{"id":6751,"job_id":13103,"problem_id":6,"lane_id":null,"type":"explore","user_id":90,"model":"muse-spark","provider":"unknown","report_md":"# Job 13103 (route 300, repair): return #4310 brought to a checkable state — outcome progress\n\n## Question\nDoes the t = 13 connected path (padded tunnel 20 at L = 63) give a Q24 rate within 2x of path 42 and a Q25..Q29 gain over path 31 matching its +3 tunnel bits? #4310 tried to answer this with a generator benchmark and was rejected as unverifiable. This return repairs it by comparison, without running the experiment.\n\n## What was compared (in order)\n- #4290 (route's own): served generator, Stage D fix, detection method, and per-CPU-s baselines (path 42 at L=61: Q24 59.6, Q25 317, Q26 0.066). No path 31 rates measured.\n- #4305 (route's own): candidate c0083 exists with static T9 22, z3 padded tunnel 20 at L=63, SAT rows 1..24, 380 conditions. No generator run.\n- #4307 (route 301): t=13 connect works in the same generator class on its own paths (path 6 at L=63: Q24 6.79, Q25 18505, Q26 1.58 per CPU-s). Those rates are for paths 6/100, not c0083, and are not transferred.\n- #4306 (route 296): S-differential UNSAT under padding; a different differential family, set aside.\n- #4310 itself: files [], no recipe, no logs; its quoted #4305 figures are correct but restated.\n\n## Decisive repair finding: the \"10 vs 20\" tunnel discrepancy is a header-alignment artifact\nThe converter #4310 used emits conditions arrays of length 34 with Q1 at index 0, but the generator indexes conditions by row number (C0[t] is row t; the canonical header builder emits length 35 with a dummy index 0, confirmed against served c31.h). Fed that header, the generator's \"Q9\" was really Q10, \"Q10\" was Q11, and \"Q11\" was Q12. Recomputed from the served candidate, the misaligned header's tunnel value is exactly 0xc020c037 = 10 bits — #4310's \"10\". The three quantities are: static T9 22 bits (0xfe1cbe3f), z3 padded tunnel 20 bits (mask fa1c3e3f), misaligned-header T9 10 bits. There is no contradiction once aligned. With every row checked against the next row's conditions, bases=0 cannot be attributed to the t=13 path's architecture, and the Q13-density story is moot (the run's \"Q13\" was really Q14's row). #4310 also used an unserved generator instead of the brief's served xfg.c, and its \"54.3 Q24/s\" for path 42 appears in neither #4305 nor #4290; the recorded figure is 59.6 per CPU-s at L=61, and that figure is dropped.\n\n## Repaired package (staged locally; recipe below rebuilds it from served files)\n- Fixed converter emitting length-35 headers (dummy index 0), producing conds_t13_fixed.h (sha256 f16ad991...) with row-indexed T9 = 22 bits, matching #4305's static T9.\n- Served job8819-xfg.c, byte-identical (sha256 1f213b28...).\n- The repaired build was compile-checked only (gcc -O3, znver4 tuning, PADL=63, detection PADX11) and never executed: no DETECT rerun and no 2-seed x 300 s comparison were done here, per the job's constraints.\n\n## Outcome: progress\nThe step is not answered (\"known\" fails: #4307 measures other paths, not c0083) and the old step cannot simply be copied (\"promising\" fails: it presupposes a working header and DETECT-ready build, which #4310's attempt shows is broken). The new next_step replaces it: repaired header plus served generator at PADL=63, a 4 s detection for PADX11, a DETECT gate, and only then the 2-seed x 300 s comparison against path 31 (L=63) and path 42 (L=61) on one machine, with the original success/failure bars.\n","patch":null,"cpu_hours":0,"hashes":{},"author_rung":null,"status":"recorded","final_rung":"recorded","created_at":"2026-11-01T00:09:39.308Z","repo_url":null,"commit":null,"cites":{"returns":[4310]},"tokens":{"log":"summary","input":0,"models":{},"output":0,"source":"none","entries":0,"cache_read":0,"cache_write":0,"observed_models":[]},"paper_slug":null,"revision_path":null,"revision_sha":null,"recipe_md":"Rebuild the repaired package from served files (no new experiment):\n1. Fetch the served candidate: /files/24b69064ce98fa3b5a0184b8569ef97e98d3c31d0fa484b4f4505b1b4efd93f0?raw=1 — must hash to 24b69064ce98fa3b5a0184b8569ef97e98d3c31d0fa484b4f4505b1b4efd93f0 (cand_t13_c0083.txt, 2922 bytes). Fetch the served generator: /files/1f213b28eb3059ee4291eb2e27a2356496980676c98a1cee8dfc2394c758d6ef?raw=1 — must hash to 1f213b28eb3059ee4291eb2e27a2356496980676c98a1cee8dfc2394c758d6ef (job8819-xfg.c, 15435 bytes).\n2. Convert the candidate with the fixed converter: the published txt2conds.py logic, except each conditions array is emitted as length 35 with a leading 0 dummy (index 0 unused, rows 1..34 follow), matching the generator's row indexing and the canonical header builder. Command form: python3 txt2conds_fixed.py cand_t13_c0083.txt conds_t13_fixed.h — expected sha256 f16ad9917e85819999c9e78c3ad8d2cdb67489a7d773e27a7a9da489cb9a1889 (2276 bytes). Check: entry 0 is 0 in every array, and the row-indexed tunnel value (Q9 free, Q10 not forced 1, Q11 not forced 0) is 0xfe1cbe3f = 22 bits, matching #4305's static T9. The unfixed converter reproduces the old local header (sha256 6f04b648...) and gives 0xc020c037 = 10 bits — that is the misaligned value behind #4310's \"10\", do not use it.\n3. Compile-check only (no execution in this repair): gcc -O3 -march=znver4 -mtune=znver4 -pthread -DCONDS='\"conds_t13_fixed.h\"' -DPADL=63 -DPADX11=0xffffffffu -o xfg_t13_detect xfg_t13.c — expected exit 0. Do NOT run selftest as a gate: it is served-path-specific. Do NOT run DETECT or benchmarks here; they belong to the next_step.\n4. Next_step execution (not done here): build the detection arm as above, run 4 s, read Q11_T9_always_zero from the JSON output for the measurement PADX11 (cross-check only: #4305's z3 padded mask fa1c3e3f implies z3-lost bits 0x04008000; detection, not z3, sets PADX11). Build the measurement arm with that PADX11. Gate: if DETECT bases/Q24 are still zero with the aligned header, stop and report generator failure confirmed for this configuration. If nonzero, run 2 seeds x 300 s on one machine for c0083 (L=63), path 31 (L=63) and path 42 (L=61); report Q24/Q25/Q26 per CPU-s with seeds, wall/CPU times and exit statuses, and judge against the next_step bars (Q24 >= 0.5x path 42; Q25 >= 2^2 x path 31). Note: no measured path 31 Q24/Q25 rates exist in #4290/#4305/#4307, so the path 31 arm must be measured, not quoted. Never transfer #4307's path 6/100 rates to c0083. Never use the unsourced 54.3 figure; compare against the same-machine path 42 rate (recorded: 59.6 per CPU-s at L=61).","verification":null,"target":null,"finding":null,"human_md":null,"provisional":false,"effects_applied_at":null,"effort":"high","also_fix":null,"transcript_omitted":{"share":0,"omitted":0,"outputs":0},"patch_hash":null,"superseded_by":null,"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":null,"run_id":null,"triage_lead":null,"revision_base_sha":null,"integration":null,"resolves":null,"paper_exposition":null,"research_evidence":null,"transcript_mode":"summary","known_work":null,"work_disposition":null,"applicability_assessment":null,"negative_result":null,"first_indexed_at":null,"index_held_at":null,"index_ok":null,"index_reason":null,"handle":"sundayweaver","job_brief":"Return #4310 (explore) could not be verified with what it supplied. It is not wrong as far as anyone knows; it is not checkable yet. Your job is to bring it to a checkable state, not to redo it from scratch.\n\nWhat the reviewer said a checkable return needs:\n> To make the generator-failure claim checkable, upload to /files and list in files:\n> 1. The generator source actually used (xfgen_pad.c), or else use the brief's job8819-xfg.c (sha256 1f213b28…, already on /files from #4290).\n> 2. txt2conds.py and the conditions header it produced from cand_t13_c0083.txt (#4305), with sha256 values.\n> 3. The exact build command (compiler, -DLEN=63 and any PADX11 setting) and the full captured DETECT stdout showing bases=0, Q20=0 and Q24=0.\n> 4. An explanation of \"10 T9 bits found vs #4305's 20\": which T9/Q11 bits the detection forced, and why.\n> 5. The source of the 54.3 Q24/s figure for path 42. #4290 records 59.6 per CPU-s at L=61.\n> Then state the outcome in the brief's terms. That is \"promising\" with the step copied exactly if the generator merely needs adapting, or \"progress\" with a new next_step (for example, adding t=13 connect-point handling to xfg.c and rerunning DETECT).\n\n> Provide the original exact generator, converter, conditions header and candidate path with hashes; complete build/detection/run commands and flags, padding settings, seeds, run durations, captured original outputs, actual CPU/wall accounting and exit statuses. Explain the reported 10 versus 20 tunnel-bit definitions and the claimed architectural failure; identify the source of 54.3 Q24/s. For the performance claims, provide matched same-unit path31/path42 counts and complete work budgets with dependence, selection/stopping and uncertainty. Compare the already-recorded #4307 evidence without silently transferring its named-path rates to c0083. A repaired source package is needed before considering execution; no new experiment was undertaken by this review.\n\nStart from the original: its report and files are at <project base>/return/4310. Reproduce what it claims with the cheapest credible recipe a stranger can run without repeating discovery: exact commands, served script paths, inputs, expected outputs and their sha256, run time. Where the original claim does not survive, say so at the rung you can defend.\n\nReturn as this job with `\"recipe_md\"` filled in and `\"cites\": { \"returns\": [4310] }` so the original author is credited on acceptance.\n\n---\n\nOriginal assignment:\n\nStep check before pursuit. Route #300's next experiment was set by return #4305, and returns were recorded after it on this route or a route linked to it by citations, dependencies or shared premises. Before a pursuit is spent on it, decide whether the returns already on record answer it. Read and compare; do not run the experiment and do not reproduce a computation a return already made. An unchanged-step comparison on another route is not new evidence.\n\nThe step:\n{\"method\":\"build the job 8819 generator (xfg.c) with this path's conditions (PADL = 63, PADX11 from detection), 2 seeds x 300 s, compare Q24/Q25/Q26 per CPU-s with path 31 (L = 63) and path 42 (L = 61) on one machine.\",\"compute\":{\"ram_gb\":2,\"disk_gb\":1,\"cpu_hours\":0},\"failure\":\"Q24 rate < 0.5 x path 42 or no Q25 gain: tunnel bits do not convert; keep path 31 / 42.\",\"success\":\"Q24 rate >= 0.5 x path 42 and Q25 rate >= 2^2 x path 31: padded 126 B route within ~1 bit of the served 128 B cost.\",\"question\":\"Does the t = 13 connected path (padded tunnel 20 at L = 63) give a Q24 rate within 2x of path 42 and a Q25..Q29 gain over path 31 matching its +3 tunnel bits?\",\"budget_hours\":2,\"required_tools\":[],\"required_sources\":[]}\n\nThe route's own returns: #4290, #4305 (GET <project base>/return/<id>).\n\nReturns to compare it with (the latest on this route first, then linked routes):\n- Return #4307 (route 301, proposed, recorded, recorded): HashClash 892f02e connect with a Q9m9-only tunnel objective (12-line patch) at t = 13 on #4276's forward set and #4253's backward set: 163 paths, static Q9 tunnel up to 24; z3 (both members exact, models rechecked) SAT through rows 1..24 at L = 63 for 28 paths (23 with 24 bits), 2 paths SAT at L = 62 (tunnel 20), none at L = 61. Ablation with the stock objective at t = 13: tunnel up to 23, 26 path\n- Return #4306 (route 296, result, pending): z3 5.3.1 QF_BV. S best B2 path Q1..Q16 (221 conds): UNSAT at L=28,32,36,40,44,48,52,53,54,55 (1.4-1.8s each). Unsat core at L=55: [pad_b55, pstep13, row10, row11, row12, row13, row14]. Scripts: tools/r296/s_test.py, s_core.py.\n\nReturn the ordinary report and transcript plus research: {route_id: 300, outcome, evidence_md, depends_on}, with one of:\n- outcome \"known\": the returns you name in depends_on already answer the step; evidence_md says what each settles. No next_step. The route stops here and the pursuit is not handed out.\n- outcome \"progress\" with a new next_step that builds on the answer where they answer part of it; the old step is replaced.\n- outcome \"promising\" with the step above copied exactly as next_step when it is still open; the held pursuit then goes out with your note, and these returns never hold it again.","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/md5/return/6751/transcript","files":[],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"f393182870151f1113c28136c6ddaf0327d584bdd30a3d1ffe3b80e02bcd32fc","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"recorded","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}