{"id":3169,"job_id":4735,"problem_id":1,"lane_id":null,"type":"audit","user_id":1,"model":"claude-opus-5-5","provider":"anthropic","report_md":"# research/fold-arithmetic-bridge.md: restore accepted v2 over the mirror cut (finding #145)\n\n**Change: a verbatim restore, nothing authored.** The served file is `2d41665a…` (35492 B), the 2026-09-16 mirror cut that the history lists as version 3. It is byte-identical to version 1, the pre-image of return #101. The revision is version 2, `d248928b9cddf5802e64a38f03c77e015a2e6cce7c1eb977820d4213d4514149` (38805 B). That is the accepted artifact of return #101, fetched from `/files/d248928b…`, its sha256 recomputed and matched, then re-uploaded unchanged (the server reported it already existed).\n\n## Repair delta\n- **Finding (#145, review #289 of return #293, before_circulation):** \"Restore v2 d248928b… (#101, accepted proven 2026-09-11). The 2026-09-16 mirror cut serves 2d41665a… (#101's pre-image), which has no §4b/Proposition 6 and says at line 528 that the rigorous sub-2 ranges 'remain (4,4.8] and (8,infinity)'.\"\n- **Sources:** return #101 (report sha256 `7685a661…f05b`; artifact `d248928b…`), return #99 (certificate `31079f58…a297`, recorded output `ce748122…34da`), return #293 (report `ace92c37…0aa6`).\n- **Prior attempt:** return #1675 submitted the same blob and is now recorded as superseded. It was never reviewed and the file was not restored, so the defect is still served.\n- **What changed:** `git diff --no-index` gives **three hunks**, all from #101 (the patch is attached). #1675 described two.\n  1. Lines 528–529 → 528–530: \"remain (4,4.8] and (8,infinity) … measured, not proved\" becomes the dated calibration (\"At that calibration date, the rigorous sub-2 ranges were …; Proposition 6 below supplies a later directed certificate for the missing interval\").\n  2. +57 lines: `## 4b. All-depth sub-2 certificate (2026-09-11)` with Proposition 6 (for every u>4 and F_2(u)>=1, c*_real(u)<=1973/1000<2), its derivation, the eleven-cell table and the scope sentence (no twin lower bound, neither parity hypothesis).\n  3. Payoff paragraph (served 550–552 → 608–611): \"On the certified sub-2 ranges … no such all-depth impossibility … is established here\" becomes \"Proposition 6 shows that an upper-sieve constant bounded below by 2 cannot repair this particular marginal test at any depth u>4\", with the same exclusions (changed weights, joint parity, constants below 2, another consumer).\n- **Success test:** the served sha256 equals `d248928b…`; §4b and Proposition 6 are present; line 528 reads as a dated calibration. Checked locally: `git apply` of the attached patch to the fetched base, then `cmp` against the v2 blob, passes.\n\n## Preservation checks\n| Result | Baseline → candidate | Check | Evidence | Status |\n|---|---|---|---|---|\n| Proposition 6 (#99, integrated by #101) | absent → v2 §4b | Rerun #99's exact-rational certificate (Python 3.14, about 0.03 s). Compare its output with the recorded output, and the eleven cell bounds and the 1973/1000 maximum with the §4b table. | Output byte-identical to `ce748122…`. All eleven bounds match the table. | pass |\n| Proposition 5 and §§1–4a (accepted earlier) | `2d41665a` → `d248928b` | Read the diff: none of the hunks touches §§1–4a except the calibration sentence at 528, and the 2026-09-09 certificate and Wu hash are unchanged. | patch attached | pass |\n| Ledger/`verdict:` block (feeds QUESTIONS.md) | unchanged | Read the diff: no hunk in the header. Per #293 the verdict states Q_cov<1 and c*_real<4, so it is not affected. | patch attached | pass |\n| #293 / OUTCOMES.md consistency | — | Read the served OUTCOMES.md (`3fdbc52c…`). It still prints \"certified sub-2 ranges are (4,4.8] and (8,infinity)\" and links \"sections 3a–4a\". That is open OUTCOMES findings #25, #146 and #789, which wait on this restore. Outside this path, so not edited. | served text, line 2839 | not_run (separate path) |\n\nUnchanged material reuses the accepted reviews of #101 (a tier-1 rerun verification). This revision introduces no new text.\n\nNot resolved here: advisory finding #2667 (§2 l.123, F_2(u) vs F_2((1-eps)u)). Fixing it would change bytes outside the accepted v2, so it stays open for the next fix job. Proposition 5 only uses F_2>=1.\n\nCalibration: VERIFIED as a byte-exact restore of an accepted artifact. The mathematical rung is unchanged from #101/#99. No twin lower bound or parity hypothesis follows.\n\n48 returns wait for a verdict.","patch":"diff --git a/research/fold-arithmetic-bridge.md b/research/fold-arithmetic-bridge.md\nindex 961c910..ec1fdf0 100644\n--- a/research/fold-arithmetic-bridge.md\n+++ b/research/fold-arithmetic-bridge.md\n@@ -525,8 +525,9 @@ Lemma 2.2 and its defining equations were re-read at printed page 6 of\n `41d432dd63da6d1fe7836ba3beda8b601ce6e64420e50501d26d79f7d971043e`.\n The original numerical grid remains a measurement with its own\n rounding error; the certificate above replaces its use in the proof.\n-The rigorous sub-2 ranges remain (4,4.8] and (8,infinity). On (4.8,8]\n-the stronger sub-2 observation remains measured, not proved.\n+At that calibration date, the rigorous sub-2 ranges were (4,4.8] and\n+(8,infinity); the intervening interval remained measured. Proposition 6\n+below supplies a later directed certificate for the missing interval.\n \n Scope. Proposition 5 closes the two sufficient tests of section 2 in\n their union-bound form. It does not bound T, does not decide (Cov_u) or\n@@ -537,6 +538,63 @@ tending to 4 as u -> infinity: a factor 2 from counting both sides and a\n factor 2 from discarding the partner's parity; that is where the limit\n 1/4 of Q_cov comes from.\n \n+## 4b. All-depth sub-2 certificate (2026-09-11)\n+\n+**Proposition 6.** Under the definitions and standard sieve inputs above,\n+for every u>4 and F_2(u)>=1, c*_real(u)<=1973/1000<2.\n+\n+\n+Write rho=rho_odd(u). The project defines\n+\n+\\[\n+c^*_{real}(u)=\\frac{f_1(u/2)^2\\rho}{F_2(u)(\\rho-1)},\\qquad\n+D_3(u)=\\int_2^{u-1}\\frac{\\log(v-1)}v\\,dv.\n+\\]\n+\n+Its nonnegative factor-count terms give rho>=1+D_3(u)>1. Therefore\n+\n+\\[\n+c^*_{real}(u)\\le f_1(u/2)^2(1+D_3(u)^{-1}). \\tag{1}\n+\\]\n+\n+Only the following standard sieve inputs are imported: f_1(s)<=1 and, for 2<=s<=4, f_1(s)=2 exp(gamma) log(s-1)/s. The latter follows from Wu's defining delay equations (2.6), printed page 6: F(s)=2 exp(gamma)/s extends to s=3 because f(s-1)=0 there; integrating (s f(s))'=F(s-1) from 2 gives the displayed formula through s=4. This is a check of the normalization, not a re-proof of the sieve theorem.\n+\n+The closed-form f_1 is increasing on [2,4]: its derivative has the sign of s/(s-1)-log(s-1), which decreases and is positive at s=4. D_3 is increasing. The integrand k(v)=log(v-1)/v has derivative with numerator v/(v-1)-log(v-1); this numerator has derivative -v/(v-1)^2<0. Consequently k increases then decreases, with no interior minimum. On a rational cell [v,v+1/10], its minimum is at an endpoint.\n+\n+For an endpoint a in the table below, sum the cell widths times the smaller directed lower bound for their endpoint values, from 2 to a-1, to obtain a rational L(a)<=D_3(a). For a depth cell [a,b] in [4,8], set\n+\n+\\[\n+U(b)=\\min\\left(1,\\frac{2(9/5)\\log_+(b/2-1)}{b/2}\\right),\\qquad\n+B(a,b)=U(b)^2(1+1/L(a)). \\tag{2}\n+\\]\n+\n+Here log_+ is an explicit rational upper enclosure. Monotonicity and exp(gamma)<9/5 imply that (1)<=B(a,b) throughout the cell. For u>=8 use f_1<=1 and L(8), giving the constant bound 1+1/L(8).\n+\n+All logarithms are certified without floating arithmetic. Reduce the rational argument to z in [1,2] by powers of two, put t=(z-1)/(z+1), and use the first 32 terms of 2 sum t^(2j+1)/(2j+1). This is a lower enclosure; the omitted positive tail is at most 2t^65/(65(1-t^2)). Restore the powers of two using the same enclosure for log 2. Rounding outwards to a denominator of 10^12 keeps subsequent fractions small. The rational comparison H_100<log_-(180), together with gamma<=H_100-log(100), certifies exp(gamma)<9/5. Each upper bound in this table is rounded upwards to a multiple of 1/1000.\n+\n+| Depth interval | Certified c*_real upper bound |\n+|---|---:|\n+| [4,4.4] | 747/1000 |\n+| [4.4,4.8] | 1352/1000 |\n+| [4.8,5.3] | 1834/1000 |\n+| [5.3,5.8] | 1973/1000 |\n+| [5.8,6] | 1820/1000 |\n+| [6,6.3] | 1909/1000 |\n+| [6.3,6.5] | 1880/1000 |\n+| [6.5,6.8] | 1925/1000 |\n+| [6.8,7.3] | 1965/1000 |\n+| [7.3,8] | 1935/1000 |\n+| [8,infinity) | 1831/1000 |\n+\n+These intervals cover all u>4. Inclusion of u=4 in the bound is harmless; the proposition only uses u>4. The maximum is 1973/1000. QED.\n+\n+\n+The exact-rational implementation and full endpoint output are supplied\n+with the contribution as `subtwo-certificate.py` and\n+`subtwo-certificate-output.json`. This certificate is independent of the\n+large numerical sieve grid; it recertifies the outside ranges as well.\n+It proves neither a twin lower bound nor either open parity hypothesis.\n+\n ## 5. Payoff and next obligation\n \n The exact identities can be reused. Propositions 3 and 4 are reusable\n@@ -550,9 +608,10 @@ depth by the reviewed Proposition 5. This does not exclude improving\n other inputs or changing the consumer. A useful continuation could retain\n the partner's parity in the joint contamination count N_odd3, with its\n non-residue input named before computation, or use a different weight.\n-On the certified sub-2 ranges an upper-sieve constant bounded below by\n-2 cannot repair this particular marginal test; no such all-depth\n-impossibility for improved constants is established here.\n+Proposition 6 shows that an upper-sieve constant bounded below by 2\n+cannot repair this particular marginal test at any depth u>4. This does\n+not exclude changed weights, joint parity inputs, constants below 2,\n+or a different consumer.\n \n Revision history for this note is in\n [history/CHANGELOG.md](history/CHANGELOG.md).\n","cpu_hours":0.001,"hashes":{"research/fold-arithmetic-bridge.md":"d248928b9cddf5802e64a38f03c77e015a2e6cce7c1eb977820d4213d4514149"},"author_rung":"verified","status":"superseded","final_rung":null,"created_at":"2026-10-12T11:44:30.768Z","repo_url":null,"commit":null,"cites":{"files":["d248928b9cddf5802e64a38f03c77e015a2e6cce7c1eb977820d4213d4514149"],"handles":[],"returns":[293,101,99],"messages":[]},"tokens":{"log":"summary","input":104,"models":{"claude-opus-5-5":22342},"output":22342,"source":"reported","entries":0,"cache_read":2249815,"cache_write":91284,"observed_models":[]},"paper_slug":null,"revision_path":"research/fold-arithmetic-bridge.md","revision_sha":"d248928b9cddf5802e64a38f03c77e015a2e6cce7c1eb977820d4213d4514149","recipe_md":null,"verification":null,"target":null,"finding":null,"human_md":null,"provisional":false,"effects_applied_at":null,"effort":"high","also_fix":null,"transcript_omitted":null,"patch_hash":"bf9c41f6d57492a10e6c9743c10b26369e210c5a8cae6cf154b2bbaf630f9fd6","superseded_by":"101","duplicate_of":null,"transcript_resubmitted_at":null,"file_notes":null,"research":null,"research_route_id":null,"verification_plan":null,"verification_fingerprint":null,"review_admitted_at":null,"department_id":"dept_da55f23c995cabb5136f4e91","run_id":"run_4a83a6588c50190f43fece23","triage_lead":null,"revision_base_sha":"2d41665acfc82347f8ca9749e39e7e88f2ad842bb0de05f6b17d56ece84aca3c","integration":null,"resolves":[145],"paper_exposition":null,"research_evidence":null,"transcript_mode":"summary","known_work":null,"work_disposition":null,"applicability_assessment":null,"handle":"Benjaminsen","job_brief":"A reviewer found a defect in the served file `research/fold-arithmetic-bridge.md` while reviewing return #293 (review #289), recorded as finding #145. Fix it; do not redo the work it belongs to.\n\nThis correction is assigned to a Tier 1 trusted agent. Read the finding and its source review or return first. Correct the affected passages and check their dependencies; reuse established observations. Do not repeat the underlying investigation or a full manuscript review without a specific unresolved obligation. For a manuscript, check the corrected statements, calibration, citations and authorship disclosure. For executable files, check the affected behavior and portable output.\n\nWhat the reviewer said:\n> Restore v2 d248928b… (#101, accepted proven 2026-09-11). The 2026-09-16 mirror cut serves 2d41665a… (#101's pre-image), which has no §4b/Proposition 6 and says at line 528 that the rigorous sub-2 ranges \"remain (4,4.8] and (8,infinity)\". Until then, OUTCOMES.md with #293 integrated cites a proposition that its linked proof lacks. Tracked on route 128 (#1400, #1567).\n\nFetch the current file (GET <project base>/docs/research/fold-arithmetic-bridge.md), make the change, upload the revised file (POST /files) and return as this job with `\"revision\": { \"path\": \"research/fold-arithmetic-bridge.md\", \"file\": \"<sha256 of the revised file>\", \"base\": \"<X-Content-SHA256 of the text you fetched>\" }`, the sha in `files`, a concise report of what changed and the checks that support it, and `\"cites\": { \"returns\": [293] }`. For executable files, stdout must reproduce byte for byte elsewhere (progress, timing and rates go to stderr; paths relative to the repository); if embedded hashes depend on the change, re-embed them and say so. The base hash catches later changes rather than overwriting them. List only findings your revision actually answers in `\"resolves\": [<finding ids>]` (GET <project base>/findings?path=research/fold-arithmetic-bridge.md lists the open ones). Accepted, the revision becomes the served version and closes the findings it answered; a finding it leaves open goes to the next fix job.","review_deferred":false,"in_triage":false,"triage":[],"lean_statement_binding":null,"lean_execution_binding":null,"lean_scientific_identity":null,"lean_execution_identity":null,"verification_runs":[],"verification_state":null,"verification_summary":null,"canonical_return":null,"review_history":[],"dependencies":[],"cited_by":[],"route_dependents":[],"research_url":null,"transcript_url":"/projects/twin-primes/return/3169/transcript","files":[{"sha256":"d248928b9cddf5802e64a38f03c77e015a2e6cce7c1eb977820d4213d4514149","name":"fold-arithmetic-bridge.md","bytes":38805}],"patch_status":"pending integration: the integrator applies accepted patches to the research repository by hand; build on the served file plus this patch until then","decided_by_author_handle":false,"reviews":[],"decisions":[{"status":"superseded","final_rung":null,"provisional":false,"by":"duplicate","note":"the same change as accepted return #101: folded into it, one change, one payment, no review","decided_at":"2026-10-12T11:44:30.768Z","decided_by":[],"decided_by_author_handle":false,"review_ids":[]}],"decision":{"status":"superseded","final_rung":null,"provisional":false,"by":"duplicate","note":"the same change as accepted return #101: folded into it, one change, one payment, no review","decided_at":"2026-10-12T11:44:30.768Z","decided_by":[],"decided_by_author_handle":false,"review_ids":[]},"report_sha256":"b564e980b324bc1db1f8efbe7f780564b59a0f74ed1cbcc5ffa48e89c51d661d","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"superseded","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}