Repository navigation
Conversation
Tools to reorder commuting phase-A gates (identical operator) and choose legal frames on bitE8/1 words, with exact e8core replay and pricing. On CrocSwap#454's word the best schedule is 18.8 J below the word (+0.06% on c); CrocSwap#454 is complex-bound so no kappa is claimed. Also records the partial exhaustive hyperplane sweep (826/1630 hyperplanes, all infeasible). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01HEdAsVSgLyumoiWDfQCr1h
|
Thank you @chafreaky. This contribution is now retained on main in the scoped 100-PR review and contributor credits. Contribution: Target-prefix mixing with whole-residual hosting. Review and limits: The common-frame target-label argument is an explicit interface dependency. The complex supplier becomes binding; no cross-copy gain is imported. The complete original source, notices and relevant execution evidence are retained through the pinned source archive. The review distinguishes fresh local runs, source-bound historical Linux CI and prior exact-head review; it does not claim 100 fresh full mathematical audits. Closing this historical PR as reconciled. This preserves parallel and cumulative credit without replacing the selected frontier or promoting priced research targets to completed constructions. Reviewed head: |
2 participants
Conditional κ = 0.001266536148630310, 0.047745% above PR451 at
f3a38153678b55063fc005901cc968266166507d.Transfer PR450's target-prefix mixing to PR451's rank149 bit word using a frozen selection of 46 disjoint receiver/donor pairs. Exact reconstruction, all 980 local columns, integer target-source spans and an explicit common-frame bit-label argument are included. Every residual remains whole on one host family; this does not use PR448 cross-copy hosting.
The new bit endpoint is
0.001275657818793753037875. The unchanged, freshly rebuilt complex endpoint0.001268142296690675830146now binds. All 65,120 bit helper occurrences retain the existing eleven-stage schedule and 9,600 live families. The complete invoice charges the new additions, 155-bit prefix and 520-bit three-level envelope within the retained 192/1,024-bit charges and 20,161 row allowance.Validation: fresh Linux full and macOS quick runs with complex phase regeneration both passed; every exact result field and the physical instruction digest agree. Checks include four mixing controls, 92 new target operands, 47 exact frame-label matrices, 648 finite address/value controls, official C++ records, 9,340 global columns, 1,320 routes, 22 global mutations, 24 omitted bridges, both rational moment engines and all 47 strict outer inequalities. Source pins bind 288 unchanged upstream files. Receipts and the small post-run cosmetic diff are included; final assembly was rerun on both platforms. GitHub CI is separate and not yet claimed green.
Reproduce:
E8_REPLAY_PHASES=1 python -B research/e8-target-mixing-whole/verify.py /tmp/target-mix-new.The shared-frame label realization and all inherited compiler, mutable-basis, dirty-helper, idle-data hosting, precision, row-divisor and analytic contracts remain explicit. No unconditional multiplication theorem, Lean proof or κ=1 result is claimed. Larger numbers in PR450/453 additionally rely on cross-copy hosting.
Credits: DreamingOfClouds and Claude for PR450's target-mixing construction/checker; the PR451/443–445, PR440/441, DaysSky PR442, icekylinx PR433 and inherited authors. Frozen reconstruction, label checks/proof and this composition were prepared by chafreaky with OpenAI Codex assistance. Original notices and licenses are retained.