# specs/ index — orientation map for plans, todos, and references > **⚠️ [`BACKLOG.md`](BACKLOG.md) is the single canonical open-items list.** Everything > below it is either the historical execution record (the per-phase `*-plan` files) or the > status/claims layer. For "what is next", read `BACKLOG.md` — nothing else. **Start here.** This indexes every planning / todo / reference doc in the corpus, with one-line status. Updated 2026-08-02. When starting a session, read this first, then the LIVE doc for the tranche you are on. ## Live (actionable) — pick up work here | Doc | What it is | Status | |---|---|---| | [`CSD-CHARTER.md`](CSD-CHARTER.md) | **★ The north star — read before framing any claim or scoping work.** **Σ is the floor and is everything** (deriving Σ is a non-question); **QM arises from the *epistemic* Ω-region partition {Ωᵢ(M)} on ℂℙ + the *ontic* typicality volume on Σ** (→ Born; dynamics + measurement follow). **The goal now is to complete the reconstruction of QM from Σ and Ω-regions** — not to accumulate more QM on the epistemic side. The **record layer** (MD-1) that does this is **BUILT** (2026-07-25, `record-layer-plan.md §4`): measurement = `context + unknown microstate → record` on the base×fibre Σ, outcome probabilities = the Kähler moment map, Born = the LLN over the unknown microstate. Enforced by the `csd-foundations` agent + `check-claims.sh`. | **REWRITTEN 2026-07-24** to the actual CSD ontology (the earlier "derive Σ / SO-1-as-goal" framing was a retired error). The anti-drift anchor. | | [`reconstruction-status.md`](reconstruction-status.md) | **★ The A1–A7 audit — "how far is finite QM from Σ from closed?"** Per-axiom rows (proved / discharged / closed-as-posit / explicitly scoped, each with theorem names), the §2a scoping decisions, and the net line. **As of 2026-08-03: no genuinely open A-row at the formalisable level** — A2's Hamiltonian-generation sub-question (reopened 2026-08-02 by the second external review on the torus-flux correction) was CLOSED 2026-08-03 by the `ℂℙ^K` smooth pointer witness (`smoothWitnessClosure`, `rampedU_schrodinger`); the symplectic/moment-map *reading* stays the §2a-scoped prose boundary shared with A1/A3. A5/A6/A7 discharged; A2/A5 formalisable halves landed. | **THE status truth for the reconstruction.** Read §2a before claiming anything open or closed. | | [`sigma-fibre-contextuality.md`](sigma-fibre-contextuality.md) | **★ Structural lesson about Σ (2026-07-27).** Born = ontic typicality volume for all `N` (theorem); but *where measurement contextuality lives in Σ* is dimension-dependent — on the projective base for the qubit (`qubitBorn`, proven), and in the fibre for `N ≥ 3` **by decision, not by proof** (base-only is constrained by a machine-checked chain but remains open in both directions — retracted 2026-07-28; NOT Gleason — CSD is contextual). Not a defect: a constraint on Σ's structure. The fibre resolution (softmax/Gumbel) is CSD-legit; deriving it is the open frontier. | **READ before touching general-`N` context-fixing.** Prevents re-litigating the `N≥3` wall. | | [`record-layer-plan.md`](record-layer-plan.md) | **The record layer (MD-1) — kinematic interface BUILT 2026-07-25 (§4); MD-1 CLOSED 2026-08-31** (author decision; the machinery is complete, and which physical interaction an apparatus realises is a modelling input outside the Lean claim by design — `R-015`). General-`N` A7 was discharged 2026-08-02 (`reconstruction-status.md` A7, fibred reading canonical). Residuals: the chart→arena transport of Hamiltonian generation (`R-016`, open mathematics) and the A1-sector status of the fibred Σ.** Measurement as `context + unknown microstate → record` on the base×fibre Σ, the piece that completes "how QM works" from Σ+Ω. Formalized end-to-end (foundational-triple, no `sorry`, axiom-pinned): fibre Born partition, P5 `RecordSemantics`, rates = the Kähler moment map (`MomentMapRace`), Born = LLN over the unknown microstate (`Measurement.bornMeasurement_frequency`), the base×fibre ontic Σ (`FibredSigma`), realized on the corpus's actual Σ (`KSigmaRecord.born_frequency_region_eq_record`), arbitrary observables (`BasisMeasurement`). **★ 2026-07-26: the qubit context-fixed Born rule is now formalized end-to-end** (`LF4/QubitBorn.lean` `qubitBorn`) — the epistemic hemisphere partition `{H±(n)}` is provably context-fixed (prep-independent) and Born is derived from the ontic μ_FS typicality; see §2. | **Kinematic interface BUILT 2026-07-25; qubit A7 Born DONE 2026-07-26.** *Addendum 2026-08-02:* **general-`N` A7 DISCHARGED** (fibred reading canonical — `GlobalBasin`/`globalBasin_born`, author decision; see `reconstruction-status.md` A7 row) and the **dynamical measurement layer is COMPLETE through the join arena** (records created/persist/exclusive, dynamical Born, rank-one Lüders on the swap arena, degenerate Lüders on `ℙ(ℂ^{N+N})` — `SwapClosure`/`UnifiedArena`/`JoinArena`). Read with `CSD-CHARTER.md`. | | [`pointer-witness-plan.md`](pointer-witness-plan.md) | **The compact Kähler pointer witness `ℂℙ^K` — the smooth-Hamiltonian measurement dynamics plan** (the reopened A2 row's confirmed route, author decision 2026-08-03: Paper C A2 / TN6's smooth architecture is retained and this witness must satisfy it). Arena `ℂℙ^{N-1} × T²_λ × ℂℙ^K`, continuously modulated Hermitian coupling, records up to a stated `ε` (the `no_everywhere_correlation` price), globally-Hamiltonian generation at the formalisable level. Brick ladder 0–5 with the reuse map. | **COMPLETE 2026-08-03 — all five bricks landed the same day** (eight `SigmaLayer/Pointer*` modules; `smoothWitnessClosure`, `rampedU_schrodinger`). Kept as the execution log; the Lüders composition and ε-Born LLN extensions live in `BACKLOG.md`. | | [`connectivity-manifest.md`](connectivity-manifest.md) | **THE single source of truth for end-to-end connectivity claims.** Per-link status (Kähler geometry → Σ → flow → projected flow → Born / Schrödinger), each row CONNECTED (with a named backing theorem) or BROKEN/CONDITIONAL/TRIVIAL-WITNESS-ONLY (with the exact gap), plus the sequenced fix course C1–C6. No doc may claim a stronger connection than a CONNECTED row here. | **CREATED 2026-07-07 after a provenance audit found the "one posited object ⇒ both pillars" claim was NOT realized in code.** Enforced by `scripts/check-connectivity.sh`. **Current state (2026-07-26): L4/L5/L6/L8/L9 CONNECTED** (a genuine `Φ≠id` sector yields both pillars on one model; the record layer is built and Born is ontic-grounded, L5); L1 is the Kähler-form Mathlib gap; L7 ("derive Σ") is retired as a non-question. Read before making any pillar/connectivity claim. | | [`active-todo.md`](active-todo.md) | **THE persistent session work queue** — the full task list (Category / Complexity / Status / Blocked-by) that survives session loss; re-seed the in-memory task tracker from it. Includes the #16 D1c plan, and the #15 open-system-empirical next-tranche scoping. | **SUPERSEDED by [`BACKLOG.md`](BACKLOG.md)** (the banner above): last updated 2026-06-29, kept as historical record. Do not seed work from it. | | [`metrology-plan.md`](metrology-plan.md) | **Quantum sensing / metrology validation program** (parameter estimation, not static states): A1 Ramsey interferometry (volume-preserving phase flow → `cos²(φ/2)` fringe, reuses `MalusVolume`), A2 QFI = Fubini-Study metric, A3 Heisenberg `1/N` via dilated manifolds, A4 decoherence as open symplectic drift. | **A1 (Ramsey) / A2 (QFI = FS-metric pullback) / A3 (Heisenberg `F_Q^GHZ = N²` via the GHZ probe) DONE 2026-06-28; A4 (decoherence = D1) open.** A3: `Heisenberg.lean`, `ghz_qfi = N²` vs `sqlQFI = N`, `heisenberg_advantage` (N-fold), genuine `N`-qubit `Fin (2^N)` GHZ family + genuine derivative, foundational triple only. A4 tied to D1. | | [`tn4-measurement-architecture.md`](tn4-measurement-architecture.md) | **Orientation map for the measurement stack** (TN4): the single architectural view — pure → density → channels → {projective, POVM} → Naimark → Born → CSD volume → LF5 dynamics — annotated with the real theorem names, the two readings (QM-validity vs CSD-ontic; operational vs ontic stratum), where Born comes from (the DH cluster, derived not postulated), and the honest D1 / posited-Σ / `Φ=id` residue. Descriptive, not a plan. | **Reference doc** (written 2026-06-27). Read first when picking up measurement-theory work. | | [`carve-out-plan.md`](carve-out-plan.md) → **D1** | **The open frontier.** Exercise real measurement *dynamics* on `Σ`; the measurement-dynamics frontier (entangled de-isolation dynamics; the record layer itself is now built). | **OPEN at the deeper strata.** `Φ≠id` (non-collapse) done 2026-06-05 (`LF4/ObservableFlow.lean`); collapse semantics DECIDED 2026-06-09 (§6: the de-isolation reading); **the measurement-event half is now exercised at the single-system projective tier — LF5 COMPLETE 2026-06-11** (`measurement_flow_born_frequency`); **the per-microstate outcome map is DONE 2026-06-14** (LF5-F: `bornRegion_pairwiseDisjoint` → `vnPointerOutcome` → `measurement_flow_outcome_frequency`, a single union event per pointer). Remaining: entangled/non-local de-isolation dynamics; `SectorData` instances still `Φ = id`. (The record layer / MD-1 is now BUILT — 2026-07-25.) | | [`frozen-base-obstruction-scoping.md`](frozen-base-obstruction-scoping.md) | **Scoping for the `H_int(M)` generation brick** (2026-09-01, `csd-foundations` checked). The negative result the frozen-base witnesses invite: generation by a `C¹` joint scalar plus a pointwise-frozen base forbids exact outcome correlation. Carries the wall-check (base-marginal invariance is Diracs, not geometry; back-reaction is NOT forbidden — `IsJointLift.moment_marginal_unchanged`; the base-moving quadrant is already occupied by `joinSwap_eq_flowTimeOne`) and the withdrawn-draft table. | **ALL FOUR BRICKS DONE** (0–2 2026-09-01, 3 2026-09-02) (`ChartIntegralCurve.lean` 5 pins; `FrozenBase.lean` 3 pins; brick 2 in three halves — `UntriggeredFlow.lean` 4 pins, `UntriggeredReadout.lean` 3 pins, `UntriggeredVolume.lean` 4 pins: chart-volume preservation via Mathlib's Haar/determinant API + uniqueness; brick 3 — `ArenaTorus.lean` / `JointLift.lean` / `HamiltonianShift.lean` + staged `InvariantTwist.lean`, 13 pins: a back-reacting `IsJointLift` **on the arena** with Liouville preservation via an invariant-twist lemma, no disintegration; the shift it carries generated chart-level by the stroke Hamiltonian). Bricks 0–2 chart-level, brick 3 arena-level except generation; `R-016` (now exactly the manifold identification `ι_X ω = dH`) and `R-015` (the boundary) both stand. | | [`way-theorem-scoping.md`](way-theorem-scoping.md) | **Scoping for the Wigner–Araki–Yanase brick** (2026-09-02, expert-review row A). The QM-side conservation-law measurement theorem (absent from the corpus), stated with the Yanase / repeatability side condition and a SWAP sharpness witness; the CSD reading as a *scope* theorem at the `IsJointLift` level (the stroke is a skew product on the arena, not a tensor-product isometry with an additive conserved `L`, so WAY's hypotheses have no record-layer instance, while the LF5 `vnUnitary` tier meets them with an engineered `L` and the conclusion holds there — the "conserves only what it reads" framing was withdrawn, §5); two negative findings — the record trilemma is **not** WAY in disguise, the `ε` of `ε`-Born is **not** a WAY bound. Not an `H_int` brick. | **Brick 0 LANDED** 2026-09-02 (`Empirical/QM/WignerArakiYanase.lean`, 7 pins); **Brick 1 LANDED** 2026-09-02 (`Empirical/CSD/WignerArakiYanase.lean`, 7 pins: `no_joint_hilbert_map` — the pointer image of a joint lift is not a function of the Hilbert data; the quantifier is a robustness one, weaker than WAY's); brick 2 (quantitative WAY) deferred behind the Ozawa error–disturbance row. | | [`generation-from-records-scoping.md`](generation-from-records-scoping.md) | **Scoping for brick 2 of the Q11 arc — generation from records** (2026-09-02, `csd-foundations` checked, 18 findings folded). The generation premise `hgen` of `compositeAlgReconstruction` / `composite_dim_eq` (the "why `⊗`" reconstruction) had been *called* local tomography in prose with no theorem behind it. The brick is a **premise conversion** (the Q11 template of `StatisticsRigidity.lean`): record-level local tomography — equal joint record rates in every pair of local bases force equal composite densities — is proved equivalent to `hgen`, both directions, under locality + star-preserving embeddings; the Kronecker composite satisfies it. ⚠️ Not a derivation of `⊗` from Σ (real QM is the standing counterexample; classical probability shows local tomography does not select `ℂ`); constraint work one level above Σ, not MD-1 progress. | **LANDED 2026-09-02** (`SigmaLayer/TensorTomography.lean`, `recordLocallyTomographic_iff_adjoin_eq_top`, `compositeAlgReconstructionOfRecords`, `kronecker_recordLocallyTomographic`; `productRecordRate_eq_bornRateBasis` in `StatisticsRigidity.lean`; Cat-1 `KroneckerAlgHom.lean`; 20 pins). The surviving posit is boundary `R-017`. | | [`local-friendliness-scoping.md`](local-friendliness-scoping.md) | **Scoping for the Local Friendliness row** (2026-09-04, expert-review row C / D12; `csd-foundations`-checked, 14 findings folded). ⚠️ §2: three statements are all called "locality" — the corpus proves (i) no-signalling CONDITIONALLY, denies (ii) Bell parameter independence, and LF is about (iii), a **joint** constraint nothing in the corpus addresses. §3: the coherent branch **is** available to CSD — Alice's setting is a forward interaction, not a time reversal, and `U†` is a unitary like any other; `sequential_no_revival` is about a later *measurement* and is narrower than its own prose. So the elimination fires and there is no empirical exposure — resting, stated plainly, on the same unitary-realisability posit as the whole gates layer. §4's deliverable is the §2 table proved as three ordered predicates with a strictness witness, NOT the LF inequality. | | [`ozawa-scoping.md`](ozawa-scoping.md) | **Scoping for the Ozawa error–disturbance twin** (2026-09-04, expert-review row B; `csd-foundations`-checked, 15 findings folded). Target is Ozawa 2003's three-term relation ONLY — the naive Heisenberg form is *not universally valid* rather than false, and no Lean claim is made about it. ⚠️ Two decisions to read before building: the abstract-`tensor` vocabulary (§3a — Mathlib `⊗[ℂ]` would strand WAY brick 2), and the CSD side being a **scope theorem**, not a volume-ratio transport bundle (§5 — `Empirical/CSD/Uncertainty.lean` is SCHEMA-MISMATCH and must not be copied). Breadth row, not reconstruction path. | | [`colbeck-renner-note.md`](colbeck-renner-note.md) | **Positioning note: which premise CSD denies in Colbeck–Renner** (2026-09-04, expert-review row D). Parameter independence, at the `Σ` level — the same escape as Bell, and a theorem there (`no_product_partition_realises_singlet`, CL-020). ⚠️ **Not** by denying free choice: CR's "free choice" bundles parameter independence with measurement independence (Ghirardi–Romano 2013, Leegwater 2016), and the corpus KEEPS measurement independence as an explicit premise (`LF3/OperationalNoSignalling.lean`). Positioning prose, not a formal result — the Lean form needs the chained-Bell family and is deliberately not queued. | | [`records-to-spacetime-scoping.md`](records-to-spacetime-scoping.md) | **Scoping for BACKLOG #38 (records to spacetime) and #39 (Σ-local, spacetime-nonlocal)** (2026-09-19; the programme's spacetime reading, written in the repository for the first time at the author's instruction). What a record is in Lean (a context-fixed basin with no location and the flow parameter as its time); four candidate relations between records assessed; what the finite compact arena can and cannot support; the theorems any account must respect; row 39 restated as a conjecture about whether the tensor cut is emergent, with (i) bounded influence and (ii) non-signalling Bell correlations already theorems relative to the cut; four bricks priced (`ST-1` the influence preorder on records, S–M; `ST-2` entanglement distance, M–L; `ST-3` the macroscopic-coordinate projection, gated on the author naming it; `ST-4` the reading into the documents, S). **Finding: no Lean shape for emergence proper, and the missing piece is a decision, not a lemma.** | SCOPED 2026-09-19 | | [`unified-record-spacetime-plan.md`](unified-record-spacetime-plan.md) | **The contract of the programme "From `Σ` to physical spacetime"** (milestone M0; BACKLOG ~~#350~~). Three registers for the shared arena `Σ_* = ℂℙ^{2M+1} × T²` of ~~#349~~: what is **assumed** (`UA-1`–`UA-20`: arena, measure, typicality reading, apparatus preparation, coupling, the cell law, the flow parameter, the Dirac propagation law, the stated laws and idealisations of the existing record and radar results, the clock's level and coupling, the two-pointer arena, the admitted interventions, the three-pointer arena, the clock on its own carrier, the composite carrier of the messenger and the clock, and the radar arena), what is **derived** (`UD-1`–`UD-241`, each with its Lean names and the arena it holds on — thirty-four on `Σ_*` since ~~#356~~, eight on its clock extension since ~~#358~~, twenty-six on the two-pointer arenas since ~~#370~~, two more on `Σ_*` since ~~#371~~, eight on the region-detector arenas since ~~#375~~ (one of them, UD-108, for the free packet), four on the pointer step and the three-pointer arena since ~~#369~~, three on the product arena of a clock beside a messenger since ~~#374~~, one more on the pointer step since ~~#374~~, fifty-nine on product rays, the composite carrier, the round trip and the four-pointer arena, about readouts, histories or real numbers on no arena (UD-131, UD-133, UD-135, UD-155, UD-156, UD-163, UD-164), on the clock arena or its uncoupled device (UD-137, UD-138, UD-141 to UD-144), or the bare five-pointer arena and its couplings (UD-139, UD-140, UD-167, UD-168), since ~~#404~~ (UD-117 for the free packet, UD-118, UD-125 and UD-127 on the region-detector arena, UD-122 for the arena alone), the rest on other arenas), and what is **owed** (`UO-1`–`UO-26`, by milestone, each with the test for a theorem that would only restate an assumption). ⚠️ §6: the cell-law premise is Posit 1 and is not derived from any dynamics; the calibration theorem of ~~#152~~ converts that premise and does not remove it. ⚠️ §5 carries two calculations bearing on gate A that are **not in Lean**. No Lean, no new posit, no new residue row. | CONTRACT (M0) | | [`unified-m6-feasibility.md`](unified-m6-feasibility.md) | **M6 feasibility and error budget** (2026-11-16; Task 1 of the research delivery plan for UO-16). Whether the existing packets, clocks and gates can meet every inequality of the top-down Lorentz statement together: the construction the theorems force (`B` and the messenger start beside one wall, `A` sits at the other), the family in `k`, the error ledger (total `≈ 16/k`, from tick discretisation), the reading theorems' size conditions, and what is not yet known in order of risk. Verdict: no obstruction found; three new required bricks. Reading and arithmetic, not Lean. | | [`unified-m6-audit.md`](unified-m6-audit.md) | **M6 audit** (2026-11-17; Task 12 of the research delivery plan for UO-16). The proof-dependency audit of `csd_operational_lorentz_from_records`: its three hypotheses, the chain from one point of the arena to the comparison of intervals, the plan's nine checks, the test of UO-16, and the findings F1 to F6. The author marked UO-16 complete on 2026-11-17 with those findings as the scope. | | [`unified-m7-feasibility.md`](unified-m7-feasibility.md) | **M7 feasibility on the unified arena** (2026-11-17; feasibility gate for UO-17). Whether the arena and theorems of M6 carry BACKLOG #296 to #298 to `d` space dimensions: which of M6's modules are free of the dimension and which are typed to the ring; the carrier layer on the torus of UD-206; four findings on paper (a two-component field in `d = 2` has no wall, walls on two axes need more components, M6's results do not transfer by the stability theorem, the exact reduction to the ring is not local); the obstruction (a messenger returns only from a flat wall, so moving observers who radar one event are collinear with it and a moving group has no trilateration chart); the three statements that can be had (S1 a resting group's spatial chart, S2 a moving observer's count against proper time, S3 a moving observer's radar from a wall); the decision for the author; an order of work that does not wait for it. Reading and arithmetic; no Lean theorem. | | [`unified-records-spacetime-task-plans.md`](unified-records-spacetime-task-plans.md) | **Plans for the outstanding tasks of the author's plan "Unified records, quantum measurement and emergent spacetime"** (2026-11-19, at the author's request). For each unfinished task (3 to 12, and the caused return of UO-26): what is in hand, the theorems to prove in order, the decisions that are the author's, and where the outcome may be an obstruction. A plan: it claims nothing, and its proposed theorem names do not exist. | | [`gleason-feasibility.md`](gleason-feasibility.md) | **Gleason's theorem, finite-dimensional — the feasibility pass** (2026-09-21). The projection form stated in the Mathlib-only tree (`Gleason/ProjectionPackage.lean`); Layers A (the frame function, restriction to completely real subspaces, the complex reduction by the equator computation) and C (from a quadratic form on the sphere to the unique density matrix) **proved**, Busch's proof re-routed through the shared engine; the core lemma on `S²` stated as `CoreLemma` (the one hypothesis of `gleason_representation_of_core`), sized against Cooke–Keane–Moran (≈3,000 lines, risk in Piron's geometric lemma and the Tychonoff extremal-value step), recommendation go in stages, BACKLOG #57(a)–(e). **All five stages landed 2026-09-21/22** (`Sphere`, `Warmup`, `Piron`, `SimpleFrame`, `Extremal`, `General`); `Gleason/Core.lean` joins them: **Gleason's theorem for `ℂᴺ`, `N ≥ 3`, proved** (`gleason_representation`, foundational triple). | | [`berry-phase-scoping.md`](berry-phase-scoping.md) | **Scoping for BACKLOG #10, Berry phase as holonomy** (2026-09-21). What the theorem is (Aharonov–Anandan: total, dynamical and geometric phase; the geometric phase is the holonomy of the canonical connection on the Hopf bundle), what Mathlib has (covariant derivatives; no parallel transport, holonomy or principal bundles), the bricks BP-1–BP-5 with prices, the CSD reading (the phase is a base quantity; the torus fibre carries no holonomy and none is claimed), and the outcome: BP-1 and BP-2 built the same day, BP-3–BP-5 are rows 54–56. | | [`feynman-continuum-scoping.md`](feynman-continuum-scoping.md) | **Scoping for BACKLOG #36(c), the continuum path integral** (2026-09-20, after rungs (a) and (b) landed). The real-time symbol `∫𝒟x e^{iS}` is not a measure (Cameron 1960); the two honest forms are Kac's Euclidean formula against Wiener measure and Nelson's time-sliced limit. Pin survey: the L² Fourier isometry, Gaussian measures and their convolution, and the Brownian-motion predicates with the weak Markov property are there; existence of Brownian motion, `C₀`-semigroups, unbounded operators, Feynman–Kac and oscillatory integrals are not. Six bricks priced, none above M–L: `FC-0` three probes (S), `FC-1` the Dyson series and the Trotter product formula for a bounded perturbation of a strongly continuous contraction semigroup (M–L, the engine), `FC-2` the heat semigroup on `L²(ℝ)` as Gaussian convolution (M), `FC-3` the time-sliced Wiener functional as an iterated Gaussian integral (M), ★★ `FC-4` Feynman–Kac, conditional on a Brownian motion (M–L), ★★ `FC-5` Nelson's theorem in operator form through the L² Fourier multiplier (L); the kernel form with the action in the integrand `FC-5′` was priced XL and alone outside the plan — **it landed 2026-10-01 at M–L (BACKLOG #44), and as an exact identity rather than the improper integral priced: see the FC-5′ row and decision D4, both corrected in the note**. **Design decision: `e^{−tH}` is defined by the Duhamel series around the free semigroup, so no generator is ever written.** Four decisions for the author (dimension, function space, waiting for Mathlib's Brownian motion, the action in the integrand). | SCOPED 2026-09-20; **FC-0 to FC-6 LANDED 2026-09-20** (Feynman–Kac and Nelson's limit with a unitary propagator, both theorems); residues FC-1′, FC-2′, FC-4′, FC-5″ priced in BACKLOG #36; **FC-5′ LANDED 2026-10-01** (BACKLOG #44), so the whole scoping note is discharged | | [`lf5-plan.md`](lf5-plan.md) | Staged plan for **LF5 (measurement dynamics, the D1 frontier)**: realise a projective measurement as a measure-preserving von Neumann de-isolation flow `Φ_vN ≠ id` on the joint `ℂℙ^{N·N−1}`, pointer regions = `blockProj` (apparatus basis, context-fixed), committed volume = Born, chained to frequencies. **Heavily reuses the POVM Naimark tranche** (block-Born = FS-volume already proved); the genuine new content is realising the static dilation isometry as a dynamical flow. | **COMPLETE 2026-06-11.** LF5-A (vN unitary, 06-09); LF5-B (measurement flow `Φ_vN ≠ id`, FS-preserving, 06-10); LF5-C (de-isolation realises the dilation: `vnNaimark`, flow carries `[ψ⊗a₀] ↦ [Vψ]`, 06-11); LF5-D (the **unconditional engine** `LF4/BornRegionUncond.lean` — retires the `hpos` genericity of the general-`N` + POVM engines, additively — + `vnDilation_pointer_volume` / `vnDilation_pointer_frequency`, 06-11); LF5-E (capstone, 06-11): **layer headline `measurement_flow_born_frequency`** (`LF5/Capstone.lean`) — `Φ_vN ≠ id` + FS-preserving + dilation realised ∀ preparations + pointer-block volume = Born + a.s. frequencies → Born, every unit `ψ`. All foundational-triple-only, AxiomAudit-pinned, Tier-A audited SOUND. **LF5-F (post-plan, 06-14): the per-microstate outcome map** — `bornRegion_pairwiseDisjoint` (`LF4/BornRegionDisjoint.lean`) → `vnPointerOutcome` → `measurement_flow_outcome_frequency` (`LF5/PointerOutcome.lean`, single union event per pointer) + the `aeece86`-owed `block_born_frequency_volume_event`; closes the outcome-function caveat. **LF5-G (post-plan, 06-15): QEC syndrome readout as a coarse-grained de-isolation flow** (`LF5/SyndromeFlow.lean`) — stabilisers computational-basis-diagonal ⟹ syndrome = fixed `synClass : Fin 8 → Fin 4`, `Φ_syn = measurementFlow 8` coarse-grained; `syndromeRegion_fs_volume` (block FS volume = syndrome Born weight = block sum of FS volumes), deterministic codeword syndrome + recovery; headline `syndrome_flow_born_volume`, projective/coherent-error tier (decoherence/partial-trace origin gated on the entangled tier). **LF5-G.2 (06-16, `LF5/SyndromeOutcome.lean`): syndrome frequency + per-microstate outcome map** — `syndrome_flow_born_frequency` (a.s. syndrome-class freq → `syndromeWeight`), `synOutcome` (= `vnPointerOutcome.map synClass`), `syndrome_flow_outcome_frequency` (single-event), completing the QEC readout to match LF5-D/F; + the machine-checked anchor `errorSyndrome_synClass3`. Remaining D1 strata: entangled/non-local tier, decoherence error model; MD-1 (the record layer — measurement on Σ). | | [`povm-plan.md`](povm-plan.md) | **POVM tranche** (DONE 2026-06-03): the ontic Born derivation extended to general (non-projective) POVMs via Naimark dilation. | **CLOSED** — P.1–P.5 done (type, dilation + Born transfer, volume reading, frequency capstone, canonical-dilation existence); all foundational-triple-only, Gleason-free. P.6 docs done. | | [`LF4-todo.md`](LF4-todo.md) | Ledger of items deferred from LF2/LF3 to LF4, numbered §1–§14, with bridge-discipline rules. | LIVE ledger. §14.2 done; general-N DH + LF3 re-route done; §2 done for pure states. ⏸ **§13 Wigner-rigidity programme PAUSED 2026-06-09 at the (2c-iii) phase-cocycle crux** — `transProb` API + `TransProbPreserving` + injectivity + frame reduction (`reducedMap` fixes basis rays) all built/audited SOUND in `Mathlib/.../Projectivization/{TransitionProbability,WignerRigidity}.lean`; residual = one Wigner normal-form lemma; decision deferred (full proof vs `wigner_fs_rigidity` axiom). See §13 banner. Other items still open. | | [`qm-empirical-tests.md`](qm-empirical-tests.md) | QM-validity regression-suite roadmap (Bell, no-cloning, teleportation, GHZ, Hardy, gates, crypto…). | LIVE. Many items done; E7/E91/BB84 + full no-broadcasting deferred (need fidelity/CPTP/LF5). | | [`qi-qec-roadmap.md`](qi-qec-roadmap.md) | Forward roadmap for the QI / QEC / algorithm branches: the four keystone gaps (entropy, channels, fidelity, LF5), the QEC tranche, algorithms + stabilisation as operational faces of D1, and the project-structure decision (topics as subfolders under QM+CSD; infra → Mathlib staging; LF5 as a new layer). | LIVE plan. **K2 channels DONE** (C1–C4); **K3 trace-distance metric COMPLETE** (`TraceDistance.lean`; defs + nonneg + distinguishability + symmetry + **triangle inequality** done 2026-06-08; only the data-processing inequality deferred); QEC 3-qubit codes + bit-flip channel DONE. **K3 COMPLETE 2026-06-09** — metric (triangle) + **data-processing inequality** (`channel_traceDist_le`, via the channel adjoint `Φ.adjoint` + variational characterisation) + closers (`traceDist_le_one` bounded `[0,1]`; `traceDist_conj_unitary` unitary invariance); all in `DataProcessing.lean`, Tier-A audited SOUND. Remaining keystones: K1 entropy (see [`k1-plan.md`](k1-plan.md)), K4/LF5. | | [`k1-plan.md`](k1-plan.md) | Staged plan for **K1 — von Neumann entropy `S(ρ) = −Tr ρ log ρ`** and the entropy/information stratum (subadditivity, SSA, data-processing, downstream Holevo/Schumacher/thermo). Mostly Cat-1 `QuantumInfo` infrastructure *beside* the CSD layers; the only thesis-relevant slice is the Landauer / ontic-entropy touchpoint (`S = log de-isolated volume`, `↔ μ_L`). K1-A reuses `TraceDistance.lean`'s `IsHermitian.cfc` + `re_trace_cfc` + eigenvalues. | **K1-A DONE 2026-06-16** (`Mathlib/QuantumInfo/Entropy.lean`): `vonNeumannEntropy` + operator-form `−Tr(ρ log ρ)` + `S ≥ 0` + pure-state `S = 0` + unitary invariance, foundational-triple-only, AxiomAudit-pinned. **Tensor additivity UNCONDITIONAL** (`vonNeumannEntropy_kronecker`, K1-A.2 DONE 2026-06-16): the Kronecker spectrum `spectral_sum_kronecker` (eigenvalues of `ρ⊗σ` = products `λρ·λσ`, in permutation-invariant spectral-sum form via charpoly + `diagonal_kronecker_diagonal`, no Mathlib Kronecker spectral theorem needed) discharges the former eigenvalue-product hypothesis; foundational-triple-only, AxiomAudit-pinned. The conditional `vonNeumannEntropy_kronecker_of_eigenvalues` is retained. `S ≤ log d` deferred (Jensen). **K1-B.1 partial trace DONE 2026-06-16** (`Mathlib/QuantumInfo/PartialTrace.lean`): `partialTrace{Right,Left}` (Mathlib has no matrix partial trace) + linearity (bundled LinearMaps) + trace preservation + tensor reduction `Tr_B (ρ⊗σ) = (Tr σ)•ρ` (trace of the traced-out factor; 2×2⊗2×2 sanity-probed) + IsHermitian/PSD preservation (via `v⊗eₖ` witnesses) + density ↦ density; foundational-triple-only, AxiomAudit-pinned, Mathlib-upstreamable. The shared prerequisite for K1-B.2 subadditivity + the gated decoherence / entangled D1 tier + the Landauer touchpoint. **K1-B.2 (Klein → subadditivity / Araki–Lieb) PARTIAL 2026-06-16** (`Mathlib/QuantumInfo/Subadditivity.lean`): **Klein's inequality LANDED** (`relEntropy_nonneg` / `klein_inequality`, `D(ρ‖σ) ≥ 0` for **σ positive-definite**) + relative entropy + the doubly-stochastic overlap matrix (`overlapV_row_sum` / `_col_sum`) + the **cross-term spectral expansion** `trace_mul_cfc_eq` (`Tr(ρ·cfc g σ) = ∑ᵢⱼ pᵢ g(qⱼ)‖Vᵢⱼ‖²`, trace of a product in DIFFERENT eigenbases) + reduced-trace identities `trace_mul_kronecker_one_right`/`_left`; all sorry-free, foundational-triple-only, AxiomAudit-pinned. Klein routes through the **scalar** `log_le_sub_one` + double stochasticity (NOT concave-Jensen, which is false at junk `log 0`). **The wall is now CLOSED — K1-B.2 subadditivity DONE 2026-06-16:** the **Kronecker-log operator split** `cfc_log_kronecker` (`log(ρ_A⊗ρ_B) = log ρ_A⊗I + I⊗log ρ_B`, PD factors) landed via the **decomposition-independent** `cfc_eq_conj_diagonal` (cfc of `U diag(d) Uᴴ` = `U diag(f∘d) Uᴴ` for ANY diagonalizing `U`, proved by **Lagrange interpolation** + charpoly-root eigenvalues — the cleaner route than the feared `StarAlgHomClass.map_cfc` continuity yak-shave); hence the headline **`vonNeumannEntropy_subadditive`** `S(ρ_AB) ≤ S(ρ_A)+S(ρ_B)` under `ρ_AB` PSD + **marginals PosDef** (NOT `ρ_AB` PD — pure entangled states covered), audited SOUND + non-vacuous at a correlated non-product state, foundational-triple-only, AxiomAudit-pinned. **`S ≤ log d` (`vonNeumannEntropy_le_log_card`, concave Jensen) + Araki–Lieb `|S_A−S_B| ≤ S_AB` (`vonNeumannEntropy_araki_lieb`, for ρ_AB PosDef) DONE 2026-06-17 — so K1-A and K1-B are COMPLETE.** Araki–Lieb built via the genuine purification route (`exists_purification` + Schmidt symmetry `pure_marginal_entropy_eq` (MMᴴ/MᴴM cospectrum) + `reshapeABR` + subadditivity + a swap-reindex for the other side); audited SOUND (Bell + asymmetric 2×3 + correlated full-rank ρ_AB probes; PD-scope honestly excludes the pure-entangled saturating case). Only the singular-marginal Araki–Lieb generalisation deferred. (Provenance: both the subadditivity wall-closure and this K1-A/B remainder were written in remote-control sessions and left uncommitted — the first with a build-breaking typo, the second build-clean; both recovered, audited SOUND, pinned, committed.) **K1-C strong subadditivity SCAFFOLD + CONDITIONAL REDUCTION LANDED 2026-06-17** (`Mathlib/QuantumInfo/StrongSubadditivity.lean`): tripartite marginals (`rhoAB`/`rhoBC`/`rhoB`/`rhoA` via reassoc + bipartite partial traces, reindexing identities proved) + the **mutual-information identity** `relEntropy_kronecker_eq_entropy_sub` (`D(ρ‖ρ_X⊗ρ_Y) = S_X+S_Y−S`, unconditional, from `cfc_log_kronecker` + reduced traces) + the **conditional reduction** `strong_subadditivity_of_relEntropy_monotone` (SSA from the data-processing inequality carried as explicit hypothesis `hDPI`; genuine reduction, correct direction `I(A:BC)≥I(A:B)`, non-vacuous on correlated ρ_ABC). **No axiom, no sorry, foundational-triple-only, AxiomAudit-pinned.** Scout CONFIRMED the deep input absent from Mathlib HEAD: operator MONOTONICITY present (`CFC.log_monotoneOn`, `monotone_rpow`) but operator CONVEXITY / Lieb / joint convexity / DPI ABSENT (explicit TODOs in `ExpLog/Order.lean`). `hDPI` (DPI = joint convexity = Lieb concavity) is the isolated wall; **FORK DECIDED 2026-06-17: route (a), BUILD the operator-convexity stratum** (genuine new math, not an axiom) — see [`operator-convexity-plan.md`](operator-convexity-plan.md). K1-D planned. | | [`lieb-dpi-scoping.md`](lieb-dpi-scoping.md) | Scoping for the last rung of the operator-convexity ladder (Lieb/Effros → `hDPI` → unconditional SSA). Walls checked first, Q10 mold. | **SCOPED 2026-09-01, recommendation = do NOT build now.** L.2/L.3a landed as transports and L.4 (`x log x` operator convex) was proved in a scoping probe, closing two named upstream Mathlib TODOs; the summit re-sized from the plan's "4–7 weeks" to **3–5 months**. ⛔ Decisive wall: physlib already proves unconditional SSA on **byte-identical pins**, which a 2026-08-17 rescan false-negatived. Payoff is smaller than it looks (CL-023 is qualified-BY-DESIGN, i.e. terminal), so the likely correct answer is to keep `hDPI` honest and do nothing. §3b weighs the bridge as long-term architecture. | | [`operator-convexity-plan.md`](operator-convexity-plan.md) | The new-math ladder to discharge K1-C's `hDPI` (data-processing / joint convexity of relative entropy) WITHOUT an axiom: L.1 operator convexity of `x⁻¹` → L.2 operator concavity of `log` → L.3 operator convexity of `x log x` → L.4 Effros perspective / joint convexity of `D` → L.5 DPI → unconditional SSA. Cat-1 (`Matrix` namespace, `Mathlib/Analysis/Matrix/OperatorConvex.lean`); multi-week. | **L.0 predicate + L.1 DONE 2026-06-17**: `OperatorConvexOn`/`OperatorConcaveOn` (genuine all-dimensions Löwner notion) + `operatorConvexOn_inv` (`x⁻¹` operator convex on `(0,∞)`, via the Schur-complement witness `fromBlocks_inv_posSemidef` + PSD-cone convexity; reuses Mathlib's `Matrix.Order` Löwner order + `PosDef.fromBlocks₁₁`). Audited SOUND (predicate all-dim/Löwner verified; L.1 holds for non-commuting PD pairs, correct direction — numeric forward-holds/reverse-fails probe), foundational-triple-only, AxiomAudit-pinned. **L.2-resolvent rungs DONE 2026-06-17**: shifted CFC↔inverse bridge `cfc_add_inv_posDef`, resolvent operator convexity `inv_shift_loewner_convex`, the **per-shift concave rung** `operatorConcaveOn_neg_add_inv` (`x ↦ −(x+s)⁻¹` operator concave on `(0,∞)`, the negation of the shifted resolvent, correct Löwner direction verified on the all-`n` predicate form), and the affine-output algebra `cfc_affine_output` + `OperatorConcaveOn.affine_output` (Step C: `c·f+d` preserves concavity for `c≥0`); all foundational-triple-only, AxiomAudit-pinned. **L.2 (log operator concavity) still WALLED** at a sharper boundary: `Matrix n n ℂ` is NOT a `CStarAlgebra` at the default instance (`infer_instance` fails), so Mathlib's C⋆-generic `CFC.log` / `tendsto_cfc_rpow_sub_one_log` / `monotone_rpow` / rpow integral-rep do not fire on the predicate's carrier — both routes shared a prerequisite `CStarMatrix ↔ Matrix` CFC/order bridge. **Bridge DONE 2026-06-18** (`Mathlib/Analysis/Matrix/OperatorConvexBridge.lean`): B.1 `cstar_cfc` (cfc naturality across the synonym `StarAlgEquiv`, the two distinct cfc instances agree via `StarAlgHomClass.map_cfc`) + B.2 `cstar_le_iff` (Löwner ↔ spectral order via `StarRingEquivClass.instOrderIsoClass.map_le_map_iff`) + `cstar_isStrictlyPositive` + **B.3 `matrix_log_le_log`** (operator MONOTONICITY of `log` on PD matrices in the Löwner order, transported from `CFC.log_le_log`); foundational-triple-only, AxiomAudit-pinned, non-vacuity witnessed. **rpow transport WALLED**: `NonnegSpectrumClass ℝ (CStarMatrix n n ℂ)` + the `Pow A ℝ` notation instance underlying `CFC.rpow_le_rpow` do not RESOLVE on `CStarMatrix` (provable as terms, not found by synthesis — the repeated-index discrimination key blocks `Fintype`/`DecidableEq` side-conditions); `log` is unaffected (needs only `[CStarAlgebra A]`). **REFRAMING LEMMA DONE 2026-06-18** (`OperatorConvex.lean`, bridge-INDEPENDENT on the `Matrix` carrier): `operatorConcaveOn_iff_concaveOn` (+ the two one-directional `operatorConcaveOn_of_concaveOn` / `concaveOn_of_operatorConcaveOn`) — `OperatorConcaveOn s f` ↔ ordinary `ConcaveOn ℝ (spectralSet s n) (fun A => cfc f A)` in the Löwner-ordered matrix codomain, for every dimension `n`; `convex_spectralSet_Ioi` (PD-cone convexity) supplies the `(0,∞)` domain-convexity side condition. Faithful (auditor-checked: genuine Löwner order, all-`n`, not a trace surrogate), foundational-triple-only, AxiomAudit-pinned. This makes Mathlib's whole `ConcaveOn` API (`ConcaveOn.add`/`.smul`/`.add_const`/Jensen) apply to operator concavity. **L.3a (`x^p` concave, `p∈[0,1]`) PARTIAL**: endpoints `operatorConcaveOn_rpow_{zero,one}` (`x^0=1`, `x^1=id`, on `Real.rpow`) LANDED via the reframing; interior `p∈(0,1)` WALLED — the matrix-carrier "cfc commutes with the integral" identity `cfc(·^p) A = ∫ cfc(integrand t) A ∂μ` needs `cfcₙ_setIntegral`, which requires `[NonUnitalCStarAlgebra (Matrix n n ℂ)]` (FAILS at default instances), so the `CStarMatrix` rpow instance-resolution wall OR a from-scratch matrix Bochner-cfc build gates it; the Schur/geometric-mean `p=1/2` route also walls (no operator geometric mean / sqrt concavity in Mathlib). No `x^p`/`log` interior concavity claimed. **A1 cfc-integral commutation + Löwner-order topology DONE 2026-06-18** (`OperatorConvex.lean` `Integral` section): `cfc_integral_commute` (`∫ cfc (g s) A ∂μ = cfc (∫ g s · ∂μ) A` on the bare `Matrix` carrier, via the spectral/entrywise route — both sides `U diag(·) Uᴴ` with `U` constant, reducing to finite scalar `integral_ofReal` identities; the from-scratch matrix Bochner-cfc build that sidesteps the failed `NonUnitalCStarAlgebra (Matrix n n ℂ)`), plus the order-topology instances `integral_concaveOn_of_integrand_ae` consumes (`isClosed_posSemidef`, `ClosedIciTopology`, `OrderClosedTopology`, `PosSMulMono`, `SMulPosMono`, `IsOrderedModule`, all scoped to the Frobenius norm). This breaches both halves of the L.3a-interior route-(ii) wall; the remaining step is the final `x^p` interior assembly. L.2/L.4/L.5 not started. | | [`csd-departures-eft.md`](csd-departures-eft.md) | The "beyond reproducing QM" thread: candidate CSD-vs-QM departures (G3b → third-order/Sorkin interference, V≈1−I → complementarity, determinism → finite-sample statistics, finite-N), their physical-test handles, the CSD-specific negative tests, and the finite-EFT correction tower. | LIVE map. Theory-gated (corrections not pinned); leading order = QM is done, subleading owed by the paper sequence. | | [`paper-candidates.md`](paper-candidates.md) | **Which landed clusters have no manuscript home.** 39 of 52 ledger claims are unattributed — LF5/LF6 are uncitable as documents by `CITATION.cff` policy, and the 21 Q17/Q19 census claims postdate every published paper. Five candidates ranked by value-per-effort: impossibility, methodology/necessity-audit, thermodynamics-typicality, rigidity, Σ-fibre; the CV/EFT chain held back behind the P3 ceiling. Carries the TH1 scope warning (first moment ≠ concentration) and names the two open errata as prerequisites. | **Authorial planning, 2026-08-19. Not a work queue** — no Lean is owed by anything on the page. | | [`eft-pillars-plan.md`](eft-pillars-plan.md) | **★ The fork that decides what "move toward EFT" means**, and the five pillars between the corpus and a field theory. **Route A** (effective field theory with spacetime *given*) versus **Route B** (the field structure of the substrate itself). Pillars: **P1** a field-structured flow + the arena bridge (the twice-observed bottleneck, highest-value structural item); **P2** composite arenas (algebra half proved, transport owed); **P3** the factorisation problem — **a CEILING, not a gap**: a chosen tensor factorisation is an epistemic construct, so Route A's spatial index is a permanent label; **P4** relativistic structure earned rather than defined (dispersion is a *definition* today); **P5** interactions past upper bounds (RG half DONE by CV-26; attainment open). | **LIVE, brought current 2026-08-19.** Read this *before* scoping any "next EFT step" — it is where the Route A/B fork and the P3 ceiling are recorded. The CV ladder it measures against is now COMPLETE (CV-1…CV-26), so its own verdict stands: **P1, not more CV**, is the next structural step; P4 is the next bounded one. | | [`nqubit-register-plan.md`](nqubit-register-plan.md) | Plan for the **n-qubit register** + quantum-algorithm branch (Deutsch–Jozsa → QFT → Grover → Shor). `QReg n = EuclideanSpace ℂ (Fin n → Fin 2)`; Cat-1 in `Mathlib/QuantumInfo/Register.lean`, algorithms in `Empirical/QM/Algorithms/`. | **COMPLETE 2026-06-08.** R1–R5 + Grover done 2026-06-06; full Shor done 2026-06-07/08 → see [`shor-plan.md`](shor-plan.md). All AxiomAudit-pinned, foundational-triple-only. Coverage breadth, not the thesis. | | [`shor-plan.md`](shor-plan.md) | Plan for **Shor's algorithm**, staged S1–S7 + assembly. In scope (finite QM + finite number theory, never field theory). | **COMPLETE 2026-06-08 — entire chain machine-checked end to end.** Quantum core (M1/M1.5/S4, `ShorCore.lean`), recovery (S5, `ShorRecovery.lean`), factoring + bridge (S6, `ShorRecovery.lean`), random-`a` success ≥ 1/2 for arbitrary odd composite `N` (S7 + S7-gen, `ShorRandomA.lean`), factoring capstone (`ShorCapstone.lean`). All foundational-triple-only, AxiomAudit-pinned, Tier-A-audited SOUND. Honest-scope deferrals (not load-bearing): constructive CF computation of `r`; two-register `r∤T` joint marginal. | | [`amplitude-amplification-plan.md`](amplitude-amplification-plan.md) | Plan for **amplitude amplification** (Brassard–Høyer–Mosca–Tapp): the generalisation Grover is an instance of — arbitrary preparation, arbitrary good set, closed-form `sin²((2j+1)θ)` amplification, `⌊π/(4θ)⌋` optimal count, Grover re-derived as the corollary and the parallel development retired (the priced atlas-extraction pilot). Cat-1, `Mathlib/QuantumInfo/AmplitudeAmplification.lean`. Amplitude *estimation* recorded as the gated stretch (AA-5, consumes `PhaseEstimation.lean`). | **EXECUTED 2026-08-29, same day (AA-1..AA-4 + the AA-5a eigenstructure/error-algebra slice).** The general theorem + optimal-count corollaries landed Cat-1; Grover rebuilt as the instance with the parallel development retired and the k-marked corollary added; pilot priced at ≈25 min for the refactor. AA-5b fully executed same day: `JointRegister.lean` + ShorCore delegation, then the assembly `AmplitudeEstimation.lean` (BHMT Thm 12 — exact mixture marginal, per-index bound, error reading). AA-6 executed in engine form (qsearch_average, BHMT Lemma 2) + mirror refinement (amplitude_estimation_pair, 4/pi^2 on the estimate) same day on author instruction. Plan closed; named residues: the straddling-kernel inequality (literal 8/pi^2) and the Thm 3 doubling schedule. Candidate 2 of the 2026-08-28 five. **R-001 and R-002 both discharged 2026-09-23** (`amplitude_estimation_bhmt`, `qsearch_expected_cost`). | | [`gottesman-knill-plan.md`](gottesman-knill-plan.md) | Plan for the **Gottesman–Knill / stabiliser mechanics** layer: the Pauli operators `X^a Z^b` with their `𝔽₂`-symplectic algebra (group law, commutation ⟺ symplectic form, character orthogonality, tracelessness), and the Clifford generators (CNOT, S, H) conjugating Paulis to Paulis with explicit `𝔽₂`-linear label maps — the Heisenberg-picture closure that IS the simulability mechanism. Cat-1, `Mathlib/QuantumInfo/Pauli.lean` + `Clifford.lean`. Honest scope: no computation model, no complexity claim; the stabiliser-group/measurement layer is the gated GK-3. | **GK-1 + GK-2 EXECUTED 2026-08-29, same session as scoping** (≈75 min; the pre-derived H-conjugation compiled on the first structural attempt); **GK-3 EXECUTED later the same day** (`Stabilizer.lean`: coherence-law design, absorption/idempotence/trace, stabState_exists with no spectral machinery; 3 more pins). Residues DISCHARGED same day ("Complete GK"): rank-equals-trace uniqueness (stabProjector_rank, stabState_unique) + the full measurement-update rule (deterministic case, expectation zero, probability 1/2, commutant update). Plan CLOSED. Candidate 3 of the 2026-08-28 five (the Gross connection). | | [`magic-plan.md`](magic-plan.md) | Plan for the **magic layer** (candidate 5, the last of the 2026-08-28 five): the T gate's provable escape from the Clifford closure — T² = S (the hierarchy descends), T X T† = (X + i·XZ)/√2 (the level-3 identity), the no-go (no c·X^aZ^b equals T X T†, forcing 1 = ±i), and the magic state |T⟩ = T·H|0⟩ with closed coordinates and unit norm. Cat-1, `Mathlib/QuantumInfo/Magic.lean`. | **EXECUTED 2026-08-29, same session as the GK completion** (≈110 min for the pair). Named residues with reasons: distillation (Bravyi–Kitaev 15-to-1, program-verification scale), universality (gate-synthesis density), the T-injection circuit. All five candidates of the 2026-08-28 thread now executed. | | [`steane-plan.md`](steane-plan.md) | Plan for the **Steane [[7,1,3]] code**: the CSS construction from the Hamming [7,4] parity rows, as the first genuine instance of the GK-3 stabiliser layer — code dimension by the general trace, logical states stabilised by all 64 group elements and orthonormal, logical X̄/Z̄ acting as a genuine encoded qubit, single-error syndromes nonzero and injective (distance 3). 3-Local, `Empirical/QM/QEC/Steane.lean`. | **EXECUTED 2026-08-29, same session as GK-3** (≈80 min for the pair). Honest scope: no recovery map, no Knill–Laflamme, no fault tolerance — three-qubit-module posture. 4 audit pins. Candidate 4 of the 2026-08-28 five. | | [`lf6-plan.md`](lf6-plan.md) | **LF6 — the entangled de-isolation tier (D1 frontier).** Extends LF5 measurement-dynamics to the entangled / non-local stratum: a deterministic de-isolation account reproducing the LF3 singlet kernel `P_st`, non-locality located honestly (in the Σ-volume engine, derived; contextuality Bell-forced), no-signalling proven. | **A.1 DONE 2026-06-28** (`LF6/ForcedContextuality.lean`, auditor-SOUND, foundational-triple): the forced-contextuality crux — `no_product_partition_realises_singlet` (no setting-local ±1 product partition of Σ reproduces the singlet, via `lhvCHSH_abs_le_two` + the singlet `2√2`; both hyps proved load-bearing), `productPartition_nonvacuous`, `engine_joint_nonfactorises` (`P_st ≠ 1/4` at aligned axes), `engine_marginal_factorises` (no-signalling, LF3 reuse). Frame: factorisation = measurement; nudge ≠ carve; non-locality and Born are two outputs of one Σ-volume engine; the sector is posited (Σ is the floor). **A.1–A.3 + B.1/B.2 DONE 2026-06-28** (`LF6/{ForcedContextuality,SingletDeisolationFlow,LocalDeisolationFlow,Decoherence}.lean`): A.2/A.3 = the `ℂℙ¹⁵` de-isolation flow `Φ ≠ id` realising the singlet dilation (`singletDeisolationFlow`, flow-realises-dilation), B.1/B.2/B.3 = the decoherence / partial-trace tier: purity-drop witness + the von Neumann entropy-increase witness (B.3, `decohere_vonNeumann_entropy_pos_of_superposition`, the pure→mixed entropy jump 0→S>0, DONE 2026-07-01). **C.1 + C.2 (general-N tier opened) DONE 2026-06-30** (`LF6/{GHZContextuality,GHZDeisolationFlow}.lean`): C.1 `no_product_partition_realises_ghz` — the GHZ forced-contextuality crux, a DETERMINISTIC all-or-nothing forcing (Mermin paradox, routes through the existing `no_lhv_assignment_for_ghz`) vs A.1's statistical CHSH bound (±1 via `pm_ae_eq` + locality both load-bearing). C.2 `ghzDeisolationFlow` — the GHZ de-isolation FLOW `Φ≠id` @ N=8, pointer-block FS volume = GHZ Born weight (composed), a.s. frequency; MINIMAL computational-basis carve (diagonal weights, NOT contextual). C.3 `GHZMerminCarve.lean` — the GENUINE dynamical contextual carve: new GHZ Pauli-context joint eigenstructure, `ghzDeisolation_blockVolume_correlation` (the sign-weighted block-volume sum = the Mermin expectation `re⟨ghz|σσσ|ghz⟩`, all four contexts) + `ghzDeisolation_carve_not_product` (four-context tie to C.1 CLOSED — the carve's OWN correlations, not a re-export; the A.2 analogue). The first DYNAMICAL realisation of GHZ non-locality. C.4 `GHZLocalFlow.lean` — the 3-party manifestly-LOCAL product flow `V₀⊗V₁⊗V₂` (three N=2 wings) realising the GHZ measurement (`ghzLocal_pullback` reuses A.3's 2-wing pullback for the inner wings), so the de-isolation needs no non-local interaction; non-locality is in the carve + entangled prep. **LF6-D (CGLMP statistical axis) CLOSED at FULL generality 2026-07-04** (`LF6/CGLMPQudit.lean`): the maximally-entangled qudit `Ψ_d`'s CGLMP violation `cglmp_maxEntangled_qudit_gt_two (hd:2≤d): cglmp d pQM > 2` for EVERY `d ≥ 2`, from `Ψ_d`'s genuine Born table (`pQM_closed`, via the `d`-th-roots-of-unity Dirichlet kernel) + a real analytic bound (`k=0` term `≥ 32/π²−8/9 > 2` uniformly in `d`, Jordan + `sin x ≤ x`); Bell force `no_lhv_realises_maxEntangled_cglmp_d`. Extends the `d=3` qutrit (LF6-D step 2, `CGLMPQutrit.lean`) to all `d`; foundational-triple, AxiomAudit-pinned. Together with the GHZ_n deterministic axis (`GHZnDeisolationFlow.lean`, LF6-E) forced non-locality is now reproduced in BOTH forms at general dimension AND general party-number. **NEXT: the wider general-`N` tier (bipartite d×d, GHZ_n); Lindblad; D1c-entangled.** | | [`thermo-plan.md`](thermo-plan.md) | **Thermodynamics track** (CSD statistical mechanics): Born weights are equilibrium volume fractions, typicality is the ensemble, the FS-measure-preserving de-isolation flow is the reversible microdynamics, entropy increase is coarse-graining. TH1 canonical typicality → TH2 second law (coarse-grained entropy monotone) → TH3 temperature/free energy → TH4 Landauer. | **TH1 DONE 2026-07-05** (`Thermo/CanonicalTypicality.lean`, auditor-SOUND, foundational-triple): `fs_first_moment` (`∫|ψ⟩⟨ψ|dμ_FS = I/N`, a genuine twirl/Schur integral -- offdiag=0 via a signFlip change-of-variables, diag=1/N via permutation unitaries + `momentMap_sum_eq_one`) + `canonical_typicality_expectation` (`E_{μFS}[Tr_E|ψ⟩⟨ψ|] = I_S/d_S`, thermal equilibrium from FS volume). HONEST SCOPE: EXPECTATION (average) only; the typical-state concentration (Levy / spherical isoperimetry, not in Mathlib) is the named residual; NOT dynamical thermalisation. **TH2 DONE 2026-07-07** (`Thermo/SecondLaw.lean`, foundational-triple): the second law as the H-theorem `vonNeumannEntropy_le_pinching` (`S(ρ) ≤ S(pinch ρ)` -- pinching/dephasing to the pointer-basis diagonal never decreases vN entropy, via Klein's inequality + the diagonal cross-term identity) + `entropy_reversible_then_coarsegrain` (fine-grained unitary conserves entropy, coarse-graining produces it) + `entropy_production_nonneg` (pure-state instance = LF6-B.3). Honest scope: the Klein support hypothesis; a SPECIFIC coarse-graining, not a universal second law. **TH3 DONE 2026-07-07** (`Thermo/FreeEnergy.lean`, foundational-triple): temperature / free energy / the Gibbs variational principle -- `gibbs_free_energy_min` (the Gibbs state `ρ_β = exp(-βH)/Z` minimises `F(ρ) = Re Tr(ρH) − T·S(ρ)`, `F(ρ_β) ≤ F(ρ)`, via Klein `β(F(ρ)−F(ρ_β)) = D(ρ‖ρ_β) ≥ 0`) + `gibbs_free_energy_eq` (`F(ρ_β) = −T log Z`) + the crux `cfc_log_gibbsState` (`log ρ_β = −βH − (log Z)·1` via `cfc_eq_conj_diagonal` on H's eigenbasis) + `gibbsState_posDef`/`_trace` (a genuine density). Honest scope: the inequality (not equality-iff-Gibbs); `[Nonempty n]`. **TH4 DONE 2026-07-07** (`Thermo/Landauer.lean`, foundational-triple): Landauer's principle in the Reeb–Wolf form -- `landauer_bound` (a system coupled by a global unitary to a bath in the Gibbs state obeys `S(ρ_S) − S(ρ_S') ≤ β·ΔQ`, entropy removed ≤ β × heat dumped) via entropy conservation (`conj_unitary`+`kronecker`) + `subadditive` + `bath_clausius` (`D(ρ_B'‖τ_B) ≥ 0` through the TH3 Gibbs log identity); `landauer_one_bit` (erasing a maximally-mixed bit costs `ΔQ ≥ T log 2 = kT ln 2`). Honest scope: full-rank marginals + initial state; the standard finite-dim setting. **THERMO TRACK TH1–TH4 COMPLETE.** TH5 (ETH / fluctuation theorem) the optional stretch. Corpus support: `vonNeumannEntropy` (K1), `vonNeumannEntropy_subadditive`, the LF6-B entropy witness, `klein_inequality`/`relEntropy_nonneg`/`cfc_eq_conj_diagonal`, K2/K3, `μ_FS`+LLN typicality. Honest posture: QM stat-mech with a CSD reading; the deterministic-microdynamics claim rests on the posited Σ / D1 residue. | | [`future-work.md`](future-work.md) | **Open-work priority table** (reference-numbered): every pending item with a stable id (SL-/LF6-/TH-/CV-/W-/EC-/HY-), its dependencies, and a complexity grade (S/M/L/XL), plus a ★ flag for the thesis-critical (record-layer / reconstruction-completing) rows and a priority read. Cite items by id ("do EC-1"). | **LIVING DOC, created 2026-07-04.** Tracks the frontier — the extensions (CV, relativistic locality, identical particles); the record layer / MD-1 is now built; SL-2 D1c-entangled), LF6 remaining, thermodynamics (TH), CV, the W-series dynamics spine (W-2/W-3/W-5 DONE 2026-07-05, the Σ-flow→Schrödinger chain; S2 finite-dim Stone DONE 2026-07-05, S1 phase lift DONE 2026-07-07 with the W5 capstone `projectedFlow_schrodinger_form` = projected flow is `exp(-itH)`-conjugation on rays; W-3 clopen datum CLOSED 2026-07-07 via the Bargmann discriminator — ALL W-RESIDUES CLOSED), hygiene/audits (HY), and the pillar-completeness deferrals (P3 tensor derivation, IP-1 identical particles). **Priority order (user-set 2026-07-07): TH track (TH-2→TH-4 + TH-1 concentration residual) → CV track (CV-1 onwards) → EC (EC-6 et al. deprioritised).** | | [`trace-distance-triangle-plan.md`](trace-distance-triangle-plan.md) | Plan for the K3 trace-distance **triangle inequality** (= trace-norm subadditivity on Hermitian matrices), via the positive/negative-part decomposition + the TR-PSD linchpin (`0 ≤ Re Tr(S·P)` for PSD `S,P`), working with the `PosSemidef` predicate (Mathlib has no Loewner `≤`) and `IsHermitian.cfc` (defined for discontinuous `f`, dodging the `sgn` continuity gap). | **DONE 2026-06-08.** `traceNorm_add_le` + `traceDist_triangle` landed (metric complete), foundational-triple-only, AxiomAudit-pinned, Tier-A-audited SOUND. Jordan primitives `posPart`/`negPart`/`posProj` exposed for reuse. Remaining K3: the *data-processing* inequality (separate, larger tranche; its load-bearing half is already built here). | | [`channels-plan.md`](channels-plan.md) | The K2 CPTP-channel infrastructure plan (Kraus core → Stinespring dilation → canonical channels → no-comm CPTP / QEC error channel). Built on the existing `canonicalNaimark` + `PartialTrace` primitives; the on-ramp to `Φ≠id`. | **C1–C4 DONE 2026-06-05** (`QuantumInfo.Channel`; TP/PSD/Hermitian core; Kraus↔Stinespring bridge; `unitaryChannel`/`traceOutChannel`/`mixedUnitaryChannel`; `tensorRight` + `channel_no_communication` **retiring the E3 CPTP gap**; `bitFlipChannel` error channel; all AxiomAudit-pinned). The **channel adjoint** `Φ.adjoint` (unital, positive, duality `Tr(P·Φρ)=Tr(Φ†P·ρ)`) landed 2026-06-09 with the K3 data-processing inequality; full CP / Choi-positivity (C5) remains, off critical path. | | [`carve-out-plan.md`](carve-out-plan.md) | Carve-out diagnosis (Born = volume?) + the moment-map programme + the posited-sector (Σ, the floor) / D1 frontier. | LIVE for the frontier (D1). Qubit Born-from-volume CLOSED; general-N superseded it (see in-file update). | | [`audit-sweep-plan.md`](audit-sweep-plan.md) | Extend the adversarial audit **below the Tier-A headlines**: per-module csd-lean-auditor gap sweeps over the Tier-B-and-below strata (supporting lemmas, Cat-1 staging tree, bridge bundles), with a running per-sweep ledger. **Also holds the external-review intake ledger** (codex 2026-06-11, codex 2026-08-06 — each finding classified + dispositioned with same-day source verification). | **COMPLETE 2026-06-15 — SOUND.** All 5 sweeps done (12 sub-sweeps, 167 files): zero BLOCKER/zero MAJOR; findings all MINOR/NIT docstring-honesty (fixed) + upstream-prep hygiene (logged). Carving-honesty line verified correct throughout. See the ledger + findings in the plan file. **2026-08-06 intake triaged same day** (12 findings, no S4; see the plan file). | | [`external-library-map.md`](external-library-map.md) | **★ The one-repo architecture + external-ecosystem ownership map (2026-08-06).** `csd-lean4` = the unified CSD quantum repository; Mathlib/Physlib/Lean-QIT sit underneath as (future) dependencies through `CsdLean4/Interop/`. Ownership table with statuses, the four-way check-before-implementing rule, single-provider rule, the named Lean-QIT fork (hDPI/SSA vs the E2 ladder), toolchain facts, and the consolidation-release target topology. | **Read BEFORE adding any generic quantum definition** (the four-way rule) and before the §H chaos tranche. | | [`FOUNDATIONAL-CODE-REVIEW.md`](FOUNDATIONAL-CODE-REVIEW.md) | **External review (codex, 2026-08-06): full 443-module source/statement/API/dependency review.** 38-file foundational ledger + 405-file appendix; 12 cross-cutting findings F-01…F-12, all S3-and-below (claim-alignment / hygiene — **no unsoundness, no trust escapes found**). Reviewer could not run `lake build` (sandbox network); compilation status comes from local/CI builds, not this doc. | Intake triaged 2026-08-06 in [`audit-sweep-plan.md`](audit-sweep-plan.md) (classification + disposition per finding, concrete claims re-verified both ways). | | [`VALIDATION-LEDGER.md`](VALIDATION-LEDGER.md) | **Headline-claim validation ledger** (created codex 2026-08-06; **census-extended 2026-08-13, Q17**): **52 headline claims** (canonical machine-readable source [`validation-claims.tsv`](validation-claims.tsv)) with status validated/qualified/needs-change/specialist-review, load-bearing assumptions, and an independent-check route each; **admission criteria now stated in-file** (terminal + claim-bearing + pinned + distinct), so the set is a rule, not a session artifact. Structurally enforced by `scripts/check-validation-ledger.sh` (CI, Linux) + the Headlines drift guard. | LIVE control. 9 validated / 40 qualified / 2 needs-change as of 2026-08-13; statuses move only with claim-by-claim sign-off. | | [`validation-hardening-plan.md`](validation-hardening-plan.md) | **Validation-hardening campaign (2026-08-12):** warnings-as-errors CI, the concrete-witness suite (`Tests/Witnesses/` — explicit nontrivial inhabitants of the LF1–LF6/composite/dynamics assumption packages, driven through the headline chains), standard-linter + import-hygiene guards, forward-compat + docs workflows. Validation machinery on the sufficiency/inhabitation leg — **not** reconstruction progress (witness ≠ result). | **COMPLETE 2026-08-12** — all 13 workstreams DONE same day; 8 witness modules, 22 pins (foundational-triple only), 1 new blocking guard + 3 permanent mutation probes, 2 new workflows. Final verification record in-file. | | [`necessity-audit.md`](necessity-audit.md) | **★ The necessity audit (2026-08-09): how tightly is the substrate pinned down?** Every declared headline (CL-001–030) classified by logical strength (necessity / conditional necessity / sufficiency / impossibility / instantiation): one unconditional necessity (`wigner_rigidity`), the forcing concentrated one level above the substrate, the ontic fields doing almost no work, and the two systematic conditioners named — `U(N)` in every measure-forcing result, `hTPP` in every Wigner-selection result. The practical output: which additions would narrow the space (conditioner discharge in the `fubiniStudy_forced_by_symmetry` template) vs merely lengthen the count. | **COMPLETE 2026-08-09.** The commissioning doc for Q11/Q17. Read before classifying any new headline's strength. | | [`unitary-tpp-scoping.md`](unitary-tpp-scoping.md) | **Q11 scoping (2026-08-13): discharging the `U(N)` / `hTPP` conditioners.** Maps the three operational-reconstruction shapes against Σ-primitives — Hardy survives as the minimal-scope programme, CDP's grammar declined (harvest = the already-landed tensor fragment), MM's requirement style survives with its Lie-theoretic lever deferred to the (deferred) maximal scope — and reduces **both conditioners to one record-level premise**: `transProb` is a record observable, so `hTPP` and the `U(N)`-invariance posit convert to statistics-preservation premises with the group a Wigner output. First brick named: `RecordLayer/StatisticsRigidity.lean` (= BACKLOG **Q18**). | **SCOPING COMPLETE 2026-08-13** (Q11 closed). Premise conversion, not elimination; the §13.2 trap untouched; D1 stays open. | | [`channel-rg-scoping.md`](channel-rg-scoping.md) | **CV-25 scoping (2026-08-18): channel-level RG at the cutoff.** The Stage-4 no-go (`exists_unitary_compress_not_unitary`) said RG must become a channel statement with an error budget; this pass finds the budget **provable from existing bricks**: mode tracing (`traceOutChannel`∘`modeSplit`) is CPTP today, the norm is fixed (trace distance on states, price via L2 opNorm), and the chain DPI → bridge → telescoping → Duhamel yields `ε(λ,τ,n) = 2n·\|τ\|·\|λ\|·C_v` with exactly two missing wall-checked links. First brick named: **CR-1**, the unitary-perturbation bridge `traceDist(UρU†,VρV†) ≤ 2‖U−V‖`. Level decimation recorded-not-selected (trace-decreasing; needs a leakage arm). | **SCOPING COMPLETE 2026-08-18; the arc it named (CV-26) EXECUTED the same day** — CR-1/CR-2/CR-3 landed in `Mathlib/QuantumInfo/UnitaryPerturbation.lean` + `CV/ChannelRG.lean` (7 pins), exactly as scoped, both missing links discharged. A **spent** scoping pass: kept as the feasibility record, not an open plan. | | [`q10-no-signalling-scoping.md`](q10-no-signalling-scoping.md) | **Q10 scoping (2026-08-25): no-signalling from primitives over a non-factorising Σ (C1 item 16).** ★ Walls checked BEFORE any theorem, which is the direct lesson of the Q12 arc. **W1**: the response-side primitive (pointwise `F_A(a,b,x) = F_A(a,b',x)`) is not merely too strong but **FALSE** here — it makes `(R_A,R_B)` a product partition, and `no_product_partition_realises_singlet` is fully general (Bell's Λ, no CSD hypothesis), so the route is closed by a theorem for any theory of this shape. **W2**: `OperationalNoSignalling` is a restatement of the conclusion, which is why the row exists. **W3**: its one-`μ`-across-contexts fixture **is** measurement independence (already disclosed). ★ **W4**: the corpus **already derives** no-signalling from a primitive — `CV.composite_no_signalling`, from **disjoint mode support**, needing no product form on the state. ⇒ the search space is **"what is the Σ-level analogue of disjoint support"**, not "which response condition is weak enough". ★ Records that **Q10 and Q12 exit at the same place**: independence as a primitive on Σ. | **OPEN, scoped.** Next brick **Q10-w**: settle the fibre-factorisation gate on paper — it may retract the doc's own §4 conjecture, and is meant to. ⚠️ Framing not yet `csd-foundations`-checked. | ## Done / historical (context, mostly closed) | Doc | What it is | Status | |---|---|---| | [`general-n-dh-plan.md`](general-n-dh-plan.md) | General-N Duistermaat–Heckman / Born-from-Kähler-volume plan (Slices A–E). | **CLOSED 2026-06-02.** Headline + Born lift + apex + empirical capstone + N=2 reduction all done. | | [`plan-b-detail.md`](plan-b-detail.md) | Plan B: discharge the qubit DH axiom `fs_moment_pushforward_uniform` (Gaussian + FS-uniqueness route). | CLOSED 2026-05-31 (axiom is now a theorem). | | [`partial-trace-plan.md`](partial-trace-plan.md) | Partial-trace Cat-1 tranche (`PartialTrace.lean`, reduced density). | DONE. Tooling now feeds the POVM tranche. | | [`pre-LF4-plan.md`](pre-LF4-plan.md) | Pre-LF4 plan: option-(B) chain design, the PureSingletPreparation bundle. | DONE (note: the chain was re-routed off Busch 2026-06-02; this doc describes the older Busch-mediated form). | | [`LF1-plan.md`](LF1-plan.md), [`LF2-plan.md`](LF2-plan.md), [`LF3-plan.md`](LF3-plan.md) | Per-layer construction plans. | DONE (LF1/LF2/LF3 all merged + verified). | | [`projectivization-plan.md`](projectivization-plan.md) | Mathlib-staging plan for the `Projectivization` topology/measure API. | DONE (the `CsdLean4/Mathlib/` staging tree). | | [`empirical-csd-bridge-plan.md`](empirical-csd-bridge-plan.md) | Discipline for `Empirical/CSD/` bridge files (load-bearing-hypothesis rules). | LIVE discipline doc; see `../BRIDGE-OBLIGATIONS.md` for the ledger. | ## Reference (not todos) | Doc | What it is | |---|---| | [`expert-system-prompt.md`](expert-system-prompt.md) | The project's expert-posture / system-prompt reference. | | [`spec-to-lean.md`](spec-to-lean.md) | Spec-to-Lean mapping notes. | | [`review-surface-findings.md`](review-surface-findings.md) | First-run findings of `scripts/check-review-surface.sh` (2026-08-05): the expert-review triage — thin defs, no-API defs, single-use statements, proof-style and name-shape outliers — as questions for a human review pass, with honest coverage of Ilin & Nugent's five categories. Baseline: `docs/review-surface-baseline-2026-08-05.txt`. | | [`../README.md`](../README.md) | **Introductory** front page (rewritten 2026-08-02): plain-language claim, honesty frame, doc map. Detail lives in `docs/`. | | [`../docs/TOUR.md`](../docs/TOUR.md) | The detailed tour: precise claims with theorem names, the results-by-pillar table, the measurement story, reader-type pathways. | | [`../docs/PATHS.md`](../docs/PATHS.md) | Reading paths by quantum sector (foundations, dynamics, measurement, entanglement, QI, crypto, algorithms, thermodynamics). | | [`../MATHLIB-GAPS.md`](../MATHLIB-GAPS.md) | Mathlib gaps this project hit (what each blocks) + the staged upstream candidates and the suggested first batch. | | [`../EMPIRICAL.md`](../EMPIRICAL.md) | Per-test index of the empirical suite (both branches: file, headline theorem, claim, source). | | [`../CLAUDE.md`](../CLAUDE.md) | Working-with-the-code guide (architecture, conventions, module chains). | | [`POSITS.md`](POSITS.md) | **The posit register** — what the corpus assumes rather than derives, each with its characterisation theorem and what would discharge it (1 cell law, 2 sector, 3 Liouville preservation, 4 typicality reading, 5 calibrated bank, 6 unitary class, 7 composite structure, 8 measurement independence, 9 product form of the preparation measure). Carries the correspondence table to the external review's Part II numbering, and defines what "frontier" means: permanent boundary vs open mathematics vs open foundations. | | [`cr-queue.md`](cr-queue.md) | **External-review queue (CR-1…CR-16)** with the author's execution order and per-item status. Recorded 2026-09-05 because it previously existed only in a chat transcript. ⚠️ Its posit numbering is not the repository's — see `POSITS.md`. | | [`../AXIOMS.md`](../AXIOMS.md) | Axiom posture: **zero imported axioms** (`busch_effect_gleason` proved and de-axiomatised 2026-07-21, §2.2; `invariant_measure_uniqueness` removed 2026-06-04, §2.1); every theorem reports only the foundational triple. Structural assumptions/posits in §3 (posited Σ, typicality reading, swap calibration, the cell law §3.10), registered with discharge conditions in [`POSITS.md`](POSITS.md); per-theorem audit §5. | | [`../CONVENTIONS.md`](../CONVENTIONS.md) | Three-category (Cat-1/2/3) discipline, naming, namespace rules. | | [`../BRIDGE-OBLIGATIONS.md`](../BRIDGE-OBLIGATIONS.md) | Ledger of load-bearing Empirical/CSD bridge fields with LF4-todo cross-refs. | | [`../PLACEHOLDERS.md`](../PLACEHOLDERS.md) | Placeholder / stub tracking. | ## Current state of the programme (one paragraph) *(Rewritten 2026-08-02; the previous version of this paragraph predated the record and dynamical layers and still named a standing axiom.)* The corpus machine-verifies finite-dimensional QM as theorems about one ontic model `Σ = ℂℙ^{N-1} × T²`: **both pillars** (Schrödinger evolution as the projected sector flow; Born as Fubini–Study typicality volume, Gleason-free, POVMs included) plus — as of the 2026-08 arc — **measurement as a process**: records dynamically created from a ready state, persistent, mutually exclusive, Born-weighted (`swap_sector_born`), with the rank-one Lüders update a pushforward theorem (`swap_luders_born`) and the **degenerate** Lüders update realised on the projective join `ℙ(ℂ^{N+N})` as a Liouville-preserving unitary, pointwise (`join_block_luders`). The **unified arena** (`unifiedArenaClosure`) carries isolated dynamics, measurement dynamics, Born frequencies, and mixed states on ONE space with ONE Liouville measure family, including the isolate→measure→isolate round trip. The **A1–A7 map has no genuinely open row** — A2's Hamiltonian-generation sub-question (reopened 2026-08-02, torus-flux) was closed 2026-08-03 by the `ℂℙ^K` smooth pointer witness; the landed witness classification is **piecewise rigid symplectic translation with null seam set** (`ShearDiscontinuity`/`PiecewiseHamiltonian`). **Operational finite-QM closure was declared 2026-08-06 (`v1.1.0-finite-qm-closed`)**: the README fixes the operational definition and the requirement→theorem role map; `reconstruction-status.md` carries the declaration against its net line. The one remaining track inside the repository is the foundations frontier (`CSD-CHARTER.md`). (B6 upstreaming and §C paper alignment were both moved off this repository's queue 2026-08-06, author decision — neither is a repo need; `CONVENTIONS.md` §7 and the BACKLOG §C banner carry the records. The staging discipline stays, and the corpus remains the source of truth for the paper–corpus divergences.) The **empirical suite** covers every flagship test on both branches (QM proof + CSD ontic twin; `../EMPIRICAL.md`), with the crypto tranche consuming the dynamical layer. The corpus imports **zero axioms** — every theorem reports only `[propext, Classical.choice, Quot.sound]` — is sorry-free, and is regression-protected by `CsdLean4/Tests/AxiomAudit.lean` + `check-claims.sh`; build with `lake build` **and** `lake build CsdLeanTests` (the second target runs the audit). Open work: the single canonical list is [`BACKLOG.md`](BACKLOG.md) (headlines: the join-arena measurement-protocol plumbing (M), any-basis first measurement (M–L), composable finite-key QKD (XL), Shor-9 (L)). ## C1 correction (2026-08-10) - [`publication-errata.md`](publication-errata.md) — claims in non-editable manuscripts that the formal work has since falsified (E-1 type separation, E-2 the nudge). - [`c1-correction-plan.md`](c1-correction-plan.md) — the C1 correction: phasing, design corrections, and the finding that work-order item 7 was FALSE. ## Busch–Gleason review integration (2026-09-29) - [Full theorem-package review](reviews/2026-09-28-gleason-submission.md): 16 local modules, pinned source, mathematical assessment and executable evidence. - [Applied corrections and validation](reviews/2026-09-29-gleason-integration.md): source fixes, expanded dependency guard and reproduction commands for the integrated work. ## Current file-review evidence - [Measurement and record review](reviews/2026-09-30-measurement-records.md): three full reviews, mathematical improvements, current inventory reconciliation and the selected 30-file pilot. - [Corpus execution plan](corpus-review-plan.md): review order, pilot and renewal queue.