Tooling: 1.50x faster P111 RUP checking with identical full replay

[mightyOil]

This opt-in tooling package replaces assignment dictionaries with signed epoch storage in Gareth Davidson's PR1097 ordered RUP checker. Seven alternating pairs on the same pinned proofs measured 1.50x checker speedup for the large c9 trace and 1.067x full fresh-process replay speedup on macOS/CPython 3.14.7. Full runs retain formula regeneration, the original PR620 encoder comparison, positive calibration, parsing, every deduction and hint, inherited controls, and complete expected-receipt equality. Filesystem caches are warm; raw measurements and implementation hashes are in BENCHMARK.json.

Status: verification tooling; no new kappa or multiplication-runtime claim. Only research/p111-rup-epochs-seed91/ is added. Frozen sources, upstream/, selected frontier and historical evidence remain unchanged.

Validation:

python3 -I -O -B research/p111-rup-epochs-seed91/verify.py

Expected: PASS, 37,044 proof additions, 1,409,475 checked hints, 272,759 differential/independent-oracle cases, 12,277 truth-table entailment checks and 12 packaging negative controls. The complete canonical PR1097 receipt remains ab65ab6dba482fb9c9741f7ff10235c972ed03052197ea1795374fd539a67d84. Relocation, equivalent gzip recompression, preloaded-module rejection and a calibrated poisoned-bytecode control also pass under isolated optimized Python. The loader executes authenticated source bytes, rather than unchecked cached imports. This finite implementation audit is not a formal proof of the checker or multiplication theorem; selected frontier and native LRAT replays were not rerun.

Base: 1b59b624d21924f8e3537cdc1ac52d3be6eda2fe. PR1097 reviewed dependency: ad0cc538d7f827461da6364c3c0e09176b10bcae; its complete source inventory and decoded proof payloads are pinned. The raw checker keeps the predecessor's parsed-proof/valid-original-literal contract and uses O(variables) dense storage. Full inherited source-order and physical/all-size/analytic limitations remain explicit.

Seed 91 assigns verification engineering. CLAIM.json records three narrow candidates, the local claim, bounded scope and refreshed overlap checks through #1186. No pinned coordination issue existed; the three discussion threads were checked. Concurrent #1184 targets emitted rational matrix auditing, not these propositional proof traces. Related #1059/#1092/#1099/#1137/#1144/#1151 are credited without combining performance claims.

Contributor: Philipp (@mightyOil), with substantial OpenAI Codex assistance and scoped AI reviews. Primary credit: Gareth Davidson (@bitplane), PR1097; Joel Pulikkan (@GamingPuzzled), PR620; the DreamingOfClouds geometry lineage; and the full predecessor notices including LRAT/DRAT-trim contributors. Apache-2.0; existing source licenses retained. No independent human review or Linux CI result is claimed.

[Mister-Meeseeks]

100-PR batch review at f699b54dbafa7409a4cd5a9741af9ad7b638870d.

Merged to main, preserving the original contributor commits and package attribution.

Reviewed scope: Epoch-array ordered RUP checker with full retained traces, semantic differential tests, entailment checks and corruption controls; no proof step is omitted.

The pinned scoped command passed; the batch ledger contains its command, input hashes, environment and log. This incorporates the research/tooling contribution and does not change the selected κ or discharge the inherited multiplication-theorem assumptions.

Batch review, attribution and evidence. Thank you for the contribution.