From 1d8267eaaf7e75b91e4ff93d13a0e95a10d47b37 Mon Sep 17 00:00:00 2001 From: Claude Date: Sat, 31 Oct 2026 17:48:46 +0000 Subject: [PATCH 1/3] Add a source-bound status audit of the elder-selection chain Classify every link of the Theorem A/B/C and lower-bound chain under the owner's five-value taxonomy (Proved / Proved under assumptions / Conditional / Conjectural / Refuted), with the event each estimate is about, its proof form, every nonauthor reader, exact source identities and the open obligation per link. Record the formal countermodel of Math-#543 (constant-one pair law) as the refutation of the interface-level implication, four surface inconsistencies, an independence table and a metrics baseline. Propose PROOF_INDEX / STATUS wording for the integrator without editing any canonical surface. CLAIMS.json is the machine-readable dependency list; check_claims.py verifies the cited blob, byte and SHA-256 identities and the dependency graph (custody only, not a Layer A auditor); test_check_claims.py covers tampered digests, wrong lengths, unknown statuses, dangling dependencies and unreviewed "Proved" entries in both interpreter modes. Scientific effect NONE. Answers the owner's assessment of 2026-10-31 (main#686 comment 6546811519, SHA-256 73ea6544...). Co-Authored-By: Claude Fable 5.1 Claude-Session: https://claude.ai/code/session_017Mi3hxjaxV45x6zo6o1ee3 --- .../AUDIT.md | 256 ++++ .../CLAIMS.json | 1247 +++++++++++++++++ .../README.md | 18 + .../check_claims.py | 137 ++ .../sources/OWNER_ASSESSMENT_20261031.txt | 144 ++ .../test_check_claims.py | 96 ++ 6 files changed, 1898 insertions(+) create mode 100644 reviews/elder_selection_chain_audit_20261031/AUDIT.md create mode 100644 reviews/elder_selection_chain_audit_20261031/CLAIMS.json create mode 100644 reviews/elder_selection_chain_audit_20261031/README.md create mode 100644 reviews/elder_selection_chain_audit_20261031/check_claims.py create mode 100644 reviews/elder_selection_chain_audit_20261031/sources/OWNER_ASSESSMENT_20261031.txt create mode 100644 reviews/elder_selection_chain_audit_20261031/test_check_claims.py diff --git a/reviews/elder_selection_chain_audit_20261031/AUDIT.md b/reviews/elder_selection_chain_audit_20261031/AUDIT.md new file mode 100644 index 00000000..0d456edf --- /dev/null +++ b/reviews/elder_selection_chain_audit_20261031/AUDIT.md @@ -0,0 +1,256 @@ +# Elder-selection chain: source-bound status audit under the owner's five-value taxonomy + +**Object:** `CL-ELDER-SELECTION-CHAIN-AUDIT-20261031-v1`. **Author:** Anthropic Claude, Claude Code session +`session_017Mi3hxjaxV45x6zo6o1ee3`, delegated AI work for Dylan Roy. **Kind:** source-bound review record. It states, +for every link of the Theorem A/B/C chain and the lower-bound chain, which event the link is about, what its proof is, +who read it, and what remains open, using the owner's five values (Proved / Proved under assumptions / Conditional / +Conjectural / Refuted). **Scientific effect: NONE.** It changes no theorem, review verdict, label, flag, register or +canonical surface; section 8 proposes wording for the integrator (B-003). It is not a register and creates none +(steward section 7). Same GitHub account as every lane: organizational-independence credit 0; human review 0. + +**Trigger.** The owner's assessment received in this session at 2026-10-31T15:14:11Z, recorded verbatim at +[main#686 comment 6546811519](https://github.com/d6g8k5htny-coder/main/issues/686#issuecomment-6546811519) and in +[`sources/OWNER_ASSESSMENT_20261031.txt`](sources/OWNER_ASSESSMENT_20261031.txt) (7,395 bytes, SHA-256 +`73ea6544478c3c8d0bfbae26ce58d4b04325b9f641fde31b9c6d3385eb3b187a`). Its items 1, 2, 4, 6 and 7 are addressed here; +item 3 (unrestricted parameter estimates) is only located, not audited; item 5 (external specialist review) is the +owner's to commission, and section 5 says what to hand the reviewers. + +**Source cuts.** Math- `19a55750451b8016b50d1d20be6a18c021807ecb`; main `4712fc13b99a4e4df68481663657b2952ce858fe`; +Math-#543 head `9e89980bfa4f281e00550f481202033775f35485`. Every cited file is bound by repository, commit, path, git +blob, byte length and SHA-256 in [`CLAIMS.json`](CLAIMS.json); `python3 -B -S check_claims.py` verifies those identities +and the dependency graph against the local object store. The checker certifies custody only. + +## 0. Summary + +The chain already contains the two things the owner's items 1 and 2 ask for, and lacks the third. + +- **The actual event is defined and its deterministic core is kernel-checked.** `p_r` is the `Q^W`-probability of the + Borel event `{d_f(M) = f(S)}` on the Morse distinct-value locus, with `d_f(M)` the maximin connection level (P + section 8). Every estimate of the chain refers to this `p_r`, including both lower bounds (E0). The step from the + local good event `G_r` to "S is the global elder partner" is the deterministic cap theorem, and it is kernel-checked + on the torus from the C⁴ hypotheses of CAP (1) (D1). That theorem is the topological witness of item 2: (C1) a + ceiling on the cap, (C2⁺) a frontier bound attained only at `S`, (C3) a path through `S` to an older point staying + above `f(S)`. No critical point elsewhere can change the pairing once these hold. +- **The probabilistic links are prose, read by two AI reviewers, one of them this session.** `Q^W(G_r^c) ≤ C r³` (A-II), + the normalizer floor (A-I), the Kac–Rice bridge (B, C) and both lower bounds (L2, L3) are complete written proofs + whose nonauthor reads are by xAI/Grok and by this session only. For the lower bounds, the cap second pass and the + reconciliation, this session is the sole nonauthor reader. There is no human review and no review outside those two + providers anywhere in the chain. +- **No kernel-checked statement of any probabilistic theorem about the actual event exists.** The statement register + makes `p` a primitive field. The formal countermodel the owner refers to is Math-#543's `oneLaw` (`p ≡ 1`): it proves + the upper shape and contradicts the two-sided shape, so at the interface level "upper selection ⇒ positive lower rate" + is Refuted (F1), and the register's shapes do not pin down the theorems (F2). Closing this needs the regression law as + a measure in Lean, which the register itself classifies as Mathlib-blocked. + +Status histogram over the 23 claims: Conditional 1, Conjectural 4, Proved 12, Proved under assumptions 3, Refuted 3. Open obligations listed: 35. Claims whose only nonauthor +reader is this session: D1, L2, L2b, L3. + +| ID | Claim | Event | Status | Evidence | Nonauthor readers | Open | +|---|---|---|---|---|---|---| +| E0 | The elder event and p_r | elder | **Proved** | prose | Anthropic session_017, xAI/Grok | 1 | +| D1 | Deterministic cap theorem: CAP (1) on the cylinder forces the global elder partner | deterministic | **Proved** | kernel + prose | Anthropic session_017 | 2 | +| A-I | Full-normalizer floor Z_r >= z_* r^2 and Z_r / r^2 -> z_0 > 0 | normalizer | **Proved** | prose | Anthropic session_017, xAI/Grok | 2 | +| A-II | Boundary layer: Q^W(G_r^c) <= C r^3 (P (7.8)) | good-jet | **Proved** | prose | Anthropic session_017, xAI/Grok | 3 | +| A | Theorem A (P 1.1): 0 <= 1 - p_r <= C r^3 on compact B, K, all frames, 0 < r <= r_* | elder | **Proved** | prose | Anthropic session_017, xAI/Grok | 3 | +| L2 | Theorem E1: planar lower bound 1 - p_r >= c r^3 | elder | **Proved** | prose | Anthropic session_017 | 2 | +| L2b | Corollary E2: c r^3 <= 1 - p_r <= C r^3 in d = 2 | elder | **Proved** | prose | Anthropic session_017 | 0 | +| L3 | Every-d lower bound (OA-ELDER-DIMENSION-LIFT Theorem A (A2)-(A3)) | elder | **Proved** | prose | Anthropic session_017 | 2 | +| L3b | CL-ELDER-LOWER-ALL-D Theorem E1_d (soft-eigenplane lift) | elder | **Conjectural** | prose | none located | 1 | +| LP-P | Lemma P, Lemma P', Proposition 3.3: the extra-window-saddle criterion on the pinned planar cubic | deterministic | **Conjectural** | prose | none located | 1 | +| LP-LP | Theorem LP: liminf r^{-3}(1 - p_r) >= a_1 + a_2 = nu_1 + nu_2 - beta_far (planar) | elder | **Conditional** | prose | none located | 2 | +| LP-EQ | Identified selection-failure law r^{-3}(1 - p_r) -> alpha_1 + alpha_2 (Theorem S planar, Theorem F every d; Conjecture LP=) | elder | **Conjectural** | prose | none located | 2 | +| W | Window critical-point counts: E_{Q^W} N_{r,j}(torus minus pins, between-pin height window) <= C r^3 | candidate-count | **Proved** | prose | Anthropic session_017, xAI/Grok | 2 | +| F5 | Identification of the maximin level with the H0 bar death and of elder-selected pairs with finite bars | deterministic | **Proved under assumptions** | prose | Anthropic session_017, xAI/Grok | 1 | +| B | Theorem B (P 1.2): compact-window densities and the O(ell^{2/3}) difference | bars-vs-candidates | **Proved** | prose | Anthropic session_017, xAI/Grok | 2 | +| C | Theorem C (P 1.3): unrestricted leading density nu_cand^all ~ nu_eld^all ~ c_{d,L} ell^{-1/3} | bars-vs-candidates | **Proved** | prose | Anthropic session_017, xAI/Grok | 2 | +| U | Theorem U: the unrestricted selection difference does not vanish | bars-vs-candidates | **Proved** | prose | OpenAI | 1 | +| C7 | Candidate C7: unrestricted difference rate 0 <= nu_cand^all - nu_eld^all <= C ell^{2/3} | bars-vs-candidates | **Refuted** | prose | OpenAI | 0 | +| RHO | Rejected density rho_rej(ell) = nu_cand^all - nu_eld^all = O(1) (C7-total Theorem K; ledger DP-07) | bars-vs-candidates | **Conjectural** | prose | none located | 1 | +| F1 | Interface-level implication "upper selection shape implies two-sided rate shape" for every PairLawInterface | formal-interface | **Refuted** | kernel | OpenAI | 1 | +| F2 | The register shapes are kernel-checkable statements of Theorems A, B, C, E2, E4 about the actual event | formal-interface | **Refuted** | register text | OpenAI | 1 | +| F3 | Formal elder bridge: trap data give the C263 death and one ordinary-H0 interval summand | deterministic | **Proved under assumptions** | kernel | OpenAI | 1 | +| F4 | Compact Morse adapter: the trap data exist for every nonessential nondegenerate maximum of a smooth periodic Morse-distinct function | deterministic | **Proved under assumptions** | prose | OpenAI | 2 | + +Evidence "kernel" means a Lean kernel check with only `propext`, `Classical.choice`, `Quot.sound`; "prose" means a +written proof. "Nonauthor readers" names providers; every one is an AI lane on the shared GitHub account. + +## 1. The actual event, and where every estimate attaches (owner item 1) + +**Definition (P l.290–297, E1 lower-bound packet section 1, E2 replacement for measurability).** For fixed pins and a +sample `f`, + + d_f(M) = sup { min_t f(γ(t)) : γ(0) = M, f(γ(1)) > f(M) }, sup ∅ = −∞. + +The elder event is `{d_f(M) = f(S)}` on the Morse distinct-value locus, which has full `Q` and `Q^W` measure for +each fixed `r, b, k, R` (P section 8). It is Borel: `{d_f(M) > h}` is a countable union of rational polygonal-path +tests. `p_r(b,k,R) := Q^W_r(elder event)`. On that locus the event is "S is the global ordinary superlevel elder death +partner of M" (F5). + +**Lean.** The deterministic sets are definitions in `frontiers/cap_first_exit_lean_20261002/CapFirstExit.lean`: +`connectionLevels f M b` (l.124, the maximin set), `CapHyp.joinedLevels` (l.211, closed-superlevel merge levels) and +`CapHyp.olderPeakLevels` (l.332, levels at which the component of `M` holds a strictly higher local maximum). Under +`CapHyp` they have the same greatest element `s = f(S)` (`maximin_eq` l.141, `elder_death_level` l.217, +`elder_death_level_peak` l.338). C263 `SuperlevelBar.lean` defines the Euclidean strict-clearance death with an explicit +empty default (TF-SCOPE). What does not exist in Lean: the event as a set of sample paths, the law `Q_r`, the weight +`W_r`, and `p_r` as a measure of the event. In `specifications/lean_statements_20261010/Specifications/Lifetime.lean` +the pair law is + + structure PairLawInterface ... where + p : ℝ → ℝ → ℝ → OrthonormalBasis (Fin d) ℝ (E d) → ℝ + p_nonneg : ... p_le_one : ... + +with the docstring "no field ties `p` to them" (l.455–467). That is the whole of the formal gap of item 1. + +**The formal countermodel.** Math-#543, `cbz/lifetime_pair_law/CounterexampleE4.lean` (blob `3b7712e4`, 1,743 bytes): +`oneLaw` sets `p := fun _ _ _ _ => 1`; `oneLaw_upper` proves `ParentUniformSelectionA` for it with `C = 0`; +`oneLaw_refutes_twoSided` derives `c (r_*/2)³ ≤ 0` from `PlanarTwoSidedRateE4` and contradicts `c > 0`. Lean-checked +at Math- `5a470bfd` (run 40957238766), with the three standard axioms; two OpenAI reviews passed the derivation at +scope and an Anthropic different-family review is open at this cut. The register's own round-4 triage already named +`p ≡ 0` and `p ≡ 1` as falsifying inhabitants (REG-SCOPE). Both say the same thing: the interface fixes neither the +upper nor the lower statement, so nothing formal connects the estimates to the event. + +**Where each estimate attaches.** + +| Estimate | What it is about | How it reaches the elder event | +|---|---|---| +| A-I floor `Z_r ≥ z_* r²` | the normalizer | denominator of every `Q^W` bound | +| A-II `Q^W(G_r^c) ≤ C r³` | the jet / sup-norm event `G_r` | through D1: `G_r ∩ Morse ⊆ elder event`, so `1 − p_r ≤ Q^W(G_r^c)` | +| D1 cap theorem | deterministic | the inclusion itself, kernel-checked | +| L2, L3 lower bounds | an explicit subevent of `elder event^c` (`d_f(M) ≥ b − k r³/4 > f(S)`) | direct: `1 − p_r ≥ Q^W(subevent) ≥ c r³` | +| W window counts | critical points in the height window | not used by Theorem A; explain the failure mechanism; not elder statements | +| B, C densities | bars (`ν_eld`) against candidates (`ν_cand`) | `ν_eld` integrates `p_r` against the candidate intensity (E2 Kac–Rice, P sections 9–11, 13); bars = elder-selected pairs by F5 | +| U, C7, RHO | the difference `ν_cand^all − ν_eld^all` | far pairs separated by a level curve are candidates and not bars | + +**Assumptions per step** are the `obligations` lists in `CLAIMS.json`. The two that carry the chain are: the full +normalizer floor and the boundary-layer integration (A-I, A-II: prose, no numerical constants, no formalization), and +the classical identification of elder death with bar death on the Morse locus (F5: an interface clause in Lean). + +## 2. The extra-critical-point problem (owner item 2) + +Two different things are true, and the chain keeps them apart. + +**Where an extra critical point cannot interfere.** On `G_r` the cap theorem gives more than a local picture. CAP +section 4: "Every critical point in D lies on the ridge, hence D contains precisely M and S." CAP (13): the ridge rises +to `g(2r) ≥ b + (11/96) r³ > b` inside the cylinder, so the older point `z` is reached without leaving `D`. CAP section +5: every path from `M` to a point above `b` crosses the frontier of `C`, where `f ≤ s`. Hence `d_f(M) = s` whatever +`f` does outside `D`, and whatever critical points exist there. This is the content of `CapHyp.barrier` (l.114), +`maximin_eq` and `elder_death_level_peak`, composed on the torus in `torus_elder_death_level_blocks` (l.5256) from the +C⁴ hypotheses of CAP (1). So the topological witness the owner asks for exists, is explicit, and is kernel-checked; its +one open item is a nonauthor alignment review of the Lean statements against CAP (I-4). + +**Where an extra critical point is exactly the failure.** Off `G_r` the pairing fails only through a competing saddle: +on the limiting cubic, `(M,S)` is the elder pair iff there is no extra strict-window critical point, and otherwise `M` +dies at the highest extra window saddle (LP-P; Math-#170 Theorem E part 3). The lower bounds build such a configuration +with probability `≥ c r³` by an explicit path above `f(S)` to a point above `b` (L2 planar, L3 every `d`). So "bound the +probability of configurations that violate the conditions" is answered on both sides: `≤ C r³` (A) and `≥ c r³` (L2, +L3), with the exact limit `α₁ + α₂` only a candidate (LP-EQ). + +**Good configurations at leading order.** P section 13 removes the compact-mark restriction by dominated convergence: +the unnormalized intensity has the target-growth majorant (13.4), `p_r → 1` pointwise by Theorem A on compact +neighbourhoods (P l.415), and the far population is `O(1)` (14.1). That is Theorem C's leading term (C). What does +not exist, and cannot, is a quantitative unrestricted rate: Theorem U shows `ν_cand^all − ν_eld^all ≥ c_* > 0` (U), so +candidate C7 is Refuted and the `O(1)` rejected density (RHO) is a candidate. "Counting candidate maximum–saddle pairs +is not the same as counting actual persistence bars" is therefore a theorem of the project, at leading order and in +the difference, not a gap: the two counts agree to leading order `ℓ^{−1/3}` and differ by a positive bounded-below +amount. + +**What remains open mathematically.** (a) The alignment review of D1 (I-4). (b) Formalization of anything +probabilistic (A-I, A-II, E2), blocked on a Gaussian-field measure in Lean. (c) The bar bijection on the Morse locus +(F5) as a theorem rather than an interface clause: the elder bridge (F3) and the compact Morse adapter (F4) give one +interval summand per maximum from trap data, and torus_filtration lists once-counting, interval decomposition and bar +bijection as not established. (d) Nonauthor reads of LP-P, L3b, LP-EQ and RHO. + +## 3. Status per claim (owner item 4) + +The table in section 0 is generated from `CLAIMS.json`. The assignment rule is in that file and repeated here: status +follows the located evidence, not the author's label. "Proved" requires a complete proof of a fixed statement and a +kernel check or a nonauthor read with no open defect at the stated scope; an author-side complete proof with no located +nonauthor read is "Conjectural (candidate)". Under that rule four items the owner may have taken as settled are +candidates: the Claude every-`d` lift L3b (the OpenAI lift L3 is read), the deterministic cubic criterion LP-P (two +author-side derivations agree; no read located), the identified failure law LP-EQ (Math-#170 Theorem S, Math-#175 +Theorem F) and the rejected density RHO. The owner's item 4 names "the proposed next-order expansion and +rejected-density estimate" as not established; this record agrees on the second and did not locate a next-order +expansion of `ν_eld` in the chain (none is in the headline statements). + +Numerical coefficients: Theorem C states `c_{d,L}` as an identified Gaussian/angular integral (15.2) with no value. The +SIDE24 numerical coefficient packets are outside this audit. No numerical value of `C`, `r_*`, `z_*`, `c_{B,K}`, +`c_{d,L}`, `α₁ + α₂` or `c_*` is established anywhere in this chain, and none belongs in a headline statement. + +## 4. Inconsistencies between surfaces + +| ID | Finding | Proposal | +|---|---|---| +| I-1 | main `docs/DISCOVERY_PRIORITY_LEDGER_20260925.md` DP-02 (l.18, l.157) says parent sections 2–7 are **AMEND**; main `STATUS.md` D1 (l.88) and Math- `PROOF_INDEX.md` (l.27) record ACCEPT at existential scope from the 28 September reconciliation, whose residual `R-A3-congruence` is DISCHARGED by E1 (REC `LEDGER.json`). The ledger cell was aligned with `STATUS.md` of 26 September and repeated unchanged by the 28 October refresh ([main#669 6465885945](https://github.com/d6g8k5htny-coder/main/issues/669#issuecomment-6465885945)). | Root decides which surface governs (B-003). This record reads the ledger cell as stale. | +| I-2 | REC section 2 and PROOF_INDEX l.27: "No lower bound is asserted for d ≥ 3" (true of the reconciled D1 chain) while the every-`d` lift L3 was ACCEPTed and integrated (Math-#129); `STATUS.md` D1 notes it, PROOF_INDEX does not. | Add the pointer to the PROOF_INDEX D1 row. | +| I-3 | Math-#543 review 5707632836 records `CounterexampleE4.lean` as 1,720 bytes; blob `3b7712e4` is 1,743 bytes (1,720 is the character count). The SHA-256 in the README is correct. | Note only. | +| I-4 | `frontiers/cap_first_exit_lean_20261002` has no PROOF_INDEX row and `MANIFEST.json` records `alignment_review: NOT_CLAIMED`; the only kernel-checked link of the chain is invisible to a cold-start reader of PROOF_INDEX. | Add a row: "Proved (kernel); alignment review open". | + +## 5. Independence (owner item 5) + +Every nonauthor read in this chain, by performer: + +| Reader | Claims read | Provider distinct from the author | Human | +|---|---|---|---| +| xAI Grok (G1, G2, G4, IW, I5 records) | E0, A-I (partial), A-II (partial), A, W, F5, B, C | yes | no | +| Anthropic Claude session_017 (this session) | E0, D1 (prose), A-I, A-II, A, L2, L2b, L3, W, F5, B, C; reconciler of REC and D5 | yes | no | +| OpenAI (REV-U, REV-CE4-ENG, REV-CE4-FORMAL, REV-EB, REV-CMA) | U, C7, F1, F2, F3, F4 | REV-U yes; the others same provider as the authors | no | + +This session is also the author of Theorem U and of the Lean cap module `CapFirstExit.lean`, and the sole nonauthor +reader of D1 (prose), L2, L2b and L3. A reader of this record should weigh it accordingly: the person best placed to +have missed a shared blind spot in the chain's acceptance wrote this audit. + +External specialist review is the owner's to commission. What to hand the two reviewers, without the project's +interpretation: the parent `P` with `E1` and `E2` and `CAP` (the four immutable bytes in REC section 1), the planar +lower bound `EL` and the dimension lift `DL`, and this record's `CLAIMS.json` as a map. The two questions the owner +proposes already have candidate targets here: the weakest indispensable lemma of Theorem A is the pair A-I/A-II +(the normalizer floor and the eigenvalue-coordinate boundary-layer integration, P sections 5–7), and for Theorems B and +C the marked Kac–Rice identification of `Q^W` with the Palm law (E2). A counterexample hunt should start from the +exact failure mechanism (LP-P) and from the far-pair separation of Theorem U. + +## 6. Dependency lists (owner item 6) + +`CLAIMS.json` gives each claim its statement, event, status, evidence, exact sources (repository, commit, path, blob, +bytes, SHA-256), reviews (performer, record, verdict, independence flags), `depends_on` and `obligations`. +`check_claims.py` verifies that every cited blob exists in the local object store with the recorded length and digest, +that the recorded commit maps the path to that blob when the commit is available, that every status is one of the five +values, that every `depends_on`, source and review reference resolves, and that no claim is "Proved" without a review or +kernel check. `test_check_claims.py` exercises it, including tampered-digest, wrong-length, unknown-status, dangling +dependency and unreviewed-"Proved" failures, and byte-identical output in normal and optimized Python. The checker does +not read Lean, does not run any proof, and certifies no Layer A property; it is custody, not mathematics. The 44 bound +sources include two from `main` and two from the Math-#543 head; `--repo main=DIR` checks the former, and the latter +are checked when that head is fetched. + +## 7. Metrics baseline (owner item 7) + +Computed from this record at the cuts above; the values are in `CLAIMS.json` under `metrics`. + +| Metric | Value at this cut | +|---|---| +| Claims audited | 23 (Conditional 1, Conjectural 4, Proved 12, Proved under assumptions 3, Refuted 3) | +| Open obligations listed | 35 | +| Counterexamples recorded | 4: `p ≡ 1` (F1) and `p ≡ 0` (F2) at the interface, open pending a successor interface; Theorem U against C7, resolved; LOCAL_PAIRING C-01, resolved by withdrawing the sketch | +| Claims downgraded or withdrawn | 3: C7 → Refuted; the LOCAL_PAIRING v1.1 section-5 sketch withdrawn; the C7-total v1 memo retracted | +| Independent (human or external) reviews completed | 0 | +| Distinct AI reviewer providers on the probabilistic chain | 2 (xAI/Grok; Anthropic, this session) | +| Formalized results whose alignment with the intended statement a nonauthor confirmed | 1 (F3, same provider); D1 has no alignment review; the probabilistic theorems have no formal statement about the actual event | +| Failed verifications that led to corrections | 6: congruence erratum E1; section-9 replacement E2; wording W1 and the embedding radius; Theorem U v1 → v3; LOCAL_PAIRING C-01; Math-#543 metadata | + +## 8. Proposed wording for the canonical surfaces (for the integrator; not applied here) + +- PROOF_INDEX D1 row: append "Status (owner taxonomy, 2026-10-31 audit): Theorems A, B, C **Proved** at existential + scope, prose, nonauthor AI reads by xAI/Grok and Anthropic session_017 only; no formal statement about the actual + event (register BLOCKED_INTERFACE; Math-#543 countermodel). Deterministic cap step **Proved (kernel)** in + `frontiers/cap_first_exit_lean_20261002`, alignment review open. Lower bound `1 − p_r ≥ c r³`: **Proved** for d = 2 + (EL, read) and for every fixed d ≥ 3 (DL, read, Math-#129); CL-ELDER-LOWER-ALL-D is a candidate." +- main `docs/DISCOVERY_PRIORITY_LEDGER_20260925.md` DP-02 current-state cell: resolve against `STATUS.md` D1 (I-1). +- `STATUS.md` "How to verify" or the D1 row: name the formal-coverage boundary in one sentence: "No kernel-checked + statement of Theorems A, B or C about the elder event exists; `PairLawInterface.p` is primitive." +- A PROOF_INDEX row for `frontiers/cap_first_exit_lean_20261002` (I-4). + +## 9. Non-claims + +No theorem is proved, improved, weakened or withdrawn here. No review verdict is issued on any proof; the statuses +summarize located records under a stated rule. No numerical constant is claimed. Nothing here is independent review +in the owner's sense of item 5. The checker certifies file identities, not mathematics. Item 3 (unrestricted parameter +estimates: small gaps, large births and gaps, intermediate scales, chart boundaries, constant dependence) is located in +P sections 13–14 and the D5/C6 packets but not audited; it is the natural next record. diff --git a/reviews/elder_selection_chain_audit_20261031/CLAIMS.json b/reviews/elder_selection_chain_audit_20261031/CLAIMS.json new file mode 100644 index 00000000..12d41e9c --- /dev/null +++ b/reviews/elder_selection_chain_audit_20261031/CLAIMS.json @@ -0,0 +1,1247 @@ +{ + "schema": "elder-selection-chain-audit/v1", + "object": "CL-ELDER-SELECTION-CHAIN-AUDIT-20261031-v1", + "author": "Anthropic Claude, Claude Code session_017Mi3hxjaxV45x6zo6o1ee3, delegated AI work for Dylan Roy", + "scientific_effect": "NONE", + "kind": "source-bound review record; proposes PROOF_INDEX/STATUS wording for the integrator; not a register", + "trigger": { + "what": "owner assessment received 2026-10-31T15:14:11Z", + "recorded_at": "main#686 comment 6546811519", + "file": "sources/OWNER_ASSESSMENT_20261031.txt", + "bytes": 7395, + "sha256": "73ea6544478c3c8d0bfbae26ce58d4b04325b9f641fde31b9c6d3385eb3b187a" + }, + "cuts": { + "Math-": "19a55750451b8016b50d1d20be6a18c021807ecb", + "main": "4712fc13b99a4e4df68481663657b2952ce858fe", + "Math-#543 head": "9e89980bfa4f281e00550f481202033775f35485" + }, + "taxonomy": { + "Proved": "complete written proof (prose or Lean) of a fixed statement, with a kernel check or at least one nonauthor read recording no open defect at the stated scope", + "Proved under assumptions": "complete proof whose conclusion holds under explicitly listed hypotheses or interface clauses that the project has not established for the actual model", + "Conditional": "complete argument whose conclusion consumes a cited claim whose status is not Proved", + "Conjectural": "no complete nonauthor-read proof: author-side complete proofs with no located nonauthor read (candidates) and open conjectures", + "Refuted": "a counterexample or contradiction is recorded" + }, + "assignment_rule": "Status follows the located evidence, not the author's label. A nonauthor AI read earns Proved; the evidence field says who read it. No claim in this chain has verification independent of the project's AI reviewers, and none has human review.", + "event_vocabulary": { + "elder": "elder event {d_f(M) = f(S)} on the Morse-distinct locus (P section 8); p_r is its Q^W-probability", + "good-jet": "the event G_r of P (7.1)/l.216: the CAP (1) hypotheses hold on the pinned cylinder", + "deterministic": "a statement about a fixed function, no probability", + "candidate-count": "counts of critical points or candidate pairs, not elder-selected", + "bars-vs-candidates": "finite H0 bars (nu_eld) against candidate pairs (nu_cand)", + "normalizer": "the full normalizer Z_r", + "formal-interface": "a statement about the Lean interface objects of specifications/lean_statements_20261010" + }, + "sources": { + "P": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "imports/lifetime_parent_20260925/UNIFORM_MATRIX_CAP_AND_LIFETIME.md", + "blob": "dfed3b8d318a3ab1950957f393307733a4bef3f2", + "bytes": 40261, + "sha256": "9350ad6eaba6626b93c3dedeef9e2ff816e5cdf1c8318e85fb27499141c84bc7" + }, + "E1-ERRATUM": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "imports/lifetime_parent_20260925/ERRATUM_CONGRUENCE.md", + "blob": "213594d6ca6a86fb938110f4d166d9ce275a02d0", + "bytes": 1782, + "sha256": "bad7ef609c4ad8c41ad6af562c1b6807921e19a9d556ed793ad1a0db6e202028" + }, + "CAP": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "imports/lifetime_parent_20260925/MARKED_CYLINDER_CAP_PROOF.md", + "blob": "0633aca3c2a2882b0de4399da0a75d64c2e6b2e1", + "bytes": 15160, + "sha256": "0bf922b9203c29088b12388807aa0e2ecd020485eb0f6e919679841b5b2636fc" + }, + "E2-REPAIR": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/d1_section9_borel_repair_20260925/REPAIR.md", + "blob": "fe9b9ce4999908bb3814b500ee2d0ceb0c6f704a", + "bytes": 9062, + "sha256": "845abf9f9c99d672c2a10a887b5a2e7206a3d2de3d876f35f75ff6e2dc13e62f" + }, + "REC": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/d1_chain_reconciliation_20260928/RECONCILIATION.md", + "blob": "75da2597971510f843f8d90c743950cb8c177342", + "bytes": 23312, + "sha256": "451b9d7ffee072a73fc904cab891b6693e3f233808b89df10cce1b57036b65da" + }, + "REC-LEDGER": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/d1_chain_reconciliation_20260928/LEDGER.json", + "blob": "e49a15aee0399e797388fd6f0fdcaa6845846e97", + "bytes": 14446, + "sha256": "8276335175af875c82bbb0936bb2b2d9657771c66859a37bda7dfd04e945e328" + }, + "REV-C1-FILE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/d1_theorem_a_nonauthor_20260928/REVIEW.md", + "blob": "bc11369c41ebc76de5700df3f931939ccfc88b9b", + "bytes": 24105, + "sha256": "b7328d4bab97fa2d12622ca730dc698970febc1ea8bcbc0b24555e69e2a66392" + }, + "REV-SP-FILE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/d1_cap_elder_partner_second_pass_claude_20260930/REVIEW.md", + "blob": "a653dbc7bd8bd692433587f050ab8678713c297e", + "bytes": 30771, + "sha256": "dc6cfc863b5b9f36065ded609e9834370e49bbe5fcf8bc11ec6b9bc0e53b133a" + }, + "REV-G2-FILE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/d1_section9_borel_grok_lucas_20260928/REVIEW.md", + "blob": "48ea9ad5a3ae5a1515a3533a4c776800547a92bc", + "bytes": 4287, + "sha256": "a654ea6b52181fa46371c3a2ef9ee17e82f8e365f9de24857526c6dd48154d7e" + }, + "REV-EL-FILE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/d1_elder_lower_claude_20260928/REVIEW.md", + "blob": "179c1e6d3d4005b2e406b1bd4d9584caf5465793", + "bytes": 16640, + "sha256": "53bd5289a2060ce6c906023ed09a5aec457313e782135e1df86e743941839495" + }, + "REV-DL-FILE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/elder_dimension_lift_claude_20260928/REVIEW.md", + "blob": "e2b9fcdc8d24f6a4d532afb7d02627e4716158d1", + "bytes": 17030, + "sha256": "032bb4c79e4a985e5b53f5abb6bb06da4d643ec7345e4df9591e2046fb287e12" + }, + "REV-U-FILE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/c7_nonvanishing_openai_20260929/REVIEW.md", + "blob": "64794e5801b70d950c7e1202f7b6b7d07055cbb6", + "bytes": 10807, + "sha256": "68b6cd43351706745992c4af242abdc5bdabd46ab5ac99347f2c12b505e1a15d" + }, + "REV-IW-G-FILE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/d5_intermediate_window_grok_20260928/REVIEW.md", + "blob": "1b0998b97677ff3e2a22e4fc5d9fabe9eb6a1f76", + "bytes": 8897, + "sha256": "3295b16edf6be150e4dee233f7666d9e2de7681574caf81a99a7367898b036d3" + }, + "REV-IW-C-FILE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/d5_intermediate_window_claude_20260928/REVIEW.md", + "blob": "100ce8ecc120fcf4e3e4e6b9b039c08632e2711c", + "bytes": 14710, + "sha256": "fb3207802bf2d92ad138bc338a36f39bcd3a94440669b9aba9558e566f767976" + }, + "REV-I5-G-FILE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/d5_i5_planar_f_20260928/REVIEW.md", + "blob": "031c5364d1377e0f85ed62caadd8406fbbedc0ab", + "bytes": 3552, + "sha256": "527d66e72b079ef7c854987f7e9d44f2252b81aa0e8d9e7ef3b45deb02388659" + }, + "D5-REC": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/d5_reconciliation_20260929/RECONCILIATION.md", + "blob": "4062d76bdf3c60bc73f04295b88c245bcee395de", + "bytes": 12536, + "sha256": "14342b7eb0ad47c1c99b06be58fb6c901f5c7c72fc70f73e81de3dde50c23432" + }, + "EL": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/window_multiplicity_laws_20260928/ELDER_LOWER_AND_DENSITY_GAP.md", + "blob": "aaefc8da9668e819591dd9db0b34442394f2362e", + "bytes": 10229, + "sha256": "8ce05a43bde8e0d718f09e2da1ffe75d3ec5893489254afdbf5b58b602cdba0c" + }, + "DL": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/elder_dimension_lift_20260928/PROOF.md", + "blob": "7303bd791a68a1139251f0f6e403a9f7cc89b006", + "bytes": 23971, + "sha256": "b529fe3780e1909014b3cb144b9a74156d976bbdc19b6711d0c5b1a1da9e5669" + }, + "ELD": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/elder_lower_all_d_20260929/PROOF.md", + "blob": "7f41c9e315be0a7985d4c770a11a2cdf717bd46a", + "bytes": 16485, + "sha256": "88c6bfc7973fe832918f91e2ed1f47da3e51987ed9efc1a99648e82b2eb119e5" + }, + "LPAIR": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/local_pairing_20260930/LOCAL_PAIRING.md", + "blob": "1f8c9455b6d8bf8b22342cc705c2119a67f8dc76", + "bytes": 46742, + "sha256": "e01e1d097fb087fee013559f508ad58849da06f9a764eed1464286946d78331a" + }, + "LEG": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/local_elder_geometry_20260930/PROOF.md", + "blob": "ef2aa57959ea9f721bbf2316ce94cf616c1c9113", + "bytes": 39722, + "sha256": "f68038be79b46124b0f9b31205aa6e3682b34b6f5ef81f4a5697ae81f07cc46b" + }, + "CFE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/concave_fibre_elder_20260930/PROOF.md", + "blob": "923d3236b5291a29bd162937f172190db84dcef5", + "bytes": 23416, + "sha256": "5681896a6808f48fb7508914d0860ec508af7646e9e706fc260d935dd951c75e" + }, + "U": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/unrestricted_selection_difference_20260929/PROOF.md", + "blob": "5a55b179a974a90c65d257f3f93765cc6fb30bb8", + "bytes": 13028, + "sha256": "93a63002d32fd5c35ae994fe28b1205d8098026d6ddde19c67d4977dcf5817a9" + }, + "IW": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/intermediate_window_20260928/PROOF.md", + "blob": "f53a527ce0204fda271f24730b62b4223a5e43ec", + "bytes": 27775, + "sha256": "b3eb9456d7058b5b78149cfca4e072679beda85cd1a0293d209664a542db66cf" + }, + "C7T": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/c7_total_bounded_20260929/PROOF.md", + "blob": "28748b086ec6761ef467dc67cba475fbdf8b7455", + "bytes": 14982, + "sha256": "16c0efe372c1b1f26460aaa109287fed78e6fbd2dd00a0530998db296278dd01" + }, + "CFE-LEAN": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/cap_first_exit_lean_20261002/CapFirstExit.lean", + "blob": "01744e2cf571049d7d1557b9ed54646881d341cf", + "bytes": 326607, + "sha256": "040aa91fcbdf25ccdaced1c828dd779318217a7c41ac928f87e8d17a8d3c3ef0" + }, + "CFE-ALIGN": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/cap_first_exit_lean_20261002/ALIGNMENT.md", + "blob": "ef05eb264ea219771fa37d7ecd10985d55846c48", + "bytes": 57035, + "sha256": "91b086c04d083d2220821c8b6900e2bd070ad11f4be03bf2255bdbd83d677f21" + }, + "CFE-MANIFEST": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/cap_first_exit_lean_20261002/MANIFEST.json", + "blob": "5252de50e9dbe5f733ec5442e2c61fc81eb76e55", + "bytes": 22679, + "sha256": "630d5edeaf975918c1c1b2648e32478c06d14a8e22a2a5e43ccb587e1cf04d8f" + }, + "LIFETIME-LEAN": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "specifications/lean_statements_20261010/Specifications/Lifetime.lean", + "blob": "f2508f7f2e3fd42f22f5aa52bc5ead5ff9e73b9a", + "bytes": 51383, + "sha256": "90f4892608e953ecc696bb32ea159d1c178ceb8a1c6edfd11acfceb14af462c6" + }, + "REG-SCOPE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "specifications/lean_statements_20261010/SCOPE.md", + "blob": "7dc499e939108f8912ba73807fa0ea32592114d0", + "bytes": 286891, + "sha256": "0917ba26c1b97d78f49da78f59981fe654012f18de4ab099aaaf4e29a62b9723" + }, + "REG-README": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "specifications/lean_statements_20261010/README.md", + "blob": "2eb556620c56fc13a082fb83dbaf7981d2d2d8ef", + "bytes": 104907, + "sha256": "40ca3a949df7794f12555c9398c37daad4c380c0c463267ec019a5593cdb7bf1" + }, + "EB-LEAN": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/asymmetric_elder_bridge_20261017/ElderBridge.lean", + "blob": "a59ee9daab7b67eafd7e4f806b27244bcc5caed1", + "bytes": 3683, + "sha256": "6b0be1116f6b357b0ffbda202c572c27c2788b98bee639e228375a5ebeea1fa9" + }, + "EB-REVIEW-FILE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/asymmetric_elder_bridge_20261017/review/NONAUTHOR_ELDER_BRIDGE_REVIEW.md", + "blob": "5e744c58db65bb705ba241cb8cbeef84ead91fd5", + "bytes": 26691, + "sha256": "aab177777595d0e725b779258fd5042e78771d4f5f89427ccd979d5949c36b43" + }, + "CMA-PROOF": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/compact_morse_elder_adapter_20261018/PROOF.md", + "blob": "bbfa404260bcbfb4cbc03e2021a26ce0b95d3dcb", + "bytes": 25662, + "sha256": "39f92734da35aa4c3ad97ea9d8e194cea4ee3a0b23054a8a412f331294d74762" + }, + "CMA-REVIEW-FILE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/compact_morse_elder_adapter_20261018/reviews/NONAUTHOR_REVIEW.md", + "blob": "955357eeefdd61ff1bb69005c28b9cc45194f1ff", + "bytes": 16983, + "sha256": "ab8ab0084449e676f0466510e31f8e0a84b36616d1d4f7488f470c47a30a4ede" + }, + "CMA-ACCEPT": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "frontiers/compact_morse_elder_adapter_20261018/ROOT_ACCEPTANCE.json", + "blob": "3139c6102ea2fef01098bf1ce2715a684cc36b91", + "bytes": 1502, + "sha256": "7b5d4a92ab87abef35722ac2446849c5f96c6bb2b3bc7383fb9956dc469c83f4" + }, + "TF-SCOPE": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "companions/torus_filtration_20261011/SCOPE.md", + "blob": "a68e8497ad198981c2973e66de606640d7aaf778", + "bytes": 41948, + "sha256": "ac7ecaba59c6ff0912f3064c4cfe860dee256b593e248b0e0322a67b902892cf" + }, + "PROOF-INDEX": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "PROOF_INDEX.md", + "blob": "6b406a44acff3005b1112e932b9f053324e96632", + "bytes": 37988, + "sha256": "4ad88f3e5e43a4caaada8124eade96114336b7c80ca5cbaeb2aab934f7862c58" + }, + "CANDIDATES": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "reviews/candidates_pending_20260928/CANDIDATES.md", + "blob": "02455f0561d51990e1486e5d4fbb4229c8a34f5a", + "bytes": 9465, + "sha256": "cbaa71b55190ac234b8363614b7e57387f60c33b393ea102408eb88ec1e44513" + }, + "REACH": { + "repo": "Math-", + "commit": "19a55750451b8016b50d1d20be6a18c021807ecb", + "path": "docs/integration/2026-10-02-proof-reachability.md", + "blob": "c5dd02757723740ed3ec6368ffffbe984a9e2a16", + "bytes": 27720, + "sha256": "ead58da5dff5cc2a0bd5cdca9826177a1b2652173b8528c3429c3b0685a76345" + }, + "CE4-LEAN": { + "repo": "Math-", + "commit": "9e89980bfa4f281e00550f481202033775f35485", + "path": "cbz/lifetime_pair_law/CounterexampleE4.lean", + "blob": "3b7712e41e8a6b0569c31c549a83857d9fe2ada6", + "bytes": 1743, + "sha256": "812ba30d02f16a69abdcb8b23d28fa4a8fe68e3a016911d7f08e98ebe04cdfc7" + }, + "CE4-README": { + "repo": "Math-", + "commit": "9e89980bfa4f281e00550f481202033775f35485", + "path": "cbz/lifetime_pair_law/README.md", + "blob": "eaab733e57026c501ae5bde0cff5f5ef081f70e4", + "bytes": 2514, + "sha256": "f347dc9da375411866f11ead4867678e71a6cbf133c5f7d7060504e3652634e3" + }, + "MAIN-LEDGER": { + "repo": "main", + "commit": "4712fc13b99a4e4df68481663657b2952ce858fe", + "path": "docs/DISCOVERY_PRIORITY_LEDGER_20260925.md", + "blob": "30e8de81929f70df7fc0fc797aed6bb7cfa95f3a", + "bytes": 23444, + "sha256": "f1a4fa748faabd383249307fa1d7e7145ce0dc5ed97851944b09d510d5e83432" + }, + "MAIN-STATUS": { + "repo": "main", + "commit": "4712fc13b99a4e4df68481663657b2952ce858fe", + "path": "STATUS.md", + "blob": "d209fe1183ece223944e0267202c1f513dbe4f8d", + "bytes": 18072, + "sha256": "677bb4e77562649af4f34d98dd0fe188c3f6e54f059ec88bbe22738f997645fe" + } + }, + "reviews": { + "REV-G1": { + "performer": "xAI Grok 4.7 via Cursor, bc-ce4bf0bc", + "record": "main#63 comment 5841570965", + "verdict": "ACCEPT D1-A to D1-E", + "scope": "P sections 8-15 and the statements", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-C1": { + "performer": "Anthropic Claude, Claude Code session_017Mi3hxjaxV45x6zo6o1ee3 (the author of this record)", + "record": "Math- reviews/d1_theorem_a_nonauthor_20260928/REVIEW.md (REV-C1-FILE) and main#63 comment 5873873903", + "verdict": "ACCEPT A1-A7, CAP, Theorem A and E2 C1-C6; wording W1", + "scope": "P sections 2-8 with E1, CAP sections 1-5, E2", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-G2": { + "performer": "xAI Grok, Lucas lane (editorial transcription by session_017)", + "record": "Math- reviews/d1_section9_borel_grok_lucas_20260928/REVIEW.md (REV-G2-FILE)", + "verdict": "E2 C1-C6 ACCEPT (narrative record; runner absent)", + "scope": "E2 Borel repair", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-G4": { + "performer": "xAI Grok 4.7 via Cursor, bc-81faa729 and bc-dfbce246", + "record": "main#67 comments 5841270276 and 5841782206", + "verdict": "partial re-derivation of P (6.2), the depth-failure integral, the m=1 split, the far and M4 Markov pieces", + "scope": "P sections 6-7 (partial, while reviewing D2)", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-C2": { + "performer": "Anthropic Claude, Claude Code session_017Mi3hxjaxV45x6zo6o1ee3 (the author of this record)", + "record": "Math- reviews/d1_chain_reconciliation_20260928/RECONCILIATION.md section 6 (REC)", + "verdict": "P sections 11-12 ACCEPT; spot re-derivations of (13.1), (13.4), (15.2)", + "scope": "P sections 11-15", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-SP": { + "performer": "Anthropic Claude, Claude Code session_017Mi3hxjaxV45x6zo6o1ee3 (the author of this record)", + "record": "Math- reviews/d1_cap_elder_partner_second_pass_claude_20260930/REVIEW.md (REV-SP-FILE)", + "verdict": "CAP sections 2-5 re-derived line by line, no gap; composition with P section 7 ACCEPT at Theorem A scope", + "scope": "CAP sections 2-5 and the G_r match", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-EL": { + "performer": "Anthropic Claude, Claude Code session_017Mi3hxjaxV45x6zo6o1ee3 (the author of this record)", + "record": "Math- reviews/d1_elder_lower_claude_20260928/REVIEW.md (REV-EL-FILE)", + "verdict": "Theorem E1 ACCEPT at existential scope; Corollary E2 ACCEPT as a composition with Theorem A", + "scope": "EL (E1)-(E8) and the local multiplicity candidate", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-DL": { + "performer": "Anthropic Claude, Claude Code session_017Mi3hxjaxV45x6zo6o1ee3 (the author of this record)", + "record": "Math- reviews/elder_dimension_lift_claude_20260928/REVIEW.md (REV-DL-FILE)", + "verdict": "DL Theorem A (A2)-(A3) ACCEPT at existential scope for every fixed d >= 3", + "scope": "DL sections 2-8", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-U": { + "performer": "OpenAI / ChatGPT (nonauthor review of candidate C7 and Theorem U)", + "record": "Math- reviews/c7_nonvanishing_openai_20260929/REVIEW.md (REV-U-FILE)", + "verdict": "ACCEPT Theorem U and (U2)-(U4), compatible-density-version convention", + "scope": "U v3", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-IW-G": { + "performer": "xAI Grok", + "record": "Math- reviews/d5_intermediate_window_grok_20260928/REVIEW.md (REV-IW-G-FILE)", + "verdict": "ACCEPT (finite exact checks of the Hermite frame and rank steps)", + "scope": "IW (I3)-(I5) algebraic interfaces", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-IW-C": { + "performer": "Anthropic Claude, Claude Code session_017Mi3hxjaxV45x6zo6o1ee3 (the author of this record)", + "record": "Math- reviews/d5_intermediate_window_claude_20260928/REVIEW.md (REV-IW-C-FILE)", + "verdict": "ACCEPT (I7)-(I9) and the continuum steps at their stated scope", + "scope": "IW", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-I5-G": { + "performer": "xAI Grok", + "record": "Math- reviews/d5_i5_planar_f_20260928/REVIEW.md (REV-I5-G-FILE)", + "verdict": "ACCEPT I15-I18 and Corollary F", + "scope": "IW (I5) and remote_collision Corollary F", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-D5-REC": { + "performer": "Anthropic Claude, Claude Code session_017Mi3hxjaxV45x6zo6o1ee3 (the author of this record)", + "record": "Math- reviews/d5_reconciliation_20260929/RECONCILIATION.md (D5-REC)", + "verdict": "binds the D5 window-count sources to nonauthor verdicts; summed pin-neighbourhood bound AMEND", + "scope": "D5 components", + "provider_distinct_from_author": true, + "same_github_principal": true, + "human": false + }, + "REV-CE4-ENG": { + "performer": "OpenAI / Codex research task loop", + "record": "Math-#543 review 5706003799 and comment 6510388884", + "verdict": "PASS_SCOPED_ENGINEERING_EVIDENCE / AMEND_SCOPED_METADATA / MATH_REVIEW_PENDING", + "scope": "nine-path delta, workflow, axiom and kernel logs", + "provider_distinct_from_author": false, + "same_github_principal": true, + "human": false + }, + "REV-CE4-FORMAL": { + "performer": "OpenAI / ChatGPT GPT-6", + "record": "Math-#543 review 5707632836 and comment 6513991759", + "verdict": "PASS_SCOPED_FORMAL_DERIVATION / CONDITIONAL_INTERFACE_BOUNDARY", + "scope": "oneLaw_upper and oneLaw_refutes_twoSided against Lifetime.lean", + "provider_distinct_from_author": false, + "same_github_principal": true, + "human": false + }, + "REV-EB": { + "performer": "OpenAI / Codex /root/c261a_successor_preparation_review_20261016", + "record": "Math- frontiers/asymmetric_elder_bridge_20261017/review/NONAUTHOR_ELDER_BRIDGE_REVIEW.md (EB-REVIEW-FILE)", + "verdict": "ACCEPT_SCOPED (local conditional bridge; 49 axiom targets; three negative controls)", + "scope": "ElderClearance, ElderInterval, ElderBridge", + "provider_distinct_from_author": false, + "same_github_principal": true, + "human": false + }, + "REV-CMA": { + "performer": "OpenAI / Codex /root/c292_geometry_review, with Root acceptance", + "record": "Math- frontiers/compact_morse_elder_adapter_20261018/reviews/NONAUTHOR_REVIEW.md (CMA-REVIEW-FILE) and ROOT_ACCEPTANCE.json (CMA-ACCEPT)", + "verdict": "ACCEPT_SCOPED_CONVENTIONAL", + "scope": "the compact Morse trap construction", + "provider_distinct_from_author": false, + "same_github_principal": true, + "human": false + } + }, + "claims": [ + { + "id": "E0", + "title": "The elder event and p_r", + "event": "elder", + "statement": "For fixed pins M, S and a sample f let d_f(M) = sup{ min_t f(gamma(t)) : gamma(0) = M, f(gamma(1)) > f(M) } (empty supremum -infinity). The elder event is {d_f(M) = f(S)} on the Morse distinct-value locus; it is Borel in f for fixed pins, and p_r(b,k,R) := Q^W_r(elder event). Theorems A, B, C, E1, the dimension lift and Theorem U all refer to this p_r (P l.29, l.290-297; EL section 1 \"exactly the p_r definition\").", + "status": "Proved", + "proof_form": "prose", + "evidence": "P section 8 with the E2 section-9 replacement; reads REV-G1, REV-C1 (section 8 and E2 C1-C6), REV-G2. Formal: the deterministic level sets are Lean definitions (CFE-LEAN connectionLevels l.124, joinedLevels l.211, olderPeakLevels l.332); the sample-path event and its Q^W-measure are not formalized; in the statement register p is a primitive field (LIFETIME-LEAN l.457-467).", + "sources": [ + "P", + "E2-REPAIR", + "EL", + "CFE-LEAN", + "LIFETIME-LEAN" + ], + "reviews": [ + "REV-G1", + "REV-C1", + "REV-G2" + ], + "depends_on": [], + "obligations": [ + "Formal: define the elder event as a set of sample paths and p_r as its Q^W-measure. This needs the regression law Q_r as a measure on smooth functions; the register classifies the parent Props as Mathlib-blocked (REG-README, REG-SCOPE)." + ] + }, + { + "id": "D1", + "title": "Deterministic cap theorem: CAP (1) on the cylinder forces the global elder partner", + "event": "deterministic", + "statement": "Let f be C^4 near D = [-2r,2r] x closedBall(0,2r) with critical pins M = (-r/2,0), S = (r/2,0), f(M) - f(S) = k r^3, lambda_min(-D_y^2 f(M)) > (4/(3k)) r M3^2 and r M4 <= 3k/10. Then on C = [-2r, r/2] x closedBall(0,2r): (C1) f <= f(M) on C; (C2+) f <= f(S) on the frontier of C with equality only at S; (C3) a path from M through S to a point z with f(z) > f(M) along which f >= f(S). Hence d_f(M) = f(S); on a compact locally connected space the largest level at which the closed-superlevel component of M contains a strictly higher local maximum is f(S). D contains exactly the critical points M and S, and every path from M to a higher point crosses the frontier of C, so no critical point elsewhere changes the pairing.", + "status": "Proved", + "proof_form": "lean+prose", + "kernel_checked": true, + "evidence": "Kernel-checked: CFE-LEAN ridge_capHyp_E (l.3162), torus_capHyp_blocks (l.5214), torus_elder_death_level_blocks (l.5256), axioms propext, Classical.choice, Quot.sound only (CFE-MANIFEST), Lean author session_017. Prose: CAP sections 2-5, read by REV-C1 and re-derived by REV-SP (both session_017). MANIFEST: mathematical_acceptance false, alignment_review NOT_CLAIMED.", + "sources": [ + "CAP", + "CFE-LEAN", + "CFE-ALIGN", + "CFE-MANIFEST", + "REV-SP-FILE" + ], + "reviews": [ + "REV-C1", + "REV-SP" + ], + "depends_on": [], + "obligations": [ + "Nonauthor alignment review of the Lean statements against CAP sections 1-5 (the Lean author is the same session as both prose reviewers).", + "No PROOF_INDEX row for frontiers/cap_first_exit_lean_20261002; only the reachability inventory lists it (REACH)." + ] + }, + { + "id": "A-I", + "title": "Full-normalizer floor Z_r >= z_* r^2 and Z_r / r^2 -> z_0 > 0", + "event": "normalizer", + "statement": "P sections 2-5 read with E1 (congruence factor diag(r^{-1/2}, I)) and W1: uniform contact frame, residual independence, endpoint identities and the full normalizer floor (5.4)-(5.5), uniformly over compact marks and all frames.", + "status": "Proved", + "proof_form": "prose", + "evidence": "REV-C1 full depth on A1-A4; REV-G4 partial (certified det T_r). Reconciled in REC section 3 component (I).", + "sources": [ + "P", + "E1-ERRATUM", + "REC", + "REC-LEDGER" + ], + "reviews": [ + "REV-C1", + "REV-G4" + ], + "depends_on": [], + "obligations": [ + "No numerical z_* or r_*.", + "Not formalized." + ] + }, + { + "id": "A-II", + "title": "Boundary layer: Q^W(G_r^c) <= C r^3 (P (7.8))", + "event": "good-jet", + "statement": "G_r = { lambda_min(-A_M) > (4/(3k)) r M3^2, r M4 <= 3k/10 } (P l.216). E_Q[W_r 1_{G_r^c}] <= C r^5 by the double soft-eigenvalue factor (P section 6), the eigenvalue-coordinate integration including corank >= 2 (7.2)-(7.4), the scalar m = 1 far branch (7.5)-(7.6) and the M4 Markov bound (7.7); dividing by the floor (5.5) gives (7.8). This is a statement about the jet and sup-norm event G_r, not about the elder event.", + "status": "Proved", + "proof_form": "prose", + "evidence": "REV-C1 full depth on A5-A7; REV-G4 re-derived (6.2), the depth-failure integral, the m = 1 split and the far and M4 Markov pieces.", + "sources": [ + "P", + "REC", + "REC-LEDGER" + ], + "reviews": [ + "REV-C1", + "REV-G4" + ], + "depends_on": [ + "A-I" + ], + "obligations": [ + "No numerical C or r_*.", + "Not formalized.", + "Together with A-I this is where the r^3 of Theorem A arises; it is the natural target for the external question \"weakest indispensable lemma\"." + ] + }, + { + "id": "A", + "title": "Theorem A (P 1.1): 0 <= 1 - p_r <= C r^3 on compact B, K, all frames, 0 < r <= r_*", + "event": "elder", + "statement": "For compact B and K in (0, infinity) there are r_* > 0 and C < infinity with 0 <= 1 - p_r(b,k,R) <= C r^3 for all 0 < r <= r_*, b in B, k in K and all orthonormal frames, with p_r the Q^W-probability of the elder event (E0). Proof: 1 - p_r = Q^W(elder event^c) <= Q^W(G_r^c) by D1 on the full-probability Morse locus (P section 8), and Q^W(G_r^c) <= C r^3 by A-II.", + "status": "Proved", + "proof_form": "prose", + "evidence": "REV-G1 and REV-C1 ACCEPT at existential scope; REV-SP confirms the composition; REC section 2 records ACCEPT. Formal: none about the actual event (see F2); the register shape ParentUniformSelectionA is a predicate on an uninterpreted p.", + "sources": [ + "P", + "E1-ERRATUM", + "E2-REPAIR", + "REC", + "REC-LEDGER", + "PROOF-INDEX", + "MAIN-STATUS", + "MAIN-LEDGER" + ], + "reviews": [ + "REV-G1", + "REV-C1", + "REV-SP" + ], + "depends_on": [ + "E0", + "D1", + "A-I", + "A-II" + ], + "obligations": [ + "Formal statement about the actual event: none exists (F2).", + "Surface inconsistency I-1: main ledger DP-02 says AMEND; main STATUS.md D1 and Math- PROOF_INDEX record ACCEPT at existential scope.", + "All nonauthor reads are by xAI/Grok or by session_017; no human or external review." + ] + }, + { + "id": "L2", + "title": "Theorem E1: planar lower bound 1 - p_r >= c r^3", + "event": "elder", + "statement": "d = 2, compact birth and positive gap marks, all frames: there are c, r_* > 0 with 1 - p_r >= c r^3 for 0 < r <= r_*. The proof exhibits an event of Q^W-probability >= c r^3 on which an explicit path from M to a point above the birth stays above f(S), so d_f(M) >= b - k r^3/4 > f(S) (EL (E3)): an extra window saddle preempts S.", + "status": "Proved", + "proof_form": "prose", + "evidence": "REV-EL ACCEPT at existential scope (session_017); CANDIDATES C1 records the integration (Math-#124) and that the author's algebra.py is absent.", + "sources": [ + "EL", + "REV-EL-FILE", + "CANDIDATES" + ], + "reviews": [ + "REV-EL" + ], + "depends_on": [ + "E0" + ], + "obligations": [ + "Sole nonauthor reviewer is session_017.", + "The author's original finite check is not publicly reproducible; the review's distinct checker does not recover it (CANDIDATES C1)." + ] + }, + { + "id": "L2b", + "title": "Corollary E2: c r^3 <= 1 - p_r <= C r^3 in d = 2", + "event": "elder", + "statement": "Composition of L2 with Theorem A.", + "status": "Proved", + "proof_form": "prose", + "evidence": "REV-EL ACCEPT as a composition; REC section 2 \"Sharpness in d = 2\".", + "sources": [ + "EL", + "REC" + ], + "reviews": [ + "REV-EL" + ], + "depends_on": [ + "A", + "L2" + ], + "obligations": [] + }, + { + "id": "L3", + "title": "Every-d lower bound (OA-ELDER-DIMENSION-LIFT Theorem A (A2)-(A3))", + "event": "elder", + "statement": "For every fixed d >= 3, compact marks and all frames, 1 - p_r >= c r^3 for small r, by one soft transverse direction and the planar path (DL sections 2-8; (A25)). \"No parent upper selection theorem entered this argument\" (DL l.361).", + "status": "Proved", + "proof_form": "prose", + "evidence": "REV-DL ACCEPT at existential scope (session_017); integrated by Math-#129; main STATUS.md D1 notes it as outside the reconciliation row.", + "sources": [ + "DL", + "REV-DL-FILE", + "MAIN-STATUS" + ], + "reviews": [ + "REV-DL" + ], + "depends_on": [ + "E0" + ], + "obligations": [ + "Sole nonauthor reviewer is session_017.", + "Surface wording I-2: REC section 2 and the PROOF_INDEX D1 row say \"no lower bound is asserted for d >= 3\" (scoped to the D1 chain) while this reviewed lift exists." + ] + }, + { + "id": "L3b", + "title": "CL-ELDER-LOWER-ALL-D Theorem E1_d (soft-eigenplane lift)", + "event": "elder", + "statement": "The planar E1 lifted to every fixed d >= 2 by a different route (Anthropic session_01NMeK).", + "status": "Conjectural", + "proof_form": "prose", + "evidence": "Candidate: complete author-side proof; \"nonauthor analytic review required\"; no nonauthor read located in Math- (PROOF-INDEX #149 \"proof candidate\"; REACH \"no\" PROOF_INDEX verdict).", + "sources": [ + "ELD", + "PROOF-INDEX", + "REACH" + ], + "reviews": [], + "depends_on": [ + "E0", + "L2" + ], + "obligations": [ + "Nonauthor analytic review." + ] + }, + { + "id": "LP-P", + "title": "Lemma P, Lemma P', Proposition 3.3: the extra-window-saddle criterion on the pinned planar cubic", + "event": "deterministic", + "statement": "On the limiting pinned cubic P_theta, off a null tie set, (M, S) is the elder pair iff the cubic has no extra strict-window critical point (n(theta) = 0); if n(theta) >= 1, M dies at the highest extra window saddle (LPAIR sections 2-3). Math-#170 Theorem E (LEG section 2) proves the same deterministic statement by another route.", + "status": "Conjectural", + "proof_form": "prose", + "evidence": "Two independent author-side derivations by different providers agree (Anthropic LPAIR; OpenAI LEG); exact finite controls (pairing_check.py). The LPAIR header's \"clean-context referee\" reads are not repository records, and the located Codex review (C85, comment 5396856086) covered slice C only. Under this record's rule a complete proof without a located nonauthor read is a candidate.", + "sources": [ + "LPAIR", + "LEG" + ], + "reviews": [], + "depends_on": [], + "obligations": [ + "Slice A nonauthor read (deterministic, planar cubic; any lane)." + ] + }, + { + "id": "LP-LP", + "title": "Theorem LP: liminf r^{-3}(1 - p_r) >= a_1 + a_2 = nu_1 + nu_2 - beta_far (planar)", + "event": "elder", + "statement": "The planar selection loss is bounded below by an explicit functional of the local cluster measure (LPAIR section 4), consuming [SC], [CUB], [TSL] and Lemma 5.1 (QS-E) at their own status.", + "status": "Conditional", + "proof_form": "prose", + "evidence": "Author-side; depends on LP-P and on interfaces not audited here.", + "sources": [ + "LPAIR" + ], + "reviews": [], + "depends_on": [ + "LP-P", + "E0" + ], + "obligations": [ + "Status of the consumed interfaces [SC], [CUB], [TSL], QS-E Lemma 5.1 (outside this audit).", + "Nonauthor read of slice B." + ] + }, + { + "id": "LP-EQ", + "title": "Identified selection-failure law r^{-3}(1 - p_r) -> alpha_1 + alpha_2 (Theorem S planar, Theorem F every d; Conjecture LP=)", + "event": "elder", + "statement": "The exact limit of the scaled failure probability and the law of the replacement partner (LEG sections 7-8 Theorem S; CFE section 5 Theorem F; LPAIR section 5 Conjecture LP=).", + "status": "Conjectural", + "proof_form": "prose", + "evidence": "AUTHOR_SIDE / HOLD pending qualified source-bound review (LEG l.4, CFE l.4); no nonauthor read located; REACH lists both with no PROOF_INDEX verdict.", + "sources": [ + "LEG", + "CFE", + "LPAIR", + "REACH" + ], + "reviews": [], + "depends_on": [ + "LP-P", + "A" + ], + "obligations": [ + "Qualified nonauthor review of Theorem E/S and of the hard-fibre lemma/Theorem F.", + "Keep alpha_1 + alpha_2 and any numerical value out of the headline until then (owner item 4)." + ] + }, + { + "id": "W", + "title": "Window critical-point counts: E_{Q^W} N_{r,j}(torus minus pins, between-pin height window) <= C r^3", + "event": "candidate-count", + "statement": "Planar, compact marks, all frames and indices: the expected number of critical points other than the pins with height in I_r = (b - k r^3, b) is O(r^3) (IW (I3)-(I5) with the collar and punctured-disk inputs; D5 reconciliation). IW section 1: \"This is not an elder-selection theorem\". The proof of Theorem A does not use these counts.", + "status": "Proved", + "proof_form": "prose", + "evidence": "REV-IW-G, REV-IW-C, REV-I5-G ACCEPT at stated scope; REV-D5-REC binds them (summed pin-neighbourhood bound AMEND).", + "sources": [ + "IW", + "D5-REC", + "REV-IW-G-FILE", + "REV-IW-C-FILE", + "REV-I5-G-FILE" + ], + "reviews": [ + "REV-IW-G", + "REV-IW-C", + "REV-I5-G", + "REV-D5-REC" + ], + "depends_on": [], + "obligations": [ + "Planar only; no all-height global count (D5-REC).", + "A count of candidate configurations; it enters no elder statement of this chain." + ] + }, + { + "id": "F5", + "title": "Identification of the maximin level with the H0 bar death and of elder-selected pairs with finite bars", + "event": "deterministic", + "statement": "On the Morse distinct-value locus, d_f(M) is the death level of the superlevel H0 class born at M (elder rule), and each finite bar has exactly one local-maximum birth and one index-(d-1) merging-saddle death, so the elder-selected candidate pairs are exactly the finite bars (P l.290-297, l.445).", + "status": "Proved under assumptions", + "proof_form": "prose", + "evidence": "Classical persistence/Morse theory asserted in P and consumed by Theorems B and C; read with sections 8 and 14 by REV-G1 and REV-C1. In Lean it is an interface clause (LIFETIME-LEAN elderPartner_iff l.348, deathHeight_maximin l.341) and, for the cap case only, CapHyp.elder_death_level / elder_death_level_peak (CFE-LEAN l.217, l.338); the compact Morse adapter (CMA-PROOF) supplies one interval summand per maximum, not the bar bijection; torus_filtration (TF-SCOPE) lists once-counting, interval decomposition and bar bijection as not established.", + "sources": [ + "P", + "LIFETIME-LEAN", + "CFE-LEAN", + "CMA-PROOF", + "TF-SCOPE" + ], + "reviews": [ + "REV-G1", + "REV-C1" + ], + "depends_on": [ + "E0" + ], + "obligations": [ + "Formal bijection between finite bars and elder-selected candidate pairs on the Morse locus." + ] + }, + { + "id": "B", + "title": "Theorem B (P 1.2): compact-window densities and the O(ell^{2/3}) difference", + "event": "bars-vs-candidates", + "statement": "For B, K of positive length: nu_cand ~ nu_eld ~ c_{B,K} ell^{-1/3} and 0 <= nu_cand - nu_eld <= C_{B,K} ell^{2/3}, where nu_eld counts candidate pairs that are elder partners (E0, F5). Proof: Theorem A plus the marked Kac-Rice interface (E2) and the radial ledger and pushforward (P sections 9-11).", + "status": "Proved", + "proof_form": "prose", + "evidence": "REV-G1, REV-C1, REV-C2; REC section 2.", + "sources": [ + "P", + "E2-REPAIR", + "REC" + ], + "reviews": [ + "REV-G1", + "REV-C1", + "REV-C2" + ], + "depends_on": [ + "A", + "E0", + "F5" + ], + "obligations": [ + "The Kac-Rice representation and pushforward are prose only.", + "No numerical c_{B,K} or C_{B,K}." + ] + }, + { + "id": "C", + "title": "Theorem C (P 1.3): unrestricted leading density nu_cand^all ~ nu_eld^all ~ c_{d,L} ell^{-1/3}", + "event": "bars-vs-candidates", + "statement": "All births, gaps and separations; essential class excluded. Near pairs: dominated convergence with the target-growth majorant (13.4) and p_r -> 1 pointwise from Theorem A on compact neighbourhoods (P l.415); far pairs: nu^far <= C (14.1); c_{d,L} is the integral (13.6)/(15.2). Leading term only: no unrestricted remainder or difference rate.", + "status": "Proved", + "proof_form": "prose", + "evidence": "REV-G1, REV-C1, REV-C2 (spot re-derivations of (13.1), (13.4), (15.2)); REC section 2.", + "sources": [ + "P", + "REC" + ], + "reviews": [ + "REV-G1", + "REV-C1", + "REV-C2" + ], + "depends_on": [ + "A", + "E0", + "F5" + ], + "obligations": [ + "No unrestricted difference rate; Theorem U shows none of order ell^{2/3} exists.", + "c_{d,L} is an identified integral with no numerical value; numerical SIDE24 coefficient work is outside this audit and belongs outside the headline until derived (owner item 4)." + ] + }, + { + "id": "U", + "title": "Theorem U: the unrestricted selection difference does not vanish", + "event": "bars-vs-candidates", + "statement": "nu_cand^all(ell) - nu_eld^all(ell) >= nu_cand^far - nu_eld^far >= c_* > 0 for 0 < ell <= ell_0; hence D(ell)/ell^alpha -> infinity for every alpha > 0 and c_* <= D(ell) = o(ell^{-1/3}).", + "status": "Proved", + "proof_form": "prose", + "evidence": "REV-U ACCEPT Theorem U and (U2)-(U4) (OpenAI); author session_017.", + "sources": [ + "U", + "REV-U-FILE" + ], + "reviews": [ + "REV-U" + ], + "depends_on": [ + "C" + ], + "obligations": [ + "Whether D is bounded is not decided by U (U section 5); see RHO." + ] + }, + { + "id": "C7", + "title": "Candidate C7: unrestricted difference rate 0 <= nu_cand^all - nu_eld^all <= C ell^{2/3}", + "event": "bars-vs-candidates", + "statement": "The compact-window rate (1.2) extended to the unrestricted densities (CANDIDATES C7).", + "status": "Refuted", + "proof_form": "prose", + "evidence": "Theorem U (U2): the difference is bounded below by c_* > 0.", + "sources": [ + "U", + "CANDIDATES" + ], + "reviews": [ + "REV-U" + ], + "depends_on": [ + "U" + ], + "obligations": [] + }, + { + "id": "RHO", + "title": "Rejected density rho_rej(ell) = nu_cand^all - nu_eld^all = O(1) (C7-total Theorem K; ledger DP-07)", + "event": "bars-vs-candidates", + "statement": "A gap-mark-explicit failure intensity giving a bounded unrestricted rejected density and finite inverse moments exactly for q > -1 (C7T sections 0-3).", + "status": "Conjectural", + "proof_form": "prose", + "evidence": "Author-side proof candidate (Anthropic session_01NMeK), \"nonauthor analytic review required\" (C7T l.3); the v1 memo was retracted and a normalized-form scope erratum added; no nonauthor analytic read located. The owner's item 4 names the rejected-density estimate as not yet established; this record agrees.", + "sources": [ + "C7T", + "MAIN-LEDGER" + ], + "reviews": [], + "depends_on": [ + "A-II", + "U" + ], + "obligations": [ + "Nonauthor analytic review of Theorem K." + ] + }, + { + "id": "F1", + "title": "Interface-level implication \"upper selection shape implies two-sided rate shape\" for every PairLawInterface", + "event": "formal-interface", + "statement": "For all G, I and every PairLawInterface p: ParentUniformSelectionA -> PlanarTwoSidedRateE4 (lower half included). Refuted by the constant law p = 1: oneLaw_upper proves the upper shape and oneLaw_refutes_twoSided contradicts the two-sided shape at r = r_*/2 (CE4-LEAN). The register also names p = 0 as falsifying the universal closure of ParentUniformSelectionA itself (REG-SCOPE round-4 triage).", + "status": "Refuted", + "proof_form": "lean", + "kernel_checked": true, + "evidence": "Lean-checked at Math- commit 5a470bfd (run 40957238766, job 123740377857; axioms propext, Classical.choice, Quot.sound), relocated byte-identically to cbz/lifetime_pair_law at Math-#543 head 9e89980b; REV-CE4-ENG and REV-CE4-FORMAL PASS at scope; an Anthropic different-family review (01XJMJ, Math-#543 comment 6545265266) is open at this cut. Byte count: blob 3b7712e4 is 1,743 bytes; the \"1,720\" in the formal review counts characters (I-3).", + "sources": [ + "CE4-LEAN", + "CE4-README", + "LIFETIME-LEAN", + "REG-SCOPE" + ], + "reviews": [ + "REV-CE4-ENG", + "REV-CE4-FORMAL" + ], + "depends_on": [], + "obligations": [ + "Math-#543 integration is behind main and has pending metadata AMENDs (not this record's claim)." + ] + }, + { + "id": "F2", + "title": "The register shapes are kernel-checkable statements of Theorems A, B, C, E2, E4 about the actual event", + "event": "formal-interface", + "statement": "ParentUniformSelectionA, ParentCompactWindowDensityB, ParentUnrestrictedLeadingDensityC, PlanarElderLowerE2, PlanarTwoSidedRateE4, PlanarDensityGapE11 (LIFETIME-LEAN l.479-623) pin down the parent theorems.", + "status": "Refuted", + "proof_form": "lean", + "evidence": "The register says so itself: PairLawInterface.p has no field tying it to G or I (LIFETIME-LEAN l.455-467 docstring); the round-4 triage marks the Props BLOCKED_INTERFACE with the falsifying inhabitants p = 0 and p = 1, and \"specified refers to the statement shape at the source's object, which is not constructed\" (REG-SCOPE). F1 is the kernel-checked instance. Consequently \"does the formal verification cover the exact theorem being advertised?\" is answered no for every probabilistic statement of this chain.", + "sources": [ + "LIFETIME-LEAN", + "REG-SCOPE", + "REG-README", + "CE4-LEAN" + ], + "reviews": [ + "REV-CE4-FORMAL" + ], + "depends_on": [ + "F1" + ], + "obligations": [ + "Successor interface in which p is the Q^W-measure of the E0 event under a constructed law (the register's own route: a numbered successor interface carrying the law of p or a Kac-Rice intensity; Mathlib-blocked, estimated weeks)." + ] + }, + { + "id": "F3", + "title": "Formal elder bridge: trap data give the C263 death and one ordinary-H0 interval summand", + "event": "deterministic", + "statement": "From explicit data (closed D containing M, a height ceiling, the frontier bound, an older point outside D and a path at the death level) the literal C263 strict-clearance death is d and one ordinary-H0 interval [-g(M), -d) splits off naturally (ElderClearance, ElderInterval, ElderBridge).", + "status": "Proved under assumptions", + "proof_form": "lean", + "kernel_checked": true, + "evidence": "Local Lean 4.34.1 compilation and kernel replays, 49 named targets with only the three standard axioms, three negative controls (EB-REVIEW-FILE); REV-EB ACCEPT_SCOPED (same provider as the authors). C263 registration remains under FORMAL HOLD (TF-SCOPE).", + "sources": [ + "EB-LEAN", + "EB-REVIEW-FILE", + "TF-SCOPE" + ], + "reviews": [ + "REV-EB" + ], + "depends_on": [], + "obligations": [ + "Chart construction, torus transfer, canonical elder labels, exhaustive decomposition and any Gaussian law are outside the bridge (EB-REVIEW-FILE)." + ] + }, + { + "id": "F4", + "title": "Compact Morse adapter: the trap data exist for every nonessential nondegenerate maximum of a smooth periodic Morse-distinct function", + "event": "deterministic", + "statement": "Closed domain = closure of the strict-superlevel path component of M at the literal C263 death; height and frontier bounds; a path to a strictly older maximum at the closed death level; hence one interval summand per supplied maximum (CMA-PROOF).", + "status": "Proved under assumptions", + "proof_form": "prose", + "evidence": "Conventional proof with named classical imports (Morse lemma, inverse function theorem, quotient atlas); REV-CMA ACCEPT_SCOPED_CONVENTIONAL and Root acceptance (same provider as the author). Morse admission for the Gaussian law, a unique saddle partner, exhaustive once-counting and counts remain separate (CMA-ACCEPT).", + "sources": [ + "CMA-PROOF", + "CMA-REVIEW-FILE", + "CMA-ACCEPT" + ], + "reviews": [ + "REV-CMA" + ], + "depends_on": [ + "F3" + ], + "obligations": [ + "No Lean implementation of this adapter.", + "Unique saddle partner and the bar bijection are not supplied (see F5)." + ] + } + ], + "inconsistencies": [ + { + "id": "I-1", + "finding": "main docs/DISCOVERY_PRIORITY_LEDGER_20260925.md DP-02 (l.18, l.157) records parent sections 2-7 as AMEND (\"reopened for independent review with an additive section-5 congruence erratum\", aligned with STATUS.md of 2026-09-26, repeated by the 2026-10-28 refresh at main#669 comment 6465885945), while main STATUS.md D1 (l.88) and Math- PROOF_INDEX.md (l.27) record ACCEPT at existential scope from the 2026-09-28 reconciliation, whose residual R-A3-congruence is DISCHARGED by E1 (REC-LEDGER).", + "proposal": "Root (B-003) decides which surface governs; this record reads the ledger cell as stale and proposes the wording in section 8.", + "sources": [ + "MAIN-LEDGER", + "MAIN-STATUS", + "PROOF-INDEX", + "REC-LEDGER" + ] + }, + { + "id": "I-2", + "finding": "REC section 2 and PROOF_INDEX l.27 say \"No lower bound is asserted for d >= 3\" (scoped to the reconciled D1 chain) although the every-d lower bound DL was ACCEPTed by REV-DL and integrated (Math-#129); STATUS.md D1 carries the note, PROOF_INDEX l.27 does not.", + "proposal": "Add the DL pointer to the PROOF_INDEX D1 row (section 8).", + "sources": [ + "REC", + "PROOF-INDEX", + "DL", + "REV-DL-FILE", + "MAIN-STATUS" + ] + }, + { + "id": "I-3", + "finding": "Math-#543 review 5707632836 and its pickup record CounterexampleE4.lean (blob 3b7712e4) as 1,720 bytes; the blob is 1,743 bytes (SHA-256 812ba30d... as recorded in CE4-README). 1,720 is the character count.", + "proposal": "Note only; the SHA-256 and blob bind the source correctly.", + "sources": [ + "CE4-LEAN", + "CE4-README" + ] + }, + { + "id": "I-4", + "finding": "frontiers/cap_first_exit_lean_20261002 (D1) has no PROOF_INDEX row and its MANIFEST records alignment_review NOT_CLAIMED, so the only kernel-checked link of the chain is invisible to a cold-start reader of PROOF_INDEX.", + "proposal": "Add a row with status \"Proved (kernel); alignment review open\" (section 8).", + "sources": [ + "PROOF-INDEX", + "CFE-MANIFEST", + "REACH" + ] + } + ], + "metrics": { + "claims": 23, + "status_histogram": { + "Conditional": 1, + "Conjectural": 4, + "Proved": 12, + "Proved under assumptions": 3, + "Refuted": 3 + }, + "open_obligations_listed": 35, + "claims_whose_only_nonauthor_reader_is_session_017": [ + "D1", + "L2", + "L2b", + "L3" + ], + "kernel_checked_claims": [ + "D1", + "F1", + "F3" + ], + "kernel_checked_with_nonauthor_alignment_review": [ + "F3" + ], + "human_or_external_reviews": 0, + "distinct_ai_reviewer_providers_on_the_probabilistic_chain": [ + "xAI/Grok", + "Anthropic (session_017)" + ], + "counterexamples_recorded": [ + { + "what": "p = 1 pair-law inhabitant refutes the interface-level implication (F1)", + "source": "CE4-LEAN", + "resolved": "open: successor interface (F2)" + }, + { + "what": "p = 0 inhabitant refutes the universal closure of ParentUniformSelectionA (F2)", + "source": "REG-SCOPE", + "resolved": "open: successor interface (F2)" + }, + { + "what": "Theorem U refutes candidate C7 (unrestricted O(ell^{2/3}) rate)", + "source": "U", + "resolved": "resolved: C7 marked Refuted; Theorems B/C unchanged" + }, + { + "what": "Slice C falsifier C-01 of the LOCAL_PAIRING v1.1 section-5 margin sketch", + "source": "LPAIR", + "resolved": "resolved: sketch withdrawn in v1.2, Lemma 5.1 consumed instead" + } + ], + "claims_downgraded_or_withdrawn": [ + { + "what": "Candidate C7", + "to": "Refuted", + "source": "U" + }, + { + "what": "LOCAL_PAIRING v1.1 section-5 margin sketch", + "to": "withdrawn as a proof", + "source": "LPAIR" + }, + { + "what": "C7-total v1 memo CL-C7-TOTAL-K-SCALING-RECON", + "to": "retracted; superseded by v1.1", + "source": "C7T" + } + ], + "failed_verifications_that_led_to_corrections": [ + { + "what": "congruence factor diag(sqrt r, I) corrected to diag(r^{-1/2}, I) (E1)", + "found_by": "REV-G4 / REV-C1 (R-A3-congruence)", + "source": "E1-ERRATUM" + }, + { + "what": "section 9 Kac-Rice interface replaced (E2 v1.1) after the Borel-event objection", + "found_by": "REV-G2, REV-C1", + "source": "E2-REPAIR" + }, + { + "what": "wording W1 and the embedding radius r < L/(4 sqrt 2)", + "found_by": "REV-C1 / REC", + "source": "REC" + }, + { + "what": "Theorem U v1 -> v3 amendments", + "found_by": "REV-U", + "source": "U" + }, + { + "what": "LOCAL_PAIRING C-01", + "found_by": "OpenAI/Codex slice C review", + "source": "LPAIR" + }, + { + "what": "Math-#543 stale \"pending execution\" wording and unbound review claim", + "found_by": "REV-CE4-ENG", + "source": "CE4-README" + } + ] + } +} diff --git a/reviews/elder_selection_chain_audit_20261031/README.md b/reviews/elder_selection_chain_audit_20261031/README.md new file mode 100644 index 00000000..f8ef7f8e --- /dev/null +++ b/reviews/elder_selection_chain_audit_20261031/README.md @@ -0,0 +1,18 @@ +# Elder-selection chain audit (2026-10-31) + +Source-bound status record for the Theorem A/B/C and lower-bound chain under the owner's five-value taxonomy, +written in answer to the owner's assessment of 31 October 2026 (recorded at main#686 comment 6546811519 and in +`sources/OWNER_ASSESSMENT_20261031.txt`). Scientific effect NONE. Not a register; proposes wording for the integrator. + +- [`AUDIT.md`](AUDIT.md): the record (event per estimate, status per claim, inconsistencies, independence, + metrics baseline, proposed wording). +- [`CLAIMS.json`](CLAIMS.json): the machine-readable dependency list (sources by repository, commit, path, blob, + bytes, SHA-256; reviews; `depends_on`; obligations; metrics). +- `check_claims.py`: identity and dependency check. Run from this directory: + `python3 -B -S check_claims.py` and `python3 -B -O -S check_claims.py` (byte-identical output); + `python3 -B -m unittest test_check_claims -v` and the same with `-O`. + Add `--repo main=/path/to/main` to also verify the two `main` sources; the two Math-#543 sources verify when + that head (`9e89980b`) is present in the object store. + +Author: Anthropic Claude, Claude Code session `session_017Mi3hxjaxV45x6zo6o1ee3`, delegated AI work for Dylan Roy. +Same GitHub account as every lane: organizational-independence credit 0. diff --git a/reviews/elder_selection_chain_audit_20261031/check_claims.py b/reviews/elder_selection_chain_audit_20261031/check_claims.py new file mode 100644 index 00000000..a693af30 --- /dev/null +++ b/reviews/elder_selection_chain_audit_20261031/check_claims.py @@ -0,0 +1,137 @@ +#!/usr/bin/env python3 +"""Identity and dependency check for CLAIMS.json (reviews/elder_selection_chain_audit_20261031). + +What it checks, and nothing more: + * every source entry resolves in the local Git object store to a blob of exactly the recorded + byte length and SHA-256 (the recorded commit is also checked to map the path to that blob + when the commit is available locally); + * every claim uses a status from the owner's five-value taxonomy, cites at least one source, + and every `depends_on`, `sources` and `reviews` reference resolves; + * every review names its performer and verdict. +Sources in another repository are checked only when that repository is given with +`--repo NAME=DIR`; otherwise they are reported as SKIPPED and do not fail the run. +This is a custody check. It certifies no mathematics and no Layer A property. +Exit 0 on success, 1 on any mismatch, 2 on a usage error. +""" +import argparse +import hashlib +import json +import os +import subprocess +import sys + +TAXONOMY = ('Proved', 'Proved under assumptions', 'Conditional', 'Conjectural', 'Refuted') +HERE = os.path.dirname(os.path.abspath(__file__)) + + +def git(repo_dir, *args): + return subprocess.run(['git', '-C', repo_dir, *args], capture_output=True, check=False) + + +def check_source(key, src, repos, out): + repo_dir = repos.get(src['repo']) + if repo_dir is None: + out.append('SKIPPED %s: repository %s not given' % (key, src['repo'])) + return True + blob = git(repo_dir, 'cat-file', 'blob', src['blob']) + if blob.returncode != 0: + out.append('FAIL %s: blob %s not found in %s' % (key, src['blob'], src['repo'])) + return False + data = blob.stdout + ok = True + if len(data) != src['bytes']: + out.append('FAIL %s: %d bytes, recorded %d' % (key, len(data), src['bytes'])) + ok = False + digest = hashlib.sha256(data).hexdigest() + if digest != src['sha256']: + out.append('FAIL %s: sha256 %s, recorded %s' % (key, digest, src['sha256'])) + ok = False + at = git(repo_dir, 'rev-parse', '--verify', '--quiet', '%s:%s' % (src['commit'], src['path'])) + if at.returncode == 0: + if at.stdout.decode().strip() != src['blob']: + out.append('FAIL %s: %s:%s is blob %s, recorded %s' + % (key, src['commit'][:12], src['path'], at.stdout.decode().strip(), src['blob'])) + ok = False + else: + out.append('NOTE %s: commit %s not available locally; blob checked by identity only' + % (key, src['commit'][:12])) + if ok: + out.append('OK %s %s %d %s' % (key, src['blob'][:12], src['bytes'], src['path'])) + return ok + + +def check_structure(doc, out): + ok = True + sources, reviews, claims = doc['sources'], doc['reviews'], doc['claims'] + ids = [c['id'] for c in claims] + if len(set(ids)) != len(ids): + out.append('FAIL duplicate claim ids') + ok = False + for key, rev in reviews.items(): + for field in ('performer', 'verdict', 'record'): + if not rev.get(field): + out.append('FAIL review %s lacks %s' % (key, field)) + ok = False + for c in claims: + if c['status'] not in TAXONOMY: + out.append('FAIL %s: status %r not in taxonomy' % (c['id'], c['status'])) + ok = False + if not c.get('sources'): + out.append('FAIL %s: no source' % c['id']) + ok = False + for s in c.get('sources', ()): + if s not in sources: + out.append('FAIL %s: unknown source %s' % (c['id'], s)) + ok = False + for r in c.get('reviews', ()): + if r not in reviews: + out.append('FAIL %s: unknown review %s' % (c['id'], r)) + ok = False + for d in c.get('depends_on', ()): + if d not in ids: + out.append('FAIL %s: depends_on unknown claim %s' % (c['id'], d)) + ok = False + if c['status'] == 'Proved' and not (c.get('reviews') or c.get('kernel_checked')): + out.append('FAIL %s: Proved without a nonauthor review or kernel check' % c['id']) + ok = False + if c['status'] == 'Conditional' and not c.get('depends_on'): + out.append('FAIL %s: Conditional without depends_on' % c['id']) + ok = False + return ok + + +def run(claims_path, repos): + with open(claims_path, encoding='utf-8') as fh: + doc = json.load(fh) + out = [] + ok = check_structure(doc, out) + for key in sorted(doc['sources']): + ok = check_source(key, doc['sources'][key], repos, out) and ok + out.append('RESULT %s: %d claims, %d sources, %d reviews' + % ('PASS' if ok else 'FAIL', len(doc['claims']), len(doc['sources']), len(doc['reviews']))) + return ok, out + + +def main(argv=None): + ap = argparse.ArgumentParser(description=__doc__.split('\n')[0]) + ap.add_argument('--claims', default=os.path.join(HERE, 'CLAIMS.json')) + ap.add_argument('--repo', action='append', default=[], metavar='NAME=DIR', + help='repository checkout for sources of that repo (default: Math-=this checkout)') + args = ap.parse_args(argv) + repos = {'Math-': os.path.abspath(os.path.join(HERE, '..', '..'))} + for item in args.repo: + name, _, path = item.partition('=') + if not name or not path: + ap.error('--repo expects NAME=DIR') + repos[name] = path + try: + ok, out = run(args.claims, repos) + except (OSError, ValueError, KeyError) as exc: + print('check_claims error: %s' % exc, file=sys.stderr) + return 2 + print('\n'.join(out)) + return 0 if ok else 1 + + +if __name__ == '__main__': + sys.exit(main()) diff --git a/reviews/elder_selection_chain_audit_20261031/sources/OWNER_ASSESSMENT_20261031.txt b/reviews/elder_selection_chain_audit_20261031/sources/OWNER_ASSESSMENT_20261031.txt new file mode 100644 index 00000000..dd062d47 --- /dev/null +++ b/reviews/elder_selection_chain_audit_20261031/sources/OWNER_ASSESSMENT_20261031.txt @@ -0,0 +1,144 @@ +The project needs to make one major transition: from a system that is increasingly good at checking its own work to one that can demonstrate, independently and reproducibly, that its central mathematical claims are true. + +Based on the issues identified in the previous investigation, I would prioritize seven changes. + +1. Close the gap between the formal proof and the actual theorem + +Highest priority + +The most important weakness is the gap between abstract estimates and the actual Gaussian-field persistence problem. + +For example, the formal countermodel we discussed showed that an upper-selection estimate does not establish a positive lower rate for genuine elder-rule pairings. + +What needs to change: + +* Define the actual measurable event that says a candidate pair is an elder-rule pair. +* Prove that the event satisfies the required lower and upper bounds. +* Connect every abstract probability or counting estimate back to that event. +* Make explicit which assumptions are needed at each step. + +Why: A mathematically valid estimate about a proxy event does not automatically prove a statement about the actual persistence pairing. This is the central logical gap to close. + +2. Resolve the extra-critical-point problem + +Highest priority + +The investigation identified a case where an additional critical point can prevent a proposed maximum-saddle pair from being the actual elder-rule pair. + +What needs to change: + +* Replace raw critical-point counting with a topological witness. +* Establish conditions under which a candidate pair really survives the elder-selection process. +* Bound the probability of configurations that violate those conditions. +* Show that the good configurations contribute at the claimed leading order. + +Why: Counting candidate maximum-saddle pairs is not the same as counting actual persistence bars. Until this distinction is controlled, the leading asymptotic argument remains vulnerable. + +3. Finish the unrestricted parameter estimates + +The project has made progress on compact parameter ranges, but those estimates do not automatically extend to all relevant birth heights and gap parameters. + +What needs to change: + +* Complete the small-gap estimates. +* Prove the large-birth and large-gap tail bounds. +* Control the intermediate separation scales. +* Handle collisions and boundaries of the rank charts. +* State precisely which constants depend on dimension, covariance, and domain size. + +Why: The asymptotic coefficient requires controlling the entire parameter space, not merely the region where the local calculation is easiest. + +4. Separate proven results from research candidates + +Some developments, such as the proposed next-order expansion and rejected-density estimate, are interesting but are not yet established results. + +What needs to change: + +Give every mathematical claim an explicit status: + +Status Meaning +Proved Complete argument with checked dependencies +Proved under assumptions Valid only under clearly listed hypotheses +Conditional Depends on an unresolved lemma or theorem +Conjectural Plausible, but not established +Refuted A counterexample or contradiction has been established + +Keep conjectural coefficients and numerical estimates out of the headline theorem until their derivations are complete. + +Why: This prevents a plausible research direction from gradually becoming treated as a proven fact through repeated documentation and review. + +5. Introduce genuinely independent verification + +The existing automated reviews and audits are useful, but repeated review by related models or the same provider does not establish independent mathematical validation. + +What needs to change: + +* Obtain a review from a specialist in Gaussian random fields. +* Obtain a separate review from someone experienced in persistent homology and stochastic topology. +* Give reviewers the theorem and proof without the project’s preferred interpretation. +* Ask them to identify the weakest indispensable lemma and construct counterexamples. +* Publish the responses and unresolved objections. + +Why: A system can become very good at detecting errors that its own review process knows how to look for while missing errors outside that framework. + +6. Make the software prove its own coverage, not just pass its tests + +The recent engineering work on module ownership, mutation tests, and failure-reason checks is useful progress. + +But passing tests does not prove that every important mathematical claim is represented or that every dependency is checked. + +What needs to change: + +* Require every theorem to have a machine-readable dependency list. +* Link each dependency to its exact source, proof, and assumptions. +* Check that no theorem silently relies on an unverified declaration or imported result. +* Make incomplete proof obligations visible in the main status report. +* Preserve failed test runs and record why they failed. + +Why: This makes it harder for a polished software pipeline to create a false impression of mathematical completeness. + +7. Make self-improvement measurable + +The project appears to be improving its organization and audit procedures. The next step is to measure whether those improvements actually increase mathematical reliability. + +Track things such as: + +* Number of unresolved proof obligations. +* Number of counterexamples discovered and resolved. +* Number of claims downgraded or withdrawn. +* Number of independent reviews completed. +* Number of formalized results that match the intended mathematical statement. +* Number of failed verification attempts that led to substantive corrections. + +Why: More commits, more tests, more documentation, and more model-generated reviews are not necessarily more mathematical progress. The important metric is how many genuine uncertainties have been eliminated. + +The order I would follow + +Recommended priority sequence + +* Prove the actual elder-rule selection statement. + Connect candidate pairs to genuine persistence pairs. +* Resolve the extra-critical-point obstruction. + Establish a valid topological witness and control its failures. +* Complete unrestricted estimates. + Close all parameter tails, intermediate scales, and boundary cases. +* Obtain independent mathematical review. + Test the proof outside the project’s own review ecosystem. +* Formalize and audit the complete dependency chain. + Ensure the formal result really matches the stated theorem. +* Only then promote stronger claims. + Revisit next-order asymptotics and rejected-density estimates. + +What success would look like + +The project should eventually be able to answer five questions with concrete evidence: + +* Does the claimed exponent hold for the actual persistence bars, rather than a proxy count? +* Is the leading coefficient derived from a complete and justified limiting argument? +* Are all parameter ranges and error terms controlled? +* Can an independent specialist reproduce and challenge the argument? +* Does the formal verification cover the exact theorem being advertised? + +My overall assessment: The project does not primarily need more automation, more agents, or more elaborate review infrastructure. It needs to close the mathematical gap between local candidate-pair calculations and actual global persistence pairings. + +That is the change most likely to turn an increasingly sophisticated self-auditing research system into a genuinely reliable mathematical research system. \ No newline at end of file diff --git a/reviews/elder_selection_chain_audit_20261031/test_check_claims.py b/reviews/elder_selection_chain_audit_20261031/test_check_claims.py new file mode 100644 index 00000000..daf2db67 --- /dev/null +++ b/reviews/elder_selection_chain_audit_20261031/test_check_claims.py @@ -0,0 +1,96 @@ +import copy +import json +import os +import subprocess +import sys +import tempfile +import unittest + +HERE = os.path.dirname(os.path.abspath(__file__)) +sys.path.insert(0, HERE) +import check_claims # noqa: E402 + +REPOS = {'Math-': os.path.abspath(os.path.join(HERE, '..', '..'))} + + +def load(): + with open(os.path.join(HERE, 'CLAIMS.json'), encoding='utf-8') as fh: + return json.load(fh) + + +def run_doc(doc): + with tempfile.TemporaryDirectory() as tmp: + path = os.path.join(tmp, 'CLAIMS.json') + with open(path, 'w', encoding='utf-8') as fh: + json.dump(doc, fh) + return check_claims.run(path, REPOS) + + +class CheckClaimsTests(unittest.TestCase): + def test_committed_register_passes(self): + ok, out = run_doc(load()) + self.assertTrue(ok, '\n'.join(out)) + self.assertTrue(out[-1].startswith('RESULT PASS')) + self.assertFalse(any(line.startswith('FAIL') for line in out)) + + def test_tampered_sha256_fails(self): + doc = load() + key = sorted(k for k, s in doc['sources'].items() if s['repo'] == 'Math-')[0] + doc['sources'][key]['sha256'] = '0' * 64 + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertTrue(any(line.startswith('FAIL %s: sha256' % key) for line in out), '\n'.join(out)) + + def test_wrong_byte_count_fails(self): + doc = load() + key = sorted(k for k, s in doc['sources'].items() if s['repo'] == 'Math-')[0] + doc['sources'][key]['bytes'] += 1 + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertTrue(any(line.startswith('FAIL %s:' % key) and 'bytes' in line for line in out)) + + def test_status_outside_taxonomy_fails(self): + doc = load() + doc['claims'][0]['status'] = 'Accepted' + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertTrue(any('not in taxonomy' in line for line in out)) + + def test_dangling_dependency_fails(self): + doc = load() + doc['claims'][0].setdefault('depends_on', []).append('NO-SUCH-CLAIM') + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertTrue(any('depends_on unknown claim NO-SUCH-CLAIM' in line for line in out)) + + def test_proved_without_review_fails(self): + doc = load() + target = next(c for c in doc['claims'] if c['status'] == 'Proved') + target['reviews'] = [] + target.pop('kernel_checked', None) + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertTrue(any('Proved without a nonauthor review' in line for line in out)) + + def test_other_repository_sources_are_skipped_not_failed(self): + doc = load() + doc['sources'] = {k: v for k, v in doc['sources'].items() if v['repo'] != 'Math-'} + for c in doc['claims']: + c['sources'] = [s for s in c['sources'] if s in doc['sources']] or [next(iter(doc['sources']))] + ok, out = run_doc(doc) + self.assertTrue(ok, '\n'.join(out)) + self.assertTrue(all(line.startswith(('SKIPPED', 'RESULT')) for line in out), '\n'.join(out)) + + def test_cli_modes_agree(self): + env = dict(os.environ, PYTHONDONTWRITEBYTECODE='1') + a = subprocess.run([sys.executable, '-B', '-S', os.path.join(HERE, 'check_claims.py')], + capture_output=True, env=env) + b = subprocess.run([sys.executable, '-B', '-O', '-S', os.path.join(HERE, 'check_claims.py')], + capture_output=True, env=env) + self.assertEqual(a.returncode, 0, a.stdout.decode() + a.stderr.decode()) + self.assertEqual(a.stdout, b.stdout) + self.assertEqual(a.returncode, b.returncode) + + +if __name__ == '__main__': + unittest.main() From e87645a411cb8c9899fbd0fbde8058b8701c7474 Mon Sep 17 00:00:00 2001 From: Claude Date: Sat, 31 Oct 2026 22:31:50 +0000 Subject: [PATCH 2/3] Apply the nonauthor reads to the elder-selection audit (v1.1) S1 (Grok Bot agent 4, Math-#554 comment 6548325142; 01SHCrxv, main#684 comment 6548184976): a Proved claim may depend only on Proved claims. Theorem B is stated in the parent through the elder relation, so it depends on A and E0 only and stays Proved. Theorems C and U are stated for finite bars and consume the Morse-locus identification F5, which the record holds at Proved under assumptions, so both become Conditional on F5; the refutation of C7 records the same dependency. check_claims.py now enforces the propagation rule, with a test. F1 (01XJMJ, review 5724603343): nine tests added so that every check of check_claims.py has a test that fails without it. The taxonomy test now mutates a leaf claim so that the propagation rule cannot mask it. With one more test for the propagation rule, 17 single-check mutants of the checker are each killed by an assertion in normal and -O mode. F3 (01XJMJ, AMEND): the I-3 attribution corrected. The 1,720-byte figure is in GPT-6's pickup 6513952657 and in review 5724470777; review 5707632836 states no byte count. F4 (01XJMJ): the I-2 wording corrected. PROOF_INDEX lists the every-d lift in its PR inventory (l.107, 110, 111); only the D1 row (l.27) lacks the pointer. No source identity, review record or claim statement changes otherwise. Status histogram: Proved 10, Proved under assumptions 3, Conditional 3, Conjectural 4, Refuted 3. Scientific effect NONE. Co-Authored-By: Claude Fable 5.1 Claude-Session: https://claude.ai/code/session_017Mi3hxjaxV45x6zo6o1ee3 --- .../AUDIT.md | 44 ++++++--- .../CLAIMS.json | 53 +++++----- .../check_claims.py | 7 ++ .../test_check_claims.py | 97 ++++++++++++++++++- 4 files changed, 163 insertions(+), 38 deletions(-) diff --git a/reviews/elder_selection_chain_audit_20261031/AUDIT.md b/reviews/elder_selection_chain_audit_20261031/AUDIT.md index 0d456edf..c32cb35d 100644 --- a/reviews/elder_selection_chain_audit_20261031/AUDIT.md +++ b/reviews/elder_selection_chain_audit_20261031/AUDIT.md @@ -33,7 +33,9 @@ The chain already contains the two things the owner's items 1 and 2 ask for, and above `f(S)`. No critical point elsewhere can change the pairing once these hold. - **The probabilistic links are prose, read by two AI reviewers, one of them this session.** `Q^W(G_r^c) ≤ C r³` (A-II), the normalizer floor (A-I), the Kac–Rice bridge (B, C) and both lower bounds (L2, L3) are complete written proofs - whose nonauthor reads are by xAI/Grok and by this session only. For the lower bounds, the cap second pass and the + whose nonauthor reads are by xAI/Grok and by this session only. Theorem B is stated for elder-selected pairs and is + Proved; Theorems C and U are stated for bars and are Conditional on exactly one further input, the classical + identification of elder death with bar death (F5), which the project holds as an interface clause (v1.1, S1). For the lower bounds, the cap second pass and the reconciliation, this session is the sole nonauthor reader. There is no human review and no review outside those two providers anywhere in the chain. - **No kernel-checked statement of any probabilistic theorem about the actual event exists.** The statement register @@ -42,7 +44,7 @@ The chain already contains the two things the owner's items 1 and 2 ask for, and is Refuted (F1), and the register's shapes do not pin down the theorems (F2). Closing this needs the regression law as a measure in Lean, which the register itself classifies as Mathlib-blocked. -Status histogram over the 23 claims: Conditional 1, Conjectural 4, Proved 12, Proved under assumptions 3, Refuted 3. Open obligations listed: 35. Claims whose only nonauthor +Status histogram over the 23 claims: Conditional 3, Conjectural 4, Proved 10, Proved under assumptions 3, Refuted 3. Open obligations listed: 37. Claims whose only nonauthor reader is this session: D1, L2, L2b, L3. | ID | Claim | Event | Status | Evidence | Nonauthor readers | Open | @@ -61,9 +63,9 @@ reader is this session: D1, L2, L2b, L3. | LP-EQ | Identified selection-failure law r^{-3}(1 - p_r) -> alpha_1 + alpha_2 (Theorem S planar, Theorem F every d; Conjecture LP=) | elder | **Conjectural** | prose | none located | 2 | | W | Window critical-point counts: E_{Q^W} N_{r,j}(torus minus pins, between-pin height window) <= C r^3 | candidate-count | **Proved** | prose | Anthropic session_017, xAI/Grok | 2 | | F5 | Identification of the maximin level with the H0 bar death and of elder-selected pairs with finite bars | deterministic | **Proved under assumptions** | prose | Anthropic session_017, xAI/Grok | 1 | -| B | Theorem B (P 1.2): compact-window densities and the O(ell^{2/3}) difference | bars-vs-candidates | **Proved** | prose | Anthropic session_017, xAI/Grok | 2 | -| C | Theorem C (P 1.3): unrestricted leading density nu_cand^all ~ nu_eld^all ~ c_{d,L} ell^{-1/3} | bars-vs-candidates | **Proved** | prose | Anthropic session_017, xAI/Grok | 2 | -| U | Theorem U: the unrestricted selection difference does not vanish | bars-vs-candidates | **Proved** | prose | OpenAI | 1 | +| B | Theorem B (P 1.2): compact-window densities and the O(ell^{2/3}) difference | elder-vs-candidates | **Proved** | prose | Anthropic session_017, xAI/Grok | 2 | +| C | Theorem C (P 1.3): unrestricted leading density nu_cand^all ~ nu_eld^all ~ c_{d,L} ell^{-1/3} | bars-vs-candidates | **Conditional** | prose | Anthropic session_017, xAI/Grok | 3 | +| U | Theorem U: the unrestricted selection difference does not vanish | bars-vs-candidates | **Conditional** | prose | OpenAI | 2 | | C7 | Candidate C7: unrestricted difference rate 0 <= nu_cand^all - nu_eld^all <= C ell^{2/3} | bars-vs-candidates | **Refuted** | prose | OpenAI | 0 | | RHO | Rejected density rho_rej(ell) = nu_cand^all - nu_eld^all = O(1) (C7-total Theorem K; ledger DP-07) | bars-vs-candidates | **Conjectural** | prose | none located | 1 | | F1 | Interface-level implication "upper selection shape implies two-sided rate shape" for every PairLawInterface | formal-interface | **Refuted** | kernel | OpenAI | 1 | @@ -118,7 +120,8 @@ upper nor the lower statement, so nothing formal connects the estimates to the e | D1 cap theorem | deterministic | the inclusion itself, kernel-checked | | L2, L3 lower bounds | an explicit subevent of `elder event^c` (`d_f(M) ≥ b − k r³/4 > f(S)`) | direct: `1 − p_r ≥ Q^W(subevent) ≥ c r³` | | W window counts | critical points in the height window | not used by Theorem A; explain the failure mechanism; not elder statements | -| B, C densities | bars (`ν_eld`) against candidates (`ν_cand`) | `ν_eld` integrates `p_r` against the candidate intensity (E2 Kac–Rice, P sections 9–11, 13); bars = elder-selected pairs by F5 | +| B compact-window densities | elder-selected candidate pairs (`ν_eld`, defined through E0) against candidates (`ν_cand`) | `ν_eld` integrates `p_r` against the candidate intensity (E2 Kac–Rice, P sections 9–11); no bar identification is used | +| C unrestricted densities | finite bars (`ν_eld^all`) against candidates | the same Kac–Rice route (P section 13) plus the Morse-locus identification bars = elder-selected pairs (F5, P l.445) | | U, C7, RHO | the difference `ν_cand^all − ν_eld^all` | far pairs separated by a level curve are candidates and not bars | **Assumptions per step** are the `obligations` lists in `CLAIMS.json`. The two that carry the chain are: the full @@ -172,6 +175,14 @@ Theorem F) and the rejected density RHO. The owner's item 4 names "the proposed rejected-density estimate" as not established; this record agrees on the second and did not locate a next-order expansion of `ν_eld` in the chain (none is in the headline statements). +**Status propagation (v1.1, finding S1 of the Grok Bot a4 and 01SHCrxv reads).** A Proved claim may depend only on +Proved claims, and `check_claims.py` now enforces it. Theorem B's `ν_eld` is defined in P through the elder relation +(E0), so B depends on A and E0 only and stays Proved. Theorems C and U are stated for finite bars; their only +non-Proved dependency is F5, the Morse-locus identification of elder-selected pairs with bars, which this record holds +at Proved under assumptions because the project consumes it as a classical fact and, in Lean, as an interface clause. +So C and U are Conditional on F5 alone (v1.0 had Proved); discharging F5 as a theorem makes both Proved with no +other change. The refutation of C7 inherits the same dependency through U and is recorded as such. + Numerical coefficients: Theorem C states `c_{d,L}` as an identified Gaussian/angular integral (15.2) with no value. The SIDE24 numerical coefficient packets are outside this audit. No numerical value of `C`, `r_*`, `z_*`, `c_{B,K}`, `c_{d,L}`, `α₁ + α₂` or `c_*` is established anywhere in this chain, and none belongs in a headline statement. @@ -181,8 +192,8 @@ SIDE24 numerical coefficient packets are outside this audit. No numerical value | ID | Finding | Proposal | |---|---|---| | I-1 | main `docs/DISCOVERY_PRIORITY_LEDGER_20260925.md` DP-02 (l.18, l.157) says parent sections 2–7 are **AMEND**; main `STATUS.md` D1 (l.88) and Math- `PROOF_INDEX.md` (l.27) record ACCEPT at existential scope from the 28 September reconciliation, whose residual `R-A3-congruence` is DISCHARGED by E1 (REC `LEDGER.json`). The ledger cell was aligned with `STATUS.md` of 26 September and repeated unchanged by the 28 October refresh ([main#669 6465885945](https://github.com/d6g8k5htny-coder/main/issues/669#issuecomment-6465885945)). | Root decides which surface governs (B-003). This record reads the ledger cell as stale. | -| I-2 | REC section 2 and PROOF_INDEX l.27: "No lower bound is asserted for d ≥ 3" (true of the reconciled D1 chain) while the every-`d` lift L3 was ACCEPTed and integrated (Math-#129); `STATUS.md` D1 notes it, PROOF_INDEX does not. | Add the pointer to the PROOF_INDEX D1 row. | -| I-3 | Math-#543 review 5707632836 records `CounterexampleE4.lean` as 1,720 bytes; blob `3b7712e4` is 1,743 bytes (1,720 is the character count). The SHA-256 in the README is correct. | Note only. | +| I-2 | REC section 2 and the PROOF_INDEX D1 row (l.27): "No lower bound is asserted for d ≥ 3" (true of the reconciled D1 chain) while the every-`d` lift L3 was ACCEPTed and integrated (Math-#129). PROOF_INDEX lists the lift in its PR inventory (l.107, 110, 111) and `STATUS.md` D1 notes it; the D1 row of PROOF_INDEX does not point to it. | Add the pointer to the PROOF_INDEX D1 row. | +| I-3 | GPT-6's pickup 6513952657 and review 5724470777 (01XJMJ) on Math-#543 record `CounterexampleE4.lean` (blob `3b7712e4`) as 1,720 bytes; the blob is 1,743 bytes (1,720 is the character count). Review 5707632836 states no byte count. The SHA-256 in the README is correct. | Note only. 01XJMJ posted its own correction at [Math-#543 6548199529](https://github.com/d6g8k5htny-coder/Math-/pull/543#issuecomment-6548199529). | | I-4 | `frontiers/cap_first_exit_lean_20261002` has no PROOF_INDEX row and `MANIFEST.json` records `alignment_review: NOT_CLAIMED`; the only kernel-checked link of the chain is invisible to a cold-start reader of PROOF_INDEX. | Add a row: "Proved (kernel); alignment review open". | ## 5. Independence (owner item 5) @@ -226,8 +237,8 @@ Computed from this record at the cuts above; the values are in `CLAIMS.json` und | Metric | Value at this cut | |---|---| -| Claims audited | 23 (Conditional 1, Conjectural 4, Proved 12, Proved under assumptions 3, Refuted 3) | -| Open obligations listed | 35 | +| Claims audited | 23 (Conditional 3, Conjectural 4, Proved 10, Proved under assumptions 3, Refuted 3) | +| Open obligations listed | 37 | | Counterexamples recorded | 4: `p ≡ 1` (F1) and `p ≡ 0` (F2) at the interface, open pending a successor interface; Theorem U against C7, resolved; LOCAL_PAIRING C-01, resolved by withdrawing the sketch | | Claims downgraded or withdrawn | 3: C7 → Refuted; the LOCAL_PAIRING v1.1 section-5 sketch withdrawn; the C7-total v1 memo retracted | | Independent (human or external) reviews completed | 0 | @@ -247,7 +258,18 @@ Computed from this record at the cuts above; the values are in `CLAIMS.json` und statement of Theorems A, B or C about the elder event exists; `PairLawInterface.p` is primitive." - A PROOF_INDEX row for `frontiers/cap_first_exit_lean_20261002` (I-4). -## 9. Non-claims +## 9. Revision history + +- **v1.0, 2026-10-31, head `1d8267ea`.** First record. Nonauthor reads: 01XJMJ (Anthropic; sources and inconsistencies, + [review 5724603343](https://github.com/d6g8k5htny-coder/Math-/pull/554#pullrequestreview-5724603343), PASS_SCOPED with + F1, F3, F4), Grok Bot agent 4 (xAI; status propagation, [6548325142](https://github.com/d6g8k5htny-coder/Math-/pull/554#issuecomment-6548325142), + AMEND S1), 01SHCrxv (Anthropic; custody 44/44 and S1, [main#684 6548184976](https://github.com/d6g8k5htny-coder/main/issues/684#issuecomment-6548184976)). +- **v1.1.** S1: C and U Conditional on F5; B's dependency on F5 removed because P states Theorem B through the elder + relation; a status-propagation rule added to `check_claims.py` with a test. F1: 01XJMJ's nine checker tests added. + F3: the I-3 attribution corrected. F4: the I-2 wording corrected. No source identity, review record, claim statement + or inconsistency other than I-2 and I-3 changed. + +## 10. Non-claims No theorem is proved, improved, weakened or withdrawn here. No review verdict is issued on any proof; the statuses summarize located records under a stated rule. No numerical constant is claimed. Nothing here is independent review diff --git a/reviews/elder_selection_chain_audit_20261031/CLAIMS.json b/reviews/elder_selection_chain_audit_20261031/CLAIMS.json index 12d41e9c..ad82241e 100644 --- a/reviews/elder_selection_chain_audit_20261031/CLAIMS.json +++ b/reviews/elder_selection_chain_audit_20261031/CLAIMS.json @@ -1,6 +1,6 @@ { "schema": "elder-selection-chain-audit/v1", - "object": "CL-ELDER-SELECTION-CHAIN-AUDIT-20261031-v1", + "object": "CL-ELDER-SELECTION-CHAIN-AUDIT-20261031-v1.1", "author": "Anthropic Claude, Claude Code session_017Mi3hxjaxV45x6zo6o1ee3, delegated AI work for Dylan Roy", "scientific_effect": "NONE", "kind": "source-bound review record; proposes PROOF_INDEX/STATUS wording for the integrator; not a register", @@ -23,13 +23,14 @@ "Conjectural": "no complete nonauthor-read proof: author-side complete proofs with no located nonauthor read (candidates) and open conjectures", "Refuted": "a counterexample or contradiction is recorded" }, - "assignment_rule": "Status follows the located evidence, not the author's label. A nonauthor AI read earns Proved; the evidence field says who read it. No claim in this chain has verification independent of the project's AI reviewers, and none has human review.", + "assignment_rule": "Status follows the located evidence, not the author's label. A nonauthor AI read earns Proved; the evidence field says who read it. A Proved claim may depend only on Proved claims; check_claims.py enforces this (v1.1). No claim in this chain has verification independent of the project's AI reviewers, and none has human review.", "event_vocabulary": { "elder": "elder event {d_f(M) = f(S)} on the Morse-distinct locus (P section 8); p_r is its Q^W-probability", "good-jet": "the event G_r of P (7.1)/l.216: the CAP (1) hypotheses hold on the pinned cylinder", "deterministic": "a statement about a fixed function, no probability", "candidate-count": "counts of critical points or candidate pairs, not elder-selected", - "bars-vs-candidates": "finite H0 bars (nu_eld) against candidate pairs (nu_cand)", + "elder-vs-candidates": "elder-selected candidate pairs (Theorem B's nu_eld, defined through E0) against all candidate pairs (nu_cand)", + "bars-vs-candidates": "finite H0 bars (nu_eld^all) against candidate pairs (nu_cand^all); the bar identification is F5", "normalizer": "the full normalizer Z_r", "formal-interface": "a statement about the Lean interface objects of specifications/lean_statements_20261010" }, @@ -745,7 +746,7 @@ ], "obligations": [ "Sole nonauthor reviewer is session_017.", - "Surface wording I-2: REC section 2 and the PROOF_INDEX D1 row say \"no lower bound is asserted for d >= 3\" (scoped to the D1 chain) while this reviewed lift exists." + "Surface wording I-2: REC section 2 and the PROOF_INDEX D1 row (l.27) say \"no lower bound is asserted for d >= 3\" (scoped to the D1 chain) and do not point to this reviewed lift, which PROOF_INDEX lists only in its PR inventory (l.107, 110, 111)." ] }, { @@ -889,11 +890,11 @@ { "id": "B", "title": "Theorem B (P 1.2): compact-window densities and the O(ell^{2/3}) difference", - "event": "bars-vs-candidates", - "statement": "For B, K of positive length: nu_cand ~ nu_eld ~ c_{B,K} ell^{-1/3} and 0 <= nu_cand - nu_eld <= C_{B,K} ell^{2/3}, where nu_eld counts candidate pairs that are elder partners (E0, F5). Proof: Theorem A plus the marked Kac-Rice interface (E2) and the radial ledger and pushforward (P sections 9-11).", + "event": "elder-vs-candidates", + "statement": "For B, K of positive length: nu_cand ~ nu_eld ~ c_{B,K} ell^{-1/3} and 0 <= nu_cand - nu_eld <= C_{B,K} ell^{2/3}, where nu_eld counts the candidate pairs that are actual ordinary elder partners (P l.37: \"let nu_eld(ell) count only pairs which are actual ordinary elder partners\"), that is the elder event E0, not bars. Proof: Theorem A plus the marked Kac-Rice interface (E2) and the radial ledger and pushforward (P sections 9-11).", "status": "Proved", "proof_form": "prose", - "evidence": "REV-G1, REV-C1, REV-C2; REC section 2.", + "evidence": "REV-G1, REV-C1, REV-C2; REC section 2. Theorem B is stated through the elder relation (E0); the bar identification F5 enters Theorem C, not Theorem B (v1.1, after S1 of the Grok Bot a4 and 01SHCrxv reads).", "sources": [ "P", "E2-REPAIR", @@ -906,8 +907,7 @@ ], "depends_on": [ "A", - "E0", - "F5" + "E0" ], "obligations": [ "The Kac-Rice representation and pushforward are prose only.", @@ -918,10 +918,10 @@ "id": "C", "title": "Theorem C (P 1.3): unrestricted leading density nu_cand^all ~ nu_eld^all ~ c_{d,L} ell^{-1/3}", "event": "bars-vs-candidates", - "statement": "All births, gaps and separations; essential class excluded. Near pairs: dominated convergence with the target-growth majorant (13.4) and p_r -> 1 pointwise from Theorem A on compact neighbourhoods (P l.415); far pairs: nu^far <= C (14.1); c_{d,L} is the integral (13.6)/(15.2). Leading term only: no unrestricted remainder or difference rate.", - "status": "Proved", + "statement": "All births, gaps and separations; essential class excluded; nu_eld^all counts all finite ordinary superlevel H0 bars (P l.44). Near pairs: dominated convergence with the target-growth majorant (13.4) and p_r -> 1 pointwise from Theorem A on compact neighbourhoods (P l.415); far pairs: nu^far <= C (14.1); c_{d,L} is the integral (13.6)/(15.2). The selected pair count equals the finite-bar count by the Morse-locus identification (P l.445), which is F5. Leading term only: no unrestricted remainder or difference rate.", + "status": "Conditional", "proof_form": "prose", - "evidence": "REV-G1, REV-C1, REV-C2 (spot re-derivations of (13.1), (13.4), (15.2)); REC section 2.", + "evidence": "The Gaussian analysis is complete and read: REV-G1, REV-C1, REV-C2 (spot re-derivations of (13.1), (13.4), (15.2)); REC section 2 records ACCEPT. Its only non-Proved dependency is F5, the classical bar identification, which this record holds at Proved under assumptions (asserted in P, an interface clause in Lean); under the record's rule that makes Theorem C Conditional on F5 and on nothing else (v1.1, S1 of the Grok Bot a4 and 01SHCrxv reads; v1.0 had Proved).", "sources": [ "P", "REC" @@ -937,6 +937,7 @@ "F5" ], "obligations": [ + "Discharge F5 (the bar identification) as a theorem rather than an interface clause; Theorem C then becomes Proved with no other change.", "No unrestricted difference rate; Theorem U shows none of order ell^{2/3} exists.", "c_{d,L} is an identified integral with no numerical value; numerical SIDE24 coefficient work is outside this audit and belongs outside the headline until derived (owner item 4)." ] @@ -945,10 +946,10 @@ "id": "U", "title": "Theorem U: the unrestricted selection difference does not vanish", "event": "bars-vs-candidates", - "statement": "nu_cand^all(ell) - nu_eld^all(ell) >= nu_cand^far - nu_eld^far >= c_* > 0 for 0 < ell <= ell_0; hence D(ell)/ell^alpha -> infinity for every alpha > 0 and c_* <= D(ell) = o(ell^{-1/3}).", - "status": "Proved", + "statement": "nu_cand^all(ell) - nu_eld^all(ell) >= nu_cand^far - nu_eld^far >= c_* > 0 for 0 < ell <= ell_0, where nu_eld^all counts finite H0 bars; hence D(ell)/ell^alpha -> infinity for every alpha > 0 and c_* <= D(ell) = o(ell^{-1/3}) (the upper half of (U2) uses Theorem C).", + "status": "Conditional", "proof_form": "prose", - "evidence": "REV-U ACCEPT Theorem U and (U2)-(U4) (OpenAI); author session_017.", + "evidence": "REV-U ACCEPT Theorem U and (U2)-(U4) (OpenAI); author session_017. (U1) reads \"cannot be persistence partners\" through the bar identification F5, and (U2) consumes Theorem C, so under the record's rule Theorem U is Conditional on F5 (and on C, itself Conditional on F5 alone) (v1.1, S1; v1.0 had Proved).", "sources": [ "U", "REV-U-FILE" @@ -957,10 +958,12 @@ "REV-U" ], "depends_on": [ - "C" + "C", + "F5" ], "obligations": [ - "Whether D is bounded is not decided by U (U section 5); see RHO." + "Whether D is bounded is not decided by U (U section 5); see RHO.", + "Discharging F5 makes U Proved with no other change." ] }, { @@ -970,7 +973,7 @@ "statement": "The compact-window rate (1.2) extended to the unrestricted densities (CANDIDATES C7).", "status": "Refuted", "proof_form": "prose", - "evidence": "Theorem U (U2): the difference is bounded below by c_* > 0.", + "evidence": "Theorem U (U2): the difference is bounded below by c_* > 0 (REV-U). The contradiction consumes the bar identification F5 exactly as U does; a refutation of C7 for elder-selected pairs in place of bars needs no F5.", "sources": [ "U", "CANDIDATES" @@ -1012,7 +1015,7 @@ "status": "Refuted", "proof_form": "lean", "kernel_checked": true, - "evidence": "Lean-checked at Math- commit 5a470bfd (run 40957238766, job 123740377857; axioms propext, Classical.choice, Quot.sound), relocated byte-identically to cbz/lifetime_pair_law at Math-#543 head 9e89980b; REV-CE4-ENG and REV-CE4-FORMAL PASS at scope; an Anthropic different-family review (01XJMJ, Math-#543 comment 6545265266) is open at this cut. Byte count: blob 3b7712e4 is 1,743 bytes; the \"1,720\" in the formal review counts characters (I-3).", + "evidence": "Lean-checked at Math- commit 5a470bfd (run 40957238766, job 123740377857; axioms propext, Classical.choice, Quot.sound), relocated byte-identically to cbz/lifetime_pair_law at Math-#543 head 9e89980b; REV-CE4-ENG and REV-CE4-FORMAL PASS at scope; an Anthropic different-family review (01XJMJ, Math-#543 comment 6545265266) is open at this cut. Byte count: blob 3b7712e4 is 1,743 bytes; the \"1,720\" of GPT-6's pickup 6513952657 and of 01XJMJ's review 5724470777 counts characters (I-3).", "sources": [ "CE4-LEAN", "CE4-README", @@ -1113,7 +1116,7 @@ }, { "id": "I-2", - "finding": "REC section 2 and PROOF_INDEX l.27 say \"No lower bound is asserted for d >= 3\" (scoped to the reconciled D1 chain) although the every-d lower bound DL was ACCEPTed by REV-DL and integrated (Math-#129); STATUS.md D1 carries the note, PROOF_INDEX l.27 does not.", + "finding": "REC section 2 and the PROOF_INDEX D1 row (l.27) say \"No lower bound is asserted for d >= 3\" (scoped to the reconciled D1 chain) although the every-d lower bound DL was ACCEPTed by REV-DL and integrated (Math-#129). PROOF_INDEX does list the lift in its PR inventory (l.107 the #129 integration record, l.110 the #125 proof, l.111 the #127 review) and STATUS.md D1 carries a note; what is missing is a pointer in the D1 row itself. (v1.1 wording per 01XJMJ review 5724603343 F4.)", "proposal": "Add the DL pointer to the PROOF_INDEX D1 row (section 8).", "sources": [ "REC", @@ -1125,8 +1128,8 @@ }, { "id": "I-3", - "finding": "Math-#543 review 5707632836 and its pickup record CounterexampleE4.lean (blob 3b7712e4) as 1,720 bytes; the blob is 1,743 bytes (SHA-256 812ba30d... as recorded in CE4-README). 1,720 is the character count.", - "proposal": "Note only; the SHA-256 and blob bind the source correctly.", + "finding": "GPT-6's pickup 6513952657 and review 5724470777 (01XJMJ) on Math-#543 record CounterexampleE4.lean (blob 3b7712e4) as 1,720 bytes; the blob is 1,743 bytes (1,720 is the character count). Review 5707632836 states no byte count. The SHA-256 in the README is correct. (v1.1: attribution corrected per 01XJMJ review 5724603343 F3; v1.0 pinned it on review 5707632836.)", + "proposal": "Note only; the SHA-256 and blob bind the source correctly. 01XJMJ posted its own correction at Math-#543 comment 6548199529.", "sources": [ "CE4-LEAN", "CE4-README" @@ -1146,13 +1149,13 @@ "metrics": { "claims": 23, "status_histogram": { - "Conditional": 1, + "Conditional": 3, "Conjectural": 4, - "Proved": 12, + "Proved": 10, "Proved under assumptions": 3, "Refuted": 3 }, - "open_obligations_listed": 35, + "open_obligations_listed": 37, "claims_whose_only_nonauthor_reader_is_session_017": [ "D1", "L2", diff --git a/reviews/elder_selection_chain_audit_20261031/check_claims.py b/reviews/elder_selection_chain_audit_20261031/check_claims.py index a693af30..4bad8df4 100644 --- a/reviews/elder_selection_chain_audit_20261031/check_claims.py +++ b/reviews/elder_selection_chain_audit_20261031/check_claims.py @@ -7,6 +7,9 @@ when the commit is available locally); * every claim uses a status from the owner's five-value taxonomy, cites at least one source, and every `depends_on`, `sources` and `reviews` reference resolves; + * a claim marked Proved depends only on claims marked Proved (status propagation: a Proved + claim resting on a Conditional, Conjectural, Refuted or Proved-under-assumptions claim is + Conditional by the record's own rule); * every review names its performer and verdict. Sources in another repository are checked only when that repository is given with `--repo NAME=DIR`; otherwise they are reported as SKIPPED and do not fail the run. @@ -64,6 +67,7 @@ def check_structure(doc, out): ok = True sources, reviews, claims = doc['sources'], doc['reviews'], doc['claims'] ids = [c['id'] for c in claims] + status_of = {c['id']: c['status'] for c in claims} if len(set(ids)) != len(ids): out.append('FAIL duplicate claim ids') ok = False @@ -91,6 +95,9 @@ def check_structure(doc, out): if d not in ids: out.append('FAIL %s: depends_on unknown claim %s' % (c['id'], d)) ok = False + elif c['status'] == 'Proved' and status_of[d] != 'Proved': + out.append('FAIL %s: Proved but depends on %s, which is %s' % (c['id'], d, status_of[d])) + ok = False if c['status'] == 'Proved' and not (c.get('reviews') or c.get('kernel_checked')): out.append('FAIL %s: Proved without a nonauthor review or kernel check' % c['id']) ok = False diff --git a/reviews/elder_selection_chain_audit_20261031/test_check_claims.py b/reviews/elder_selection_chain_audit_20261031/test_check_claims.py index daf2db67..77fa40ad 100644 --- a/reviews/elder_selection_chain_audit_20261031/test_check_claims.py +++ b/reviews/elder_selection_chain_audit_20261031/test_check_claims.py @@ -51,10 +51,16 @@ def test_wrong_byte_count_fails(self): def test_status_outside_taxonomy_fails(self): doc = load() - doc['claims'][0]['status'] = 'Accepted' + # a leaf claim, so that no status-propagation failure can mask the taxonomy check + leaf = next(c for c in doc['claims'] + if not any(c['id'] in d.get('depends_on', []) for d in doc['claims'])) + leaf['status'] = 'Accepted' ok, out = run_doc(doc) self.assertFalse(ok) - self.assertTrue(any('not in taxonomy' in line for line in out)) + self.assertTrue(any(line.startswith('FAIL %s: status' % leaf['id']) and 'not in taxonomy' in line + for line in out), '\n'.join(out)) + self.assertEqual([line for line in out if line.startswith('FAIL')], + ["FAIL %s: status 'Accepted' not in taxonomy" % leaf['id']]) def test_dangling_dependency_fails(self): doc = load() @@ -81,6 +87,93 @@ def test_other_repository_sources_are_skipped_not_failed(self): self.assertTrue(ok, '\n'.join(out)) self.assertTrue(all(line.startswith(('SKIPPED', 'RESULT')) for line in out), '\n'.join(out)) + def test_path_at_commit_must_map_to_recorded_blob(self): + doc = load() + keys = sorted(k for k, s in doc['sources'].items() if s['repo'] == 'Math-') + a, b = doc['sources'][keys[0]], doc['sources'][keys[1]] + self.assertEqual(a['commit'], b['commit']) + a['path'] = b['path'] + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertTrue(any(line.startswith('FAIL %s:' % keys[0]) and 'is blob' in line for line in out), + '\n'.join(out)) + + def test_blob_absent_from_object_store_fails(self): + doc = load() + key = sorted(k for k, s in doc['sources'].items() if s['repo'] == 'Math-')[0] + doc['sources'][key]['blob'] = '0' * 40 + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertTrue(any(line.startswith('FAIL %s: blob' % key) and 'not found' in line for line in out)) + + def test_duplicate_claim_ids_fail(self): + doc = load() + doc['claims'].append(copy.deepcopy(doc['claims'][0])) + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertIn('FAIL duplicate claim ids', out) + + def test_review_without_verdict_fails(self): + doc = load() + key = sorted(doc['reviews'])[0] + doc['reviews'][key]['verdict'] = '' + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertIn('FAIL review %s lacks verdict' % key, out) + + def test_claim_without_sources_fails(self): + doc = load() + doc['claims'][0]['sources'] = [] + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertIn('FAIL %s: no source' % doc['claims'][0]['id'], out) + + def test_unknown_source_reference_fails(self): + doc = load() + doc['claims'][0]['sources'].append('NO-SUCH-SOURCE') + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertIn('FAIL %s: unknown source NO-SUCH-SOURCE' % doc['claims'][0]['id'], out) + + def test_unknown_review_reference_fails(self): + doc = load() + doc['claims'][0].setdefault('reviews', []).append('NO-SUCH-REVIEW') + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertIn('FAIL %s: unknown review NO-SUCH-REVIEW' % doc['claims'][0]['id'], out) + + def test_conditional_without_dependency_fails(self): + doc = load() + target = next(c for c in doc['claims'] if c['status'] == 'Conditional') + target['depends_on'] = [] + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertIn('FAIL %s: Conditional without depends_on' % target['id'], out) + + def test_cli_exits_nonzero_on_mismatch(self): + doc = load() + key = sorted(k for k, s in doc['sources'].items() if s['repo'] == 'Math-')[0] + doc['sources'][key]['sha256'] = '0' * 64 + with tempfile.TemporaryDirectory() as tmp: + path = os.path.join(tmp, 'CLAIMS.json') + with open(path, 'w', encoding='utf-8') as fh: + json.dump(doc, fh) + proc = subprocess.run([sys.executable, '-B', '-S', os.path.join(HERE, 'check_claims.py'), + '--claims', path], capture_output=True) + self.assertEqual(proc.returncode, 1, proc.stdout.decode() + proc.stderr.decode()) + self.assertIn(b'RESULT FAIL', proc.stdout) + + def test_proved_depending_on_non_proved_fails(self): + doc = load() + status_of = {c['id']: c['status'] for c in doc['claims']} + parent = next(c for c in doc['claims'] if c['status'] == 'Proved') + weaker = next(c for c in doc['claims'] if c['status'] == 'Proved under assumptions') + parent.setdefault('depends_on', []).append(weaker['id']) + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertIn('FAIL %s: Proved but depends on %s, which is %s' + % (parent['id'], weaker['id'], status_of[weaker['id']]), out) + def test_cli_modes_agree(self): env = dict(os.environ, PYTHONDONTWRITEBYTECODE='1') a = subprocess.run([sys.executable, '-B', '-S', os.path.join(HERE, 'check_claims.py')], From b3e5d9ffe6eb685bca39cddddb444ffb01a496a3 Mon Sep 17 00:00:00 2001 From: Claude Date: Wed, 4 Nov 2026 23:35:14 +0000 Subject: [PATCH 3/3] Apply the OpenAI amendment readback to the elder-selection audit (v1.2) RB1 (Math-#554 comment 6645040299): finish propagating S1. The section-8 proposal for the PROOF_INDEX D1 row now separates Theorems A and B (Proved) from Theorem C (Conditional on the classical bar identification F5); F5's evidence names Theorem C and, through it, Theorem U as its consumers, not Theorem B; the section-2 bar-count conclusion carries the F5 qualification; the finite-bar definition is cited at P l.47; the v1.1 revision note says that the B, C and U statement fields were reworded. RB2: check_claims.py distinguishes an absent commit (NOTE, blob checked by identity only) from a present commit whose recorded path does not exist (FAIL). A missing-path test is added: the blob, size and digest of a source are kept and ".missing" is appended to its path. 19 tests OK in normal and -O mode; 19 single-check mutants of the checker, including the new path branch and the commit-presence check, are each killed by an assertion in both modes. No source identity, review record, status or histogram changes. Scientific effect NONE. Co-Authored-By: Claude Fable 5.1 Claude-Session: https://claude.ai/code/session_017Mi3hxjaxV45x6zo6o1ee3 --- .../AUDIT.md | 38 ++++++++++++------- .../CLAIMS.json | 6 +-- .../check_claims.py | 21 ++++++---- .../test_check_claims.py | 10 +++++ 4 files changed, 51 insertions(+), 24 deletions(-) diff --git a/reviews/elder_selection_chain_audit_20261031/AUDIT.md b/reviews/elder_selection_chain_audit_20261031/AUDIT.md index c32cb35d..a350aa04 100644 --- a/reviews/elder_selection_chain_audit_20261031/AUDIT.md +++ b/reviews/elder_selection_chain_audit_20261031/AUDIT.md @@ -150,11 +150,13 @@ L3), with the exact limit `α₁ + α₂` only a candidate (LP-EQ). **Good configurations at leading order.** P section 13 removes the compact-mark restriction by dominated convergence: the unnormalized intensity has the target-growth majorant (13.4), `p_r → 1` pointwise by Theorem A on compact -neighbourhoods (P l.415), and the far population is `O(1)` (14.1). That is Theorem C's leading term (C). What does -not exist, and cannot, is a quantitative unrestricted rate: Theorem U shows `ν_cand^all − ν_eld^all ≥ c_* > 0` (U), so -candidate C7 is Refuted and the `O(1)` rejected density (RHO) is a candidate. "Counting candidate maximum–saddle pairs -is not the same as counting actual persistence bars" is therefore a theorem of the project, at leading order and in -the difference, not a gap: the two counts agree to leading order `ℓ^{−1/3}` and differ by a positive bounded-below +neighbourhoods (P l.415), and the far population is `O(1)` (14.1). That is Theorem C's leading term (C) for the +elder-selected pairs; reading it as a statement about finite bars uses the Morse-locus identification F5 (P l.445), +which is why C is Conditional on F5 alone. What does not exist, and cannot, is a quantitative unrestricted rate: +Theorem U shows `ν_cand^all − ν_eld^all ≥ c_* > 0` (U, Conditional on F5 in the same way), so candidate C7 is Refuted +and the `O(1)` rejected density (RHO) is a candidate. "Counting candidate maximum–saddle pairs is not the same as +counting actual persistence bars" is therefore, granted F5, a theorem of the project at leading order and in the +difference, not a gap: the two counts agree to leading order `ℓ^{−1/3}` and differ by a positive bounded-below amount. **What remains open mathematically.** (a) The alignment review of D1 (I-4). (b) Formalization of anything @@ -223,7 +225,8 @@ exact failure mechanism (LP-P) and from the far-pair separation of Theorem U. `CLAIMS.json` gives each claim its statement, event, status, evidence, exact sources (repository, commit, path, blob, bytes, SHA-256), reviews (performer, record, verdict, independence flags), `depends_on` and `obligations`. `check_claims.py` verifies that every cited blob exists in the local object store with the recorded length and digest, -that the recorded commit maps the path to that blob when the commit is available, that every status is one of the five +that the recorded commit maps the path to that blob when the commit is available (an absent commit is a NOTE; a present +commit whose path does not exist is a FAIL), that every status is one of the five values, that every `depends_on`, source and review reference resolves, and that no claim is "Proved" without a review or kernel check. `test_check_claims.py` exercises it, including tampered-digest, wrong-length, unknown-status, dangling dependency and unreviewed-"Proved" failures, and byte-identical output in normal and optimized Python. The checker does @@ -248,11 +251,14 @@ Computed from this record at the cuts above; the values are in `CLAIMS.json` und ## 8. Proposed wording for the canonical surfaces (for the integrator; not applied here) -- PROOF_INDEX D1 row: append "Status (owner taxonomy, 2026-10-31 audit): Theorems A, B, C **Proved** at existential - scope, prose, nonauthor AI reads by xAI/Grok and Anthropic session_017 only; no formal statement about the actual - event (register BLOCKED_INTERFACE; Math-#543 countermodel). Deterministic cap step **Proved (kernel)** in - `frontiers/cap_first_exit_lean_20261002`, alignment review open. Lower bound `1 − p_r ≥ c r³`: **Proved** for d = 2 - (EL, read) and for every fixed d ≥ 3 (DL, read, Math-#129); CL-ELDER-LOWER-ALL-D is a candidate." +- PROOF_INDEX D1 row: append "Status (owner taxonomy, 2026-10-31 audit, v1.2): Theorem A (elder selection) and + Theorem B (elder-selected compact-window pairs) **Proved** at existential scope, prose, nonauthor AI reads by + xAI/Grok and Anthropic session_017 only. Theorem C (finite bars) **Conditional** on the classical Morse-locus + identification of elder-selected pairs with bars (P l.445; an interface clause in Lean), its Gaussian analysis read + at the same scope. No formal statement about the actual event exists (register BLOCKED_INTERFACE; Math-#543 + countermodel). Deterministic cap step **Proved (kernel)** in `frontiers/cap_first_exit_lean_20261002`, alignment + review open. Lower bound `1 − p_r ≥ c r³`: **Proved** for d = 2 (EL, read) and for every fixed d ≥ 3 (DL, read, + Math-#129); CL-ELDER-LOWER-ALL-D is a candidate." - main `docs/DISCOVERY_PRIORITY_LEDGER_20260925.md` DP-02 current-state cell: resolve against `STATUS.md` D1 (I-1). - `STATUS.md` "How to verify" or the D1 row: name the formal-coverage boundary in one sentence: "No kernel-checked statement of Theorems A, B or C about the elder event exists; `PairLawInterface.p` is primitive." @@ -266,8 +272,14 @@ Computed from this record at the cuts above; the values are in `CLAIMS.json` und AMEND S1), 01SHCrxv (Anthropic; custody 44/44 and S1, [main#684 6548184976](https://github.com/d6g8k5htny-coder/main/issues/684#issuecomment-6548184976)). - **v1.1.** S1: C and U Conditional on F5; B's dependency on F5 removed because P states Theorem B through the elder relation; a status-propagation rule added to `check_claims.py` with a test. F1: 01XJMJ's nine checker tests added. - F3: the I-3 attribution corrected. F4: the I-2 wording corrected. No source identity, review record, claim statement - or inconsistency other than I-2 and I-3 changed. + F3: the I-3 attribution corrected. F4: the I-2 wording corrected. No source identity or review record changed; the + statement fields of B, C and U were reworded to name their populations (elder-selected pairs for B, finite bars for + C and U), and the inconsistency rows I-2 and I-3 were reworded. +- **v1.2.** OpenAI amendment readback ([6645040299](https://github.com/d6g8k5htny-coder/Math-/pull/554#issuecomment-6645040299), + AMEND_SCOPED_RECORD). RB1: the §8 proposal now separates Theorems A and B (Proved) from Theorem C (Conditional on + F5); F5's evidence names C and U, not B, as its consumers; §2's bar-count conclusion carries the F5 qualification; + the finite-bar definition is cited at P l.47. RB2: `check_claims.py` now distinguishes an absent commit (NOTE) from a + present commit whose recorded path does not exist (FAIL), with a missing-path test. Histogram unchanged. ## 10. Non-claims diff --git a/reviews/elder_selection_chain_audit_20261031/CLAIMS.json b/reviews/elder_selection_chain_audit_20261031/CLAIMS.json index ad82241e..f715d0bb 100644 --- a/reviews/elder_selection_chain_audit_20261031/CLAIMS.json +++ b/reviews/elder_selection_chain_audit_20261031/CLAIMS.json @@ -1,6 +1,6 @@ { "schema": "elder-selection-chain-audit/v1", - "object": "CL-ELDER-SELECTION-CHAIN-AUDIT-20261031-v1.1", + "object": "CL-ELDER-SELECTION-CHAIN-AUDIT-20261031-v1.2", "author": "Anthropic Claude, Claude Code session_017Mi3hxjaxV45x6zo6o1ee3, delegated AI work for Dylan Roy", "scientific_effect": "NONE", "kind": "source-bound review record; proposes PROOF_INDEX/STATUS wording for the integrator; not a register", @@ -868,7 +868,7 @@ "statement": "On the Morse distinct-value locus, d_f(M) is the death level of the superlevel H0 class born at M (elder rule), and each finite bar has exactly one local-maximum birth and one index-(d-1) merging-saddle death, so the elder-selected candidate pairs are exactly the finite bars (P l.290-297, l.445).", "status": "Proved under assumptions", "proof_form": "prose", - "evidence": "Classical persistence/Morse theory asserted in P and consumed by Theorems B and C; read with sections 8 and 14 by REV-G1 and REV-C1. In Lean it is an interface clause (LIFETIME-LEAN elderPartner_iff l.348, deathHeight_maximin l.341) and, for the cap case only, CapHyp.elder_death_level / elder_death_level_peak (CFE-LEAN l.217, l.338); the compact Morse adapter (CMA-PROOF) supplies one interval summand per maximum, not the bar bijection; torus_filtration (TF-SCOPE) lists once-counting, interval decomposition and bar bijection as not established.", + "evidence": "Classical persistence/Morse theory asserted in P and consumed by Theorem C (bars, P l.47 and l.445) and through it by Theorem U; Theorem B is stated through the elder relation (P l.37) and does not consume it (v1.2, RB1). Read with sections 8 and 14 by REV-G1 and REV-C1. In Lean it is an interface clause (LIFETIME-LEAN elderPartner_iff l.348, deathHeight_maximin l.341) and, for the cap case only, CapHyp.elder_death_level / elder_death_level_peak (CFE-LEAN l.217, l.338); the compact Morse adapter (CMA-PROOF) supplies one interval summand per maximum, not the bar bijection; torus_filtration (TF-SCOPE) lists once-counting, interval decomposition and bar bijection as not established.", "sources": [ "P", "LIFETIME-LEAN", @@ -918,7 +918,7 @@ "id": "C", "title": "Theorem C (P 1.3): unrestricted leading density nu_cand^all ~ nu_eld^all ~ c_{d,L} ell^{-1/3}", "event": "bars-vs-candidates", - "statement": "All births, gaps and separations; essential class excluded; nu_eld^all counts all finite ordinary superlevel H0 bars (P l.44). Near pairs: dominated convergence with the target-growth majorant (13.4) and p_r -> 1 pointwise from Theorem A on compact neighbourhoods (P l.415); far pairs: nu^far <= C (14.1); c_{d,L} is the integral (13.6)/(15.2). The selected pair count equals the finite-bar count by the Morse-locus identification (P l.445), which is F5. Leading term only: no unrestricted remainder or difference rate.", + "statement": "All births, gaps and separations; essential class excluded; nu_eld^all counts all finite ordinary superlevel H0 bars (P l.47). Near pairs: dominated convergence with the target-growth majorant (13.4) and p_r -> 1 pointwise from Theorem A on compact neighbourhoods (P l.415); far pairs: nu^far <= C (14.1); c_{d,L} is the integral (13.6)/(15.2). The selected pair count equals the finite-bar count by the Morse-locus identification (P l.445), which is F5. Leading term only: no unrestricted remainder or difference rate.", "status": "Conditional", "proof_form": "prose", "evidence": "The Gaussian analysis is complete and read: REV-G1, REV-C1, REV-C2 (spot re-derivations of (13.1), (13.4), (15.2)); REC section 2 records ACCEPT. Its only non-Proved dependency is F5, the classical bar identification, which this record holds at Proved under assumptions (asserted in P, an interface clause in Lean); under the record's rule that makes Theorem C Conditional on F5 and on nothing else (v1.1, S1 of the Grok Bot a4 and 01SHCrxv reads; v1.0 had Proved).", diff --git a/reviews/elder_selection_chain_audit_20261031/check_claims.py b/reviews/elder_selection_chain_audit_20261031/check_claims.py index 4bad8df4..fe1e88c4 100644 --- a/reviews/elder_selection_chain_audit_20261031/check_claims.py +++ b/reviews/elder_selection_chain_audit_20261031/check_claims.py @@ -3,8 +3,9 @@ What it checks, and nothing more: * every source entry resolves in the local Git object store to a blob of exactly the recorded - byte length and SHA-256 (the recorded commit is also checked to map the path to that blob - when the commit is available locally); + byte length and SHA-256; when the recorded commit is available locally, the recorded path + must exist at that commit and map to that blob (a missing path is FAIL; only an absent + commit is a NOTE); * every claim uses a status from the owner's five-value taxonomy, cites at least one source, and every `depends_on`, `sources` and `reviews` reference resolves; * a claim marked Proved depends only on claims marked Proved (status propagation: a Proved @@ -49,15 +50,19 @@ def check_source(key, src, repos, out): if digest != src['sha256']: out.append('FAIL %s: sha256 %s, recorded %s' % (key, digest, src['sha256'])) ok = False - at = git(repo_dir, 'rev-parse', '--verify', '--quiet', '%s:%s' % (src['commit'], src['path'])) - if at.returncode == 0: - if at.stdout.decode().strip() != src['blob']: + have_commit = git(repo_dir, 'cat-file', '-e', src['commit'] + '^{commit}').returncode == 0 + if not have_commit: + out.append('NOTE %s: commit %s not available locally; blob checked by identity only' + % (key, src['commit'][:12])) + else: + at = git(repo_dir, 'rev-parse', '--verify', '--quiet', '%s:%s' % (src['commit'], src['path'])) + if at.returncode != 0: + out.append('FAIL %s: path %s does not exist at commit %s' % (key, src['path'], src['commit'][:12])) + ok = False + elif at.stdout.decode().strip() != src['blob']: out.append('FAIL %s: %s:%s is blob %s, recorded %s' % (key, src['commit'][:12], src['path'], at.stdout.decode().strip(), src['blob'])) ok = False - else: - out.append('NOTE %s: commit %s not available locally; blob checked by identity only' - % (key, src['commit'][:12])) if ok: out.append('OK %s %s %d %s' % (key, src['blob'][:12], src['bytes'], src['path'])) return ok diff --git a/reviews/elder_selection_chain_audit_20261031/test_check_claims.py b/reviews/elder_selection_chain_audit_20261031/test_check_claims.py index 77fa40ad..e79ea234 100644 --- a/reviews/elder_selection_chain_audit_20261031/test_check_claims.py +++ b/reviews/elder_selection_chain_audit_20261031/test_check_claims.py @@ -174,6 +174,16 @@ def test_proved_depending_on_non_proved_fails(self): self.assertIn('FAIL %s: Proved but depends on %s, which is %s' % (parent['id'], weaker['id'], status_of[weaker['id']]), out) + def test_missing_path_at_present_commit_fails(self): + doc = load() + key = sorted(k for k, s in doc['sources'].items() if s['repo'] == 'Math-')[0] + doc['sources'][key]['path'] += '.missing' # blob, size and digest unchanged + ok, out = run_doc(doc) + self.assertFalse(ok) + self.assertIn('FAIL %s: path %s does not exist at commit %s' + % (key, doc['sources'][key]['path'], doc['sources'][key]['commit'][:12]), out) + self.assertFalse(any(line.startswith('OK %s ' % key) for line in out)) + def test_cli_modes_agree(self): env = dict(os.environ, PYTHONDONTWRITEBYTECODE='1') a = subprocess.run([sys.executable, '-B', '-S', os.path.join(HERE, 'check_claims.py')],