2n`) can. See §2a. | | **A2** Hamiltonian ontic dynamics — ⚠️ *the witness that carries it pays a trilemma horn; see `SigmaLayer/FiniteQMClosure.lean`, "Which trilemma horn this closure pays" (**seams**: unitary interaction, exact Born, a.e.-total readout), and `docs/TOUR.md`, "The three priced witnesses"* | **Formalisable half DISCHARGED 2026-08-02** (`RecordLayer/HamiltonianSignature.lean`): the witness flow carries every Hamiltonian **signature** a measure space can express — a canonical conserved energy (`onticEnergy`, ray-well-defined, conserved under the flow: `onticEnergy_flow_invariant`), which is *simultaneously* A5's exactly-projectable `h` (`onticEnergy_epsProjectable` — the A2/A5 junction); the commuting phase torus preserving every moment-map coordinate (`momentMap_phaseDiag_invariant`, `phaseDiag_comm`); plus the already-proved Liouville property and group laws, with the fibre translations measure-preserving (`ShearWitness`). **The vector-field equation `X_H = ω⁻¹dH` itself is §2a-scoped** (symplectic form + exterior derivative; verified a tooling gap). The witness uses the projected `e^{-itH}` lift. ⚠️ *Correction 2026-08-02 (external review, machine-checked same day):* the **measurement-interval** propagators (`shearEvolve`, `swapEvolve`) are **provably not continuous** (`shearEvolve_not_continuous`, `RecordLayer/ShearDiscontinuity.lean`) — they are measurable, measure-preserving, *piecewise* maps, so they are **not** time slices of any continuous flow, and the Hamiltonian-generation question is **reopened**, not Mathlib-scoped (`BACKLOG.md`, decision pending: corridor regularisation vs piecewise-Hamiltonian classification). The *isolated* dynamics (this row's flow) is untouched by the correction — `e^{-itH}` on rays is continuous; the discontinuity enters only with the de-isolating measurement interaction, exactly where `no_everywhere_correlation` proves some seam set is forced. *Resolution 2026-08-02 (user decision):* the piecewise classification landed (`RecordLayer/PiecewiseHamiltonian.lean`) — each basin cylinder carries an explicit continuous rigid translation, the seam set is null (`seam_null`). ⚠️ *Second correction, same day (second external review):* **(flux)** those translations are symplectic but **not globally Hamiltonian** on the compact torus (`ι_Xω = a·dp` closed-not-exact; `∮dp ≠ 0`), so the classification's correct name is **piecewise rigid symplectic translation**, not "piecewise Hamiltonian" — the `hᵢ = aᵢ·p_R` reading is withdrawn; **(overread)** `no_everywhere_correlation` forces an exceptional set, not discontinuity — a continuous globally-Hamiltonian propagator with seams mapped to transition states remains open. The Hamiltonian-generation question is **REOPENED** with the recommended compact-Kähler-pointer route `ℂℙ^K` (`BACKLOG.md`); *route confirmed 2026-08-03 (author decision): the smooth-Hamiltonian architecture of Paper C A2 / TN6 is retained, and the `ℂℙ^K` pointer witness is the obligation — execution plan `pointer-witness-plan.md`.* **★ DELIVERED 2026-08-03 (all five bricks, `pointer-witness-plan.md`):** `smoothWitnessClosure` — one witness carrying the measurement protocol (two-time law = exponential group property), **joint time–state continuity** (`continuous_pointerRampedEvolve`; contrast `shearEvolve_not_continuous`), Liouville preservation, a positive-measure ready state (no Dirac calibration posit — the preparation conditions on the positive-measure ready region), record creation with the ontic sector selecting the outcome, structural persistence, and the ε-Born sandwich `rⱼ − 2ε ≤ sector ≤ rⱼ + 2(N−1)ε`; and ★ `rampedU_schrodinger` — on the interaction window the propagator satisfies the **Schrödinger equation** with the explicit **Hermitian** generator `H_eff = (π/2)•couplingH w` (single-plane exponential identification `pointerRot_eq_exp` by ODE uniqueness). **The Hamiltonian-generation sub-question is discharged at the formalisable level**: the flux obstruction is gone (projective pointer, `H¹(ℂℙ^K) = 0`, generator exhibited); the symplectic/moment-map *reading* remains the same §2a-scoped prose boundary as A1/A3 (Mathlib has no symplectic API); exact records/Born are traded for the stated `ε` per `no_everywhere_correlation` — the exact-record piecewise witnesses stand alongside. ⚠️ **Corrected 2026-08-04 (independent Opus review of HEAD).* — this row said "**the** other horn", a two-horn framing, and this file (which `README.md` sends readers to for the axiom-level audit) contained **no mention of the third horn at all**. The fork is a **trilemma**: seams (piecewise), `ε`-Born (smooth witness), or Dirac calibration (`nullSeamClosure`, `RecordLayer/NullSeamWitness.lean` — continuous, measure-preserving, exact Born off a two-point seam). Whether a fourth combination is impossible is open, not claimed.* ⚠️ *Boundary sharpened 2026-08-03 (fourth external review, verified both ways):* the delivered generation is **fibrewise** — the arena propagator's generator is vertical `(0, V_Q)`, and for the joint scalar `𝓗(x,q) = μ_{H_eff(w(x))}(q)` one has `ι_Vω ≠ d𝓗` on the ε-collars where the weights vary (off the collars the weights are locally constant and the flow is genuinely locally Hamiltonian; the defect shares the Born error's `O(ε)` budget). The suppressed horizontal component is register back-reaction — `pointerEvolve_base_marginal_unchanged` states it as a theorem. The accurate name is **continuous fibrewise-Schrödinger witness**: "smooth horn" names the ε-corridor architecture, and the weight/ramp ingredients are Lipschitz, proved `Continuous`, not `C^∞` (the `Real.smoothTransition` upgrade **landed 2026-08-03**, `RecordLayer/PointerSmoothProfile.lean`; substituting those profiles into the protocol stack is the part still outstanding — `BACKLOG.md` B1). The formalisable-level discharge stands **as scoped**; the full back-reacting joint-arena flow — where the register moves mid-stroke and exact moment transport no longer applies — is the recorded **research** row in `BACKLOG.md`, a frontier, not a silent claim. ▸ *Update 2026-10-27 (BACKLOG #209):* arena-level Hamiltonian generation of a **record-creating** propagator is now proved — for the activated write (`hamiltonianFlow_activeFlow`, `RecordLayer/ActivatedHamiltonian.lean`, #195, on `ℂℙ³ × T²`) and the autonomous echo (`hamiltonianFlow_echoFlow`, `RecordLayer/EchoHamiltonian.lean`, #202, on `ℂℙ³ × T² × T²`); *2026-10-29 (BACKLOG ~~#225~~):* also for the clock coupled to the echo (`hamiltonianFlow_coupledFlow`, `RecordLayer/ClockEchoHamiltonian.lean`, on `(ℂℙ³ × T² × T²) × (ℂℙ³ × T²)`, the cross term pairing the clock's ray with the echo's pointer angle). Both are time-independent smooth Hamiltonians for the arena symplectic form, the pointer coupled through a periodic potential of a conserved angle, with no seams, no ε-collars and no flux obstruction: the pointer's field is the Hamiltonian field of `f(θ₂)`, globally. Their price is the trilemma's smooth horn, stated: the write is exact only off a positive-measure transition region (`transition_region`). What stays open, `R-016`, is the same statement for the shear and joint-lift propagators, whose strokes are symplectic but not Hamiltonian on the torus. | | **A3** smooth many-to-one projection | **Discharged 2026-09-11 (Q33).** The arena `ℂℙⁿ × T²` is an analytic manifold (`CSD.LF4.ksigma_isManifold`; `AddCircle` made a manifold in `Instances/AddCircle.lean`) and `π = Prod.fst` is `C^ω` (`manyToOneSetup_pi_contMDiff`). Lean's proofs use only measurability; the witness is genuinely many-to-one (the `T²` fibre). | | **A4** pushforward measure `π_*μL = μ_FS` | **Proved for the witness** (`productSector_hasFubiniStudyPushforward`, B1) **and forced under full unitary symmetry** (`localised_sectorPostulate_capstone`); **not derived** from arbitrary ontic dynamics. The witness has `μL = μFS ⊗ vol` built in. | | **A5** quantum-effective Hamiltonians (projectability) — ⚠️ *the unitary class is a separate posit (`specs/POSITS.md` Posit 6, review Part II Posit 4): `projectedFlow_eq_unitary_family` takes `ProjUnitary` as a hypothesis and selects the family, so the corpus proves "unitary ⇒ Schrödinger" (`CL-063`) and posits "unitary". Motivated by no-signalling (Gisin 1990), not derived from it; see `LF4/ProjectedDynamics.lean`* | **Approximate case FORMALISED 2026-08-02** (`RecordLayer/ApproxProjectability.lean` + the Duhamel bound): `EpsProjectable` (oscillation form; the derivative form is the §2a-scoped manifold refinement, substitution stated), `epsProjectable_zero_iff` (exact case ⇔ `ε = 0`, an iff), and the **shadowing theorem** — `‖H−H₀‖ ≤ ε`, `|t| ≤ T` ⇒ sector dynamics tracks true dynamics within `ε·T`. The exact fibre-invariant case `H = h∘π` was already formalised (projected flow closes, `e^{-itH}` on rays). ⚠️ Residue: the shadowing is Hilbert-side; an ontic Hamiltonian *generating* the flow is A2's row. ▸ *2026-10-29 (BACKLOG ~~#208~~):* for coupled energies on the arena the bridge is proved — the derivative form is stated and implies the oscillation form at `ε/2`, and the activated write's sector predictions are shadowed at `‖O‖·ε·T` with its sharp ontic `ε` (`activated_regime`, `SigmaLayer/ActivatedRegime.lean`). | | **A6** composites + marginal stability | **Non-factorisation PROVED 2026-08-02** (`RecordLayer/OnticComposite.lean`): the Segre embedding `([u],[v]) ↦ [u⊗v]` is **injective but not surjective** at every dimension pair ≥ 2×2 (Bell-type witness), so `Σ_AB ⊋ image(Σ_A × Σ_B)` — the ontic composite strictly exceeds the product of sectors, as a theorem. Operational tensor structure, reduced states and no-signalling already existed (`compositeTensorEquiv`, `tensorSector_no_signalling`). **Steps 2–3 DONE 2026-08-02** (`RecordLayer/OnticMarginals.lean`): ontic reduction maps (`rayDensity` unit-trace + `reduceA`/`reduceB` partial traces), and **marginal stability under local flows** — `reduceB_local_flow_invariant`: the Schrödinger flow of any `A`-Hamiltonian leaves the `B`-marginal fixed at every time (ontic no-signalling), with the `A`-marginal evolving by conjugation. **Every A6 clause the corpus can express is now a theorem.** ~~Remaining: step 4 (dynamical no-signalling through the measurement layer — optional capstone)~~ **Step 4 DONE 2026-08-03** (`RecordLayer/LocalLuders.lean` → `LocalLudersBasis.lean`): a local measurement IS a block-degenerate measurement (`toComposite_blockProj`, definitional under the index bridge), and the Born-weighted mixture of post-measurement local marginals equals the pre-measurement marginal **in every measurement basis** (`reduceA_localLudersOn_mixture`) — the distant outcome *and the basis choice* are dynamically invisible, with the post-states exactly those the join witness delivers (`BlockLudersObligation`). The eraser rides the same machinery as a process (`EraserDynamics`, `EraserSequential` — including `sequential_no_revival`: records are statistically irreversible). A6-as-philosophy (`Σ_AB` primitive) is not a formalisation target. **Generation premise in record vocabulary 2026-09-02** (`SigmaLayer/TensorTomography.lean`): `hgen` of the "why `⊗`" reconstruction ⟺ record-level local tomography (`recordLocallyTomographic_iff_adjoin_eq_top`); the surviving posit is boundary `R-017`. | | **A7** context-defined measurement partitions | **DISCHARGED at every `N` (author decision 2026-08-02: the FIBRED reading is canonical).** Paper C A7's `Ωᵢ(M)` is the TN6 two-level structure: ontic basins `Bᵢ(M) ⊆ Σ` fixed by the apparatus (`globalBasin` — no `ψ` in the definition), with the epistemic response kernel induced on the base. That structure is now realised **kinematically** (`GlobalRecordClosure`, `globalBasin_born` = `‖⟨eᵢ,ψ⟩‖²`) **and dynamically** (v0.7.0: records created from a ready state, persistent, Born-weighted, with the rank-one Lüders update — `DynamicMeasurementClosure.luders_followup`). **Base-only regions on `ℂℙⁿ⁻¹` are the completed qubit special case** (`LF4/QubitBorn.lean`), not the general mechanism; whether a base-only realisation exists at `N ≥ 3` remains the ⏸ parked `ContextFixedA7` question — still open in both directions, now characterising the special case rather than gating the axiom. *(Historical row, superseded by the decision:)* **OPEN at general `N`; DISCHARGED at `N = 2` (2026-07-26).** ⚠️ *This row read "Addressed by the record layer"; corrected 2026-07-28 — the record layer's partition is built from the preparation (`bornContext ψ`), so it does not establish `Ωᵢ(M)` from the apparatus alone.* The record layer supplies the **kinematic** interface (MD-1, `record-layer-plan.md §4`); The record layer formalizes measurement as `context + unknown microstate → record` on the base×fibre Σ, with outcome probabilities = the Kähler moment map (`MomentMapRace`, forced not injected) and frequencies = the LLN over the unknown microstate (`Measurement.bornMeasurement_frequency`). **The A7 residual — that the epistemic partition be genuinely *context-fixed* (a function of the measurement context alone, not the preparation, rather than the corpus's preparation-indexed `bornRegion ψ`) — is now discharged for the qubit** (`LF4/QubitBorn.lean` `qubitBorn`, foundational-triple, pinned): the hemisphere partition `{H±(n)}` depends only on the axis `n`, and the Born weight `|⟨n\|ψ⟩|²` is *derived* from the ontic Fubini–Study typicality volume via the CSD spread density `4(2·blochProj ψ − 1)₊`. The 7-module `CP¹` chain (`QubitReflection`→`BlochProjection`→`AxisBridge`→`QubitDipole`→`QubitCrossTerm`→`QubitBorn`, + `HatBox`) is `record-layer-plan.md §2/§4`. **★ UPDATE 2026-07-31 — a context-fixed partition now exists at every `N`, on the fibre.** `RecordLayer/GlobalBasin.lean` defines `ContextField` (a measurable simplex-valued rate *field* on the base) and `globalBasin c i = {x | x.2.1 ∈ circleCell (c.rate x.1) i}`, reading the rate **at the ontic point** — so no preparation enters the definition; `globalBasin_born` returns `‖⟨eᵢ,ψ⟩‖²`, and `RecordLayer/GlobalRecordClosure.lean` carries the five record-layer facts on it over `KSigma = ℂℙⁿ⁻¹ × T²`. The prerequisite was `LF4.measurable_momentMap`, proved by quotient descent. ⚠️ **What this closes is the preparation-indexing defect, not general-`N` A7 outright**: the partition is **fibred**, and the parked chain below concerns **base-only** regions. Whether A7 wants base-only regions is a question about the axiom, not about the Lean. Still kinematic — no `H_int`. Remaining: the general-`N` **base-only** context-fixed partition. ⚠️ **This row previously said a base-only, `U(N)`-covariant, nonnegative context-fixed density "**cannot**" reproduce Born for `N ≥ 3`, "proven numerically + operator-theoretically". That is an overclaim and is retracted (it should have gone when the same claim was retracted from `BACKLOG.md` on 2026-07-28; it survived here — a staleness bug, corrected 2026-07-30).** What is actually established is a chain of **necessary conditions** on such a density (`SigmaLayer/ContextFixedA7*.lean`): it is confined to measure `≤ 1/n` and vanishes below `(n−1)/n`, with the `N=2` / `N≥3` threshold confirmed three independent ways. That constrains a base-only construction; it does **not** refute one, and no construction has been exhibited either. **General-`N` A7 is unresolved in both directions and is ⏸ parked** ([`sigma-fibre-contextuality.md`](sigma-fibre-contextuality.md)). The working architectural conclusion — that at `N ≥ 3` contextuality should live in the **fibre** of Σ — is what the parked chain *motivates*, not what it proves; a Born fibre-partition exists (Phase-2b), but its mechanism is posited, not derived from a de-isolation dynamics. So A7-as-base-regions is proved only at the qubit, where `CP¹=S²` supplies an antipode with no `N≥3` analogue. (And this is **not** Gleason — CSD is contextual.) The pinned `FiniteQMClosure` field still names the preparation-indexed `bornRegion ψ'` cells (a mechanical residue; `KSigmaRecord.born_frequency_region_eq_record` shows its region already *is* the record-layer event). | ### 2a. Audit of this map (2026-07-29) — and the scoping decision **Audit result: all seven rows are accurate.** Every cited name resolves to a real declaration (`flow_preserves` is a `ConstraintDynamics` *field*, not a top-level theorem — correct as cited), `compact_sigma : CompactSpace Sigma` is a genuine carried field so A1's "compactness … represented" is fair, and A5's approximate case is *genuinely absent* from the corpus (no `ε`-projectability anywhere), so "not formalised" is exact. Worth stating plainly: after the A7 correction of 2026-07-28 one might expect this map to be optimistic, and it is not. **The scoping decision.** Five rows read "Partial", which invites reading the whole map as unfinished. It is not: the rows fail for *different reasons*, and only some are work. **(i) Permanently scoped — blocked on Mathlib, not on us.** ⚠️ **Re-classified 2026-09-11.** This category was written when the manifold API was absent and *staying* absent was the working assumption. The corpus then built the API itself (the G series, `generator-layer-scoping.md`). Nothing below is permanently scoped any more: one item is DISCHARGED, two have moved to (iii) as priced bricks. The heading is kept so that older citations of "§2a (i)" still resolve; read the bullets for the current state. * **A1, the exterior-calculus half** — ✅ **DISCHARGED 2026-09-07 → 2026-09-11 on the `ℂℙⁿ` base.** `dω = 0` is `Projectivization.fsForm_mextDeriv`; `ωⁿ/n! = μ_FS` is `fsVolume_eq_smul_fsMeasure` (`ω^{∧n} = (4π)ⁿ μ_FS`, the constant visible) and `riemannianVolume_fsMetric` (`vol_g = ω^{∧n}/n!`); the Kähler predicate itself is `fsForm_isKahler` (atlas sense) with `nijenhuis_fsJ_eq_zero` (tensor sense); and the sector's `liouvilleMeasure` is proved to be this volume on every `ℂℙⁿ` instance (`LF4/SectorManifold.lean`, connectivity link L1 discharged). The *formalizable core* (`IsFubiniStudyKahler`, `kahler_pointwise`) is unchanged and remains the field's type. What is NOT done: the same on the *fibred* arena `ℂℙⁿ × T²` as one Kähler manifold (the fibre's `T²` is Haar, not a Kähler factor; parity is even) — that is the arena half of `R-016′`, below. ⚠️ **Scope of this bullet, narrowed 2026-07-30.** It covers the exterior calculus **on `KSigma`** and nothing else. It was being cited to excuse the *fibred* Σ's missing Kähler structure as well — including by `CircleFibre.lean` and by the ★★ `BACKLOG.md` row. **That citation was wrong.** The fibred Σ's problem is dimension parity (see the ★ paragraph below), which no Mathlib API can repair. Do not use this bullet to classify an odd-dimensional arena as "scoped". * **A3, smoothness of `π`** — ✅ **DISCHARGED 2026-09-11 (Q33, same day it was re-classified).** `AddCircle T` is an analytic manifold (`Mathlib/Geometry/Manifold/Instances/AddCircle.lean`: a charted space and all its groupoids transport along a homeomorphism, applied to `AddCircle.homeomorphCircle`), so the arena `KSigma = ℂℙⁿ × T²` is one (`CSD.LF4.ksigma_isManifold`) and `π = Prod.fst` is `C^ω` (`manyToOneSetup_pi_contMDiff`, `LF4/SectorManifold.lean`). Lean's proofs still use only measurability, and the witness is genuinely many-to-one. * **Hamiltonian generation of the measurement witnesses** *(classified 2026-08-02, user decision; re-classified 2026-09-11)* — ➡ **moved to (iii) as `R-016′`, L.** The shear and calibrated-swap propagators are explicit, with every required property proved *of* them (correlation, measure preservation, persistence, Lüders); that they arise as time-`T` flows of `H_int = g(t)(ι+1)δ·p_R` (+ the record-triggered kick) is a symplectic-geometry statement on the joint arena. When classified, Mathlib could not express it (no manifold Hamiltonian-flow API). Since 2026-09-08 the corpus can, on `ℂℙⁿ`: `IsHamiltonianVectorField`, existence/uniqueness/smoothness of `X_H`, integral curves, and the unitary flows proved Hamiltonian (`schrodingerField_isHamiltonianVectorField`). What remains is the *arena* — the product manifold `ℂℙⁿ × T² × pointer` with a product symplectic form, and the piecewise pieces stated on it — which is a brick, not a wall. ⚠️ *Still true and still verified:* the generating Hamiltonians are written down, the claim is standard physics, nothing here is odd-dimensional; and a *discharge* (proving the propagators ARE those flows) is the `H_int` frontier, paper-side. The statement-shape is `R-016′`. ⚠️ **`R-016′` landed 2026-09-14** (`LF4/ArenaSymplectic.lean`): the arena `ℂℙᴺ × T²` is a symplectic manifold, Liouville holds on it, and the *isolated* dynamics is proved to be its Hamiltonian flow (`hamiltonianFlow_sectorEnergy_schrodinger`); the torus strokes are proved locally Hamiltonian there. The measurement propagators (`jointLift`) as time-`1` maps on the arena remain the paper-side content of `R-016`; the flux obstruction at manifold level landed the same day (`LF4/ArenaStrokeFlux.lean`: for `a ≠ 0` no function has the torus stroke as its Hamiltonian vector field), so the strokes' classification "symplectic, not Hamiltonian" is a theorem there. The arena's top-form measure is the corpus's `μ_FS ⊗ vol_{T²}` up to its total mass (`arenaVolume_eq_smul_kMuL`, `LF4/ArenaVolume.lean`, same day, by uniqueness of the invariant measures rather than the binomial identity), so arena Liouville is Liouville for `kMuL` at manifold level (`kMuL_smul_map_hamiltonianFlow`); the mass is `(N+1)·(4π)^N` (`arenaVolume_univ`, 2026-09-17), so Liouville holds for `kMuL` itself (`kMuL_map_hamiltonianFlow`). **(ii) Scoped by doctrine, not by tooling.** * **A4** — proved for the witness, which has `μL = μFS ⊗ vol` built in, and *not* derived from arbitrary ontic dynamics. Under "Σ is the floor", deriving it is a non-question. The legitimate content is constraining Σ from above, and that is exactly `localised_sectorPostulate_capstone`: `μ_FS` is **forced** under full unitary symmetry. A4 is closed in the only sense available. **(iii) Genuinely open — these are the reconstruction's remaining work.** * **A5 — approximate `(ε,T)`-projectability. NOT blocked on Mathlib**, and the highest-value item after the fibre. Only the exact case `H = h∘π` is formalised (`kSectorDataFlow_projectable`, `DynamicsBridge`), but A5's *physical* content is the approximate one, `sup‖d(δH)|_V‖ ≤ ε` over a window — that is what makes a Hamiltonian *quantum-effective* rather than arbitrary, i.e. what **selects the sector**. The ingredients exist here already (operator norms, matrix-exp differentiability, `StoneC1`). Effort **M–L**. * **A2 — a generic ontic Hamiltonian vector field `X_H`.** Partly shares A1/A3's blocker, but the physical content — a genuine, physically meaningful `Φ ≠ id` — is real work and overlaps the de-isolation `H_int` item. * **A6 — the non-factorising ontic composite** (`Σ_AB ≠ Σ_A × Σ_B` as primitive). Architecture, not tooling. * **A7 — ⏸ parked** at general `N`, discharged at `N = 2`. See [`sigma-fibre-contextuality.md`](sigma-fibre-contextuality.md). **★ A1 splits, and the tractable half is the live one.** The exterior-calculus half is scoped (i), but "is the *fibred* Σ a compact Kähler sector at all?" is a **different and unblocked** question: `FibredSigma` uses `ℂℙⁿ⁻¹ × ℝ` — non-compact fibre, measure not shown Liouville — whereas `KSigma = ℂℙⁿ⁻¹ × T²` is compact and already in the corpus. Swapping the fibre needs no Mathlib API. Since the A7 work concluded that contextuality lives in the fibre, this is now the load-bearing question. `BACKLOG.md` ★★ row. **★ UPDATE 2026-07-30 — the fibre swap half-landed, and the parity fact that reclassifies the rest.** Two modules landed on 2026-07-29/30. `RecordLayer/CircleFibre.lean` puts the active Born partition on a **compact** fibre `AddCircle 1`, with Haar a genuine probability measure and cells carrying exactly the Born weights; `RecordLayer/CircleRecord.lean` gives the compact counterpart of the P5 record semantics, with `ae_total` coming out *stronger* than on `ℝ` (whole-space, not a hand-chosen window). Two corrections attach to that work: * **It is a parallel construction, not a migration.** The landing commit was headlined "the record layer now runs on the compact fibre" and the ★★ BACKLOG row logged the re-plumbing as DONE. Both overstated it. `Measurement.lean`, `RecordLayerClosure.lean`, `FiniteQMClosure.lean` and `KSigmaRecord.lean` still run on `ℝ` with `fibreTypicality`, and nothing outside `AxiomAudit.lean` imports the circle modules. The accurate claim is *a compact counterpart of the record semantics has been proved.* * **A single circle cannot complete A1, and this is not a tooling gap.** `ℂℙⁿ⁻¹ × AddCircle 1` has real dimension `2n-1` — **odd** — and no odd-dimensional manifold admits a symplectic form (`ωᵏ` must be a volume form), hence none admits a Kähler structure. The same applies retroactively to `FibredSigma`'s `ℂℙⁿ⁻¹ × ℝ`. So the fibred Σ's missing Kähler structure was **never** blocked on Mathlib's absent exterior calculus, and classifying it under (i) above was a **misclassification**. The fix is in the corpus already: `KSigma = ℂℙⁿ⁻¹ × T²` (`LF4/KahlerInstance.lean`, `KTorus = AddCircle 1 × AddCircle 1`) has real dimension `2n`, is even, compact, and is a product of Kähler manifolds. The successor construction puts `circleCell` on **one** torus coordinate and keeps the second as its symplectic partner. Effort **S–M**, and it needs no new fibre mathematics. **Net corrected 2026-08-02 (evening): A2's Hamiltonian-generation sub-question is REOPENED (torus-flux, second external review — see the A2 row); the rest of the net line stands.** **Net for "finite QM from Σ" (updated 2026-08-02): three rows permanently scoped (A1-exterior *on `KSigma`*, A3, witness-`H_int` origin), one closed as a posit with the symmetry-forcing result (A4), **A7 discharged** (fibred canonical; base-only = the parked qubit-special-case question), **A5's approximate case formalised 2026-08-02** (oscillation predicate + exact-case iff + `ε·T` shadowing), **A2's formalisable half discharged 2026-08-02** (the Hamiltonian signature package; the vector-field equation is scoped), and **A6's expressible content proved 2026-08-02** (non-factorisation via the Segre embedding; reduction maps; marginal stability under local flows). **The A1–A7 map has no unscoped open rows left**: every row is proved, discharged, closed-as-posit, or explicitly scoped with a verified reason. The frontier is now the recorded L-scale items (degenerate Lüders via the join, engine migration, Kähler/Liouville on `KSigma`, unitary covariance/POVMs, dynamical no-signalling).** That is the honest distance to closed. **Closure declared 2026-08-06 (`v1.1.0-finite-qm-closed`).** Every L-scale item the net line above listed as the frontier has since landed (degenerate Lüders via the join 2026-08-03, POVM/instrument + sector Born 2026-08-03/04, dynamical no-signalling 2026-08-03, the smooth-witness Lüders composition 2026-08-05, degenerate-on-mixed 2026-08-06), and the README's "Operational finite-QM closure" section fixes the operational definition and maps each requirement to its carrying theorems. The scoped rows above are unchanged by the declaration — closure is *of the operational surface, under the recorded posits and scopes*, exactly as this document has always measured it. ### 2a. Target inventory (T1–T16) — inhabited reconstruction targets The A1–A7 map above is the *axiom-formalisation* status. The following is the separate inventory of *reconstruction targets* that are inhabited by proved theorems on the witness. NONE of these is an `axiom`; postulates are structure fields, bridges are named assumptions discharged per model, targets are `Prop` predicates inhabited by proved theorems. Full ledger in [`../CsdLean4/SigmaLayer/Adapters.lean`](../CsdLean4/SigmaLayer/Adapters.lean) and [`../AXIOMS.md`](../AXIOMS.md) §3.7. | # | Target | Key theorem | Status | |---|---|---|---| | T1 | Born from volume | `BornFromVolume`, LF4 `fs_born_volume_ratio_N` | proved | | T2 | Born from i.i.d. frequencies | `born_frequency_convergence_N` | proved | | T3 | Born from deterministic-flow frequencies | `BornFromFlow` predicate | **OPEN (= SO-1 face)** — a single flow cannot pin `μ_FS` (no-go below); residual gap is the Mathlib-absent pointwise Birkhoff theorem, and the unitary no-gos exclude the hypothesis | | T4 | Unitary projected dynamics | `HasUnitaryRealisation` | proved (witness) | | T5 | Schrödinger evolution | `HasHamiltonianRealisation`, `productProjectedFlow_hasHamiltonianRealisation` | proved (witness) | | T6 | Unique contextual outcome a.e. | `vnDeisolationModel_ae_total` | proved (preparation-indexed cells — MD-1) | | T7 | General conditional update | `conditionalUpdate_capstone` | proved | | T8 | Lüders update | `luders_capstone` (sharp special case of T7) | proved | | T9 | Mixed states | `mixedState_capstone`, `isPure_iff_trace_sq_one`; ensemble #8 A+B `traceForm_ensemble`, `density_isPureEnsemble`, `mixedEnsemble_capstone`; ontic-side #8 C weight `mixed_ontic_born_weight` (= `FiniteQMClosure.mixed_born`) **and frequency** `unified_mixed_born_frequency` (= `FiniteQMClosure.mixed_born_frequency`) | **proved (weights + a.s. frequency; both closed)** | | T10 | POVM by dilation | `POVMWeightsProbability`, `LF4.povm_born_frequency_volume_canonical` | proved | | T11 | Composite probabilities | joint Born-frequency capstones | per-instance (A6-gated for sector-intrinsic) | | T12 | Entangled predictions | = T14 | per-instance | | T13 | Contextuality | `NoNonContextualValuation`: Cabello-18 / Mermin-Peres / GHZ; `general_ks_noNonContextualValuation` | proved (+ general KS) | | T14 | Bell correlations | `NoLocalHiddenVariableTable` (CGLMP), `HasTsirelsonSeparation`, `bell_general_separation`, `lhv_chsh_le_two`, `qm_chsh_le_tsirelson` | proved (+ universal bounds) | | T15 | No-signalling | `HasNoSignalling` (singlet), `tensorSector_no_signalling` (operator) | proved | | T16 | Two-path interference | `HasBornInterference` (Hadamard test) | proved (derived, not postulated) | ## 3. The connectivity chain (L1–L9) See [`connectivity-manifest.md`](connectivity-manifest.md) for full evidence. | Link | Claim | Status | |---|---|---| | L1 | Kähler geometry ⇒ sector fields | DISCHARGED 2026-09-10 (W1; was PARTIAL) — volume forced; 2-form's **pointwise** compatibility core is the concrete field `kahler_pointwise : IsFubiniStudyKahler` (F-04 tightening 2026-08-06); manifold closedness `dω=0` / `ω^{∧n}/n!=μ_FS` formalised 2026-09-08 and wired 2026-09-10 (`LF4/SectorManifold.lean`) | | L2 | Σ+Φ+π ⇒ projected flow | CONNECTED | | L3 | projected flow ⇒ Schrödinger | CONNECTED — general `N`, arbitrary `H`; C¹-Stone derivation EXERCISED on the real nonzero generator (`manyToOneSchrodingerSetup_schrodinger_derived`) | | L4 | genuine `Φ ≠ id` inhabitant | CONNECTED — `rotationSetup`, `manyToOneSetup`, `unitaryFlowSetup` (4 total) | | L5 | sector ⇒ Born frequencies | CONNECTED (structural) | | L6/L8 | ONE object, both pillars, many-to-one `π` | CONNECTED — `manyToOneSchrodingerSetup_both_pillars` | | **L9** | ONE model: dynamics + measurement + records + Born + update | CONNECTED — **`FiniteQMClosure` / `unifiedFiniteQMClosure`** assembles all 11 proved-on-model facts into ONE tiered record, each field discharged by its source lemma; projective-sector posit / QM adapters / open residue documented, not encoded as fields | | **L7** ★ | Born weights derived FROM the flow | **OPEN — = SO-1** (boundary proved; ergodic face sharpened) — the sector is posited; a single flow provably cannot pin `μ_FS` (`flow_admits_invariant_ne_fubiniStudy`, `obsFlow_not_uniquely_ergodic`). The gap to `BornFromFlow` is exactly the Mathlib-absent pointwise Birkhoff theorem. Typicality itself is forced by the LLN, not this route (Papers A/B) | ## 4. The forward reconstruction — what each pillar delivers (on the witness) * **Born rule** (LF1–LF4): the Born weight `‖⟨eᵢ,ψ⟩‖²` is a Fubini–Study typicality volume; i.i.d. FS-typical trial frequencies converge a.s. to it, Gleason-free, general `N`, including **general POVMs** (`povm_born_frequency_volume`, canonical Naimark dilation from CFC `√Eᵢ`). * **Schrödinger dynamics** (the W-series, LF4): given the Kähler sector interface, Wigner rigidity + Bargmann branch selection + phase lift + a C¹ (and continuity-only) finite-dim Stone theorem force the projected flow to be `exp(-itH)` on rays. Instantiated non-trivially at general `N` (`manyToOneSchrodingerSetup`); the C¹-Stone derivation is EXERCISED on the real object at general `N` with arbitrary Hermitian `H` — `manyToOneSchrodingerSetup_schrodinger_derived`. * **Measurement** (LF5 + SigmaLayer): a measure-preserving von Neumann de-isolation flow realises the Naimark dilation; the per-microstate pointer outcome is defined a.e.; frequencies are Born. **Honest caveat (MD-1):** the outcome *cells* are the dilated Born regions `bornRegion ψ'` — preparation-indexed, not the context-fixed `Ωᵢ(M)` of Paper C A7. This is a preparation-indexed operational witness. * **Records / state update** (SigmaLayer): records are time-indexed measurable events (`flowedSemantics`), with probability conserved and flow-covariant under isolation; the record conditioning equals the Lüders update as an OPERATIONAL EQUIVALENCE (`conditioning_luders_effect_equivalence`), resting on the proved weight agreement `onticRegion_measure_eq_born`. * **Entanglement / non-locality** (LF6): Bell-forced non-factorisation in the Σ-engine at full generality — CGLMP for every `d`, GHZ/Mermin for every party count `n`, no-signalling; the universal bounds (`lhv_chsh_le_two`, `qm_chsh_le_tsirelson`, `cglmp_lhv_le_two`, `bell_general_separation`). * **Open-system dynamics** (LF6-2): the two canonical qubit dissipators as continuous quantum dynamical semigroups — T2 dephasing (`dephasingChannel`) and T1 amplitude damping (`dampingChannel`). ## 5. Adjacent arms (honest scope) * **Empirical / CSD arm** — thermodynamics (TH1–TH4), CV track (finite position/momentum, approximate CCR, oscillator spectrum), algorithms (Deutsch–Jozsa, Simon, BV, Grover, QFT, full Shor), metrology (Ramsey, Heisenberg, quantum Fisher), QEC, contextuality, channels/decoherence. These share the same complex-sector + Born + unitary primitives but are independent theorems consuming them, not a formal cascade from the SigmaLayer ledger. * **Reversible-arithmetic / circuit arm** — the general reversible quantum-arithmetic library (`Mathlib/QuantumInfo/Reversible/`) presupposes the unitary-evolution pillar; its theorems are circuit semantics and cost bounds, NOT a QM reconstruction. (The ecdsa.fail / ECDLP resource-estimation track was extracted to its own repository 2026-07-20 and is no longer present here.) ## 6. Axiom hygiene * Foundational triple only on every SigmaLayer/LF headline pin. * **Zero imported axioms** (since 2026-07-21). The last one, `busch_effect_gleason`, was proved — now `OperationalPackage.effect_gleason_representation` (`LF2/EffectGleason.lean`), foundational triple. See [`../AXIOMS.md`](../AXIOMS.md) §2.2. * No global `axiom` declarations in the Σ-layer; no `sorry`/`admit`. * Static guards (connectivity, sector-linkage, axiom-imports, claims) pass and are run in CI. ## 7. The honest frontier — what is NOT claimed *(Actionable open items in [`BACKLOG.md`](BACKLOG.md); the mission frame in [`CSD-CHARTER.md`](CSD-CHARTER.md).)* **Framing (read first).** **Σ is the floor and is everything** — deriving it is a *non-question*, not an open problem. QM (the T1–T16 inventory in §2a) is the **calculation engine** the Ω-region/volume-ratio structure computes. The goal now is to **complete the reconstruction of QM from Σ and Ω-regions** — QM genuinely *arising from* the ontic surface, not accumulated on the epistemic side. * **The record layer (MD-1) — the KINEMATIC interface is BUILT (2026-07-25), formalized end-to-end; MD-1 itself is NOT discharged.** ⚠️ *Updated 2026-07-31: the preparation-indexing is fixed on the fibre (`GlobalBasin`, `GlobalRecordClosure`); the base-only question stays parked, and `H_int` is still open, so MD-1 is still not discharged.* ⚠️ *Corrected 2026-07-28 (external review): the partition is preparation-indexed, so general-`N` A7 is open (`N=2` is done, `LF4/QubitBorn.lean`); the de-isolation Hamiltonian is not constructed (`DeIsolationInteraction.basin_rate` is a hypothesis field); and the fibred Σ below is a measurable record model, not a proven A1 ontic sector.* Measurement = de-isolation, on the model space **Σ = base × fibre** (`RecordLayer/FibredSigma.lean`): the **base** `CPN n` is the *epistemic* projective point (pinned to `[ψ]` for a sharp prep), the **fibre** carries the *ontic* record coordinate. The measurement is `context + unknown microstate → record` (`RecordLayer/Measurement.lean`): the context fixes the basin partition, the unknown microstate selects the basin it occupies, and the combined result is the P5 `RecordSemantics` record (`RecordLayer/FibreRecord.lean`, `RecordLayer/ProjectiveRecord.lean`). The outcome probabilities are the **Kähler torus moment map** (`RecordLayer/MomentMapRace.lean`, `bornRate_eq_momentMap` — read off the context, not injected; the choice of rate field is forced by torus *generation* (`RecordLayer/CellLawForced.lean`, `torusGenerated_eq_momentMap`), not by invariance), and the frequencies are the **law of large numbers over the unknown initial microstate** (`Measurement.bornMeasurement_frequency`, `ProjectiveRecord.projRecord_frequency` — no dynamical postulate; randomness = ignorance of the initial condition). This is realized on the corpus's actual Σ: `KSigmaRecord.born_frequency_region_eq_record` proves the region `FiniteQMClosure.born_frequency` lands in **is definitionally the record-layer event**. Arbitrary observables via `RecordLayer/BasisMeasurement.lean`. All foundational-triple, no `sorry`, axiom-pinned. **Only residue:** the pinned `unifiedFiniteQMClosure.records_time_physical` field still *names* the coarse `vnPointerOutcome` (a block-sum of these events); rewriting it carries no new theorem. See `record-layer-plan.md §4`. **Constraining Σ (legitimate — *not* deriving it).** Σ is the floor (deriving it is a non-question), but it is not directly seen, so **constraining its structure as tightly as possible — from above, from what it must do — is valuable work**: the more forced/less arbitrary the hidden substrate, the stronger the theory. The structure-forcing results are exactly this. μ_FS is the **unique** SU(n)-compatible measure (Paper B), realized in Lean by `fubiniStudy_forced_by_symmetry` / `LocalisedTypicality.lean` (`region_measure_symmetry_forced`) — so Σ's typicality measure is *forced*, not chosen. The companion no-go (`SectorPostulateNoGo.lean`, `flow_admits_invariant_ne_fubiniStudy`) records only that a single *epistemic unitary* flow does not time-average to μ_FS — expected, and irrelevant to CSD's mechanism (typicality is repeated-preparation ignorance over `Ω₀` on Σ, not epistemic time-averaging). These are **constraint work on Σ** — pinning the hidden substrate down — not a *research frontier*, and NOT the retired "derive Σ / SO-1 / L7 Born-from-flow" non-question. **Remaining formalization gaps (engine-level, not the thesis):** * **A6 "why ⊗"** — SUFFICIENCY + NECESSITY both proved (`SigmaLayer/TensorSolved.lean` `composite_is_tensor_product`; `SigmaLayer/TensorReconstruction.lean` `compositeAlgReconstruction`, `composite_dim_eq`). Residual: local tomography itself is the one operational axiom that cannot be derived from nothing (it singles out quantum `⊗`) — since 2026-09-02 stated in record vocabulary and proved equivalent to the generation premise (`SigmaLayer/TensorTomography.lean` `recordLocallyTomographic_iff_adjoin_eq_top`; registered boundary `R-017`); and the general non-factorising ontic composite (A6 as a primitive) is not reconstructed. * **A1 / KG-1** — ✅ *both halves proved, the constants visible* (reconciled 2026-10-27, BACKLOG #210; this bullet had said the volume identity remained, while §2a had recorded its discharge since 2026-09-11). `dω = 0` is `fsForm_mextDeriv` (2026-09-07). The volume identity, in the corpus's wedge convention: the top-power measure is `ω_FS^{∧n} = (4π)ⁿ μ_FS` (`fsVolume_eq_smul_fsMeasure`), the Riemannian volume is `vol_g = ω^{∧n}/n!` (`riemannianVolume_fsMetric`), so the textbook `ωⁿ/n! = μ_FS` holds as `ω^{∧n}/n! = ((4π)ⁿ/n!) · μ_FS` (`riemannianVolume_fsMetric_eq_smul_fsMeasure`) — equality up to that stated constant, `μ_FS` being the probability normalisation. On the arena `ℂℙᴺ × T²` the Liouville measure of `arenaForm^{∧(N+1)}` is `(N + 1)(4π)ᴺ · (μ_FS ⊗ vol_{T²})` (`arenaVolume_eq_ofReal_smul_kMuL`, BACKLOG #32). * **A5 approximate regime** — ⚠️ *narrowed 2026-09-05 (CR-11): the predicate is now populated at `ε > 0`.* `SigmaLayer/MovingFibreWitness.lean` gives `movingFibreEnergy`, an `EpsProjectable _ ε` witness for every `ε ≥ 0`, together with `movingFibreEnergy_not_projectable` — for `ε > 0` it is **not** `EpsProjectable _ 0`, so it genuinely does not factor through `π` and is not an `ε = 0` instance wearing a parameter. Every earlier instance sat at `ε = 0`, which by `epsProjectable_zero_iff` is exactly the exact case, so the approximate regime had no inhabitant at all. ▸ *Closed as a dynamical statement 2026-10-29 (BACKLOG ~~#208~~, `SigmaLayer/ActivatedRegime.lean`):* the derivative form `sup‖d(δH)|_V‖ ≤ ε` is stated on the arena and proved to give oscillation `ε/2` (`epsProjectable_of_deriv`); the activated write's energy `−(2/ħ)(J⟨X⟩ + g f(θ₂) n_B)` is `EpsProjectable` at the sharp `ε = (2|g|/|ħ|)·osc f` (`epsProjectable_activatedEnergy_iff`), not at `0`, and generates #194's `activeFlow`; and `quantum_effective_shadowing` is carried from matrices to the arena (`coupledFlow_shadowing`), so every sector expectation of that flow is shadowed by the exactly projectable sector flow to within `‖O‖·ε·T` (`activated_expectation_shadowing`). The moving-fibre witness itself is still predicate only; the dynamical statement is carried by the activated write. * **LF6-9** — the general Lindblad generator + complete positivity (the two bounded dissipators are done). * **IP-1** — identical particles / spin-statistics: two particles in the corpus (SL-T8 and BACKLOG ~~#206~~, `SigmaLayer/IdenticalParticles.lean`, 2026-10-28; the symmetrisation postulate shown consistent, not derived); `n` particles in the corpus (BACKLOG ~~#214~~ and ~~#217~~, `SigmaLayer/ManyIdenticalParticles.lean`, 2026-10-28; the Slater determinant vanishes exactly for dependent orbitals); only `±1` exchange statistics for states whose ray permutations keep (BACKLOG ~~#216~~, `SigmaLayer/ExchangeStatistics.lean`, 2026-10-28); second quantisation at finite dimension in the corpus (BACKLOG ~~#215~~, `SigmaLayer/FermionFock.lean`, `SigmaLayer/BosonFock.lean`, 2026-10-28) and the operator dictionary for fermions and bosons, one- and two-body operators included (BACKLOG ~~#220~~–~~#223~~, `SigmaLayer/FermionDictionary.lean`, `SigmaLayer/BosonDictionary.lean`, `SigmaLayer/OneBody.lean`, `SigmaLayer/TwoBody.lean`); the mixed-symmetry sectors described (BACKLOG ~~#219~~, `SigmaLayer/MixedSymmetry.lean`: orthogonal complement and dimension for every `n`, two isomorphic copies for three particles); spin–statistics proper and the general isotypic decomposition open (BACKLOG #218, #224). ### 7a. Settled non-goals — do NOT re-litigate these Two positions are **decided** and recorded here so they are not re-argued. Both are backed by machine-checked facts in the corpus. * **NG1 — CORRECTED 2026-07-24 (the earlier wording was wrong and misleading).** Do **not** call the single-trajectory / deterministic-flow account a route "CSD rejects" — **CSD *is* a single-trajectory theory** (Paper D §4.2: physical reality is one trajectory `ω(t)` on Σ). What *is* a proved dead-end is time-averaging one infinite *epistemic unitary* trajectory on `ℂℙⁿ⁻¹` (`obsFlow_not_ergodic` / `obsFlow_not_uniquely_ergodic`, `LF4/TypicalityForcing.lean`; `flow_admits_invariant_ne_fubiniStudy`, `SigmaLayer/SectorPostulateNoGo.lean`), so building out the epistemic ergodic scaffolding (`SigmaLayer/UniqueErgodicity.lean`, `BornFromFlow`, `IsErgodicForOutcomeRegions`) is **not progress**. CSD's typicality is **repeated-preparation ignorance over the prepared region `Ω₀` on Σ** (Paper D, "Note on repeated preparations"; classical-statistical-mechanics style) — each experiment is one trajectory; statistics come from not knowing the exact microstate in `Ω₀`. That is neither time-averaging one epistemic trajectory nor "fresh i.i.d. preparations *instead of* single-trajectory." Do not round the narrow epistemic-`ℂℙ` no-go up to a rejection of the ontology, and do not treat "deriving Σ" as the residue (Σ is the floor — a non-question). * **NG2 — the Busch effect-Gleason axiom is NOT needed for CSD's core claim; discharging it is cosmetic.** CSD's ontic Born rule is **Gleason-free**: it is a Fubini–Study / Duistermaat–Heckman *volume* (`bornRegion_fs_measure`, `born_frequency_convergence_N`). The former axiom `busch_effect_gleason` entered only the **operational effect/POVM stratum**, off the reconstruction path. It was **proved in-repo 2026-07-21** (`OperationalPackage.effect_gleason_representation`), taking the imported-axiom count to zero — an **audit-posture** improvement, NOT a strengthening of the CSD reconstruction. ⚠️ **What "Gleason-free" does and does not say** (governance wording, 2026-09-05). It says the Σ-side road shares **no machinery** with Gleason or Busch — a provenance claim, and a machine-checked one, by two guards: `check-import-negative.sh` (46 modules — the claim that `LF2/EffectGleason` is absent from the import closure) and `check-gleason-free.sh` (11 modules, 259 declarations, none reaching `effect_gleason_representation` in the constant graph). ⚠️ 59 module headers assert Gleason-freeness, so the two guards together cover most but not every assertion; the uncovered remainder rests on the axiom sweep. It does **not** say the Born form is derived from nothing. The record layer's rates rest on the **cell law**, which is a posit *with a characterisation*: a context field whose rates **generate** the coordinate phase rotations is exactly the moment map (`torusGenerated_eq_momentMap`), while torus **invariance** alone does not (`rate_field_not_forced_by_torus_symmetry` exhibits a rival field). Two adjacent uniqueness results support the surrounding structure without bearing on the cell law: `μ_FS` is the unique `U(N)`-invariant probability measure (`fubiniStudy_forced_by_symmetry`), and additivity on effects forces the trace form in the operational stratum (`effect_gleason_representation`). The record layer then realises the Born weights as volume fractions under a uniform fibre variable (`globalBasin_born`) by a route that uses neither — and the cell-law characterisation is **single-context**, so it does not import noncontextuality either. ⚠️ That is not a claim that no Gleason-style characterisation could exist: the frame-function route was **declined, not refuted**. See `specs/POSITS.md` Posit 1 and `specs/cell-law-scoping.md`. ## 8. Bottom line The corpus proves the **QM calculation engine on a concrete projective witness** — the *consistency floor*: the full T1–T16 target inventory inhabited, axiom-clean, Born = Ω-region volume ratio, on one witness model with μ_FS and `exp(-itH)` built in. **Σ is the floor and is everything; deriving it is a non-question.** The goal now is to **complete the reconstruction of QM from Σ and Ω-regions** — making QM genuinely *arise from* the ontic surface. The **record layer** (MD-1) is now **formalized in Lean** (2026-07-25; see §7): measurement as `context + unknown microstate → record` on the base×fibre Σ, Born = the law of large numbers over the unknown microstate, outcome probabilities = the Kähler moment map. ⚠️ *Updated 2026-07-31:* the **record layer's own** closure is now context-fixed (`GlobalRecordClosure` on `KSigma`, events built from a `ContextField` rather than from `ψ`). The residual is that the **pinned `FiniteQMClosure`** still names the preparation-indexed `bornRegion ψ'` — a separate migration on the `productDynamics` engine, not a record-layer gap. The genuine near frontier is now the **extensions** — continuous spectra (CV), relativistic locality, identical particles. (The earlier "SO-1 = derive the sector, the central frontier" framing is a retired error — §7.) ⚠️ *Frontier vocabulary superseded 2026-09-05, and this sentence with it.* `specs/POSITS.md` ("What 'frontier' means here") separates **permanent boundary** / **open mathematics** / **open foundations**. The extensions are breadth, not the frontier; the surviving foundational item is Posit 1's discharge condition — deriving `IsTorusGenerated` from the de-isolation dynamics. §1 of this file carries the corrected statement. References: [`connectivity-manifest.md`](connectivity-manifest.md), [`future-work.md`](future-work.md), [`../AXIOMS.md`](../AXIOMS.md), [`../CsdLean4/SigmaLayer/Adapters.lean`](../CsdLean4/SigmaLayer/Adapters.lean).