Repository navigation
Reconstruct a G20 native SOURCE from five prior values; ordinary BIT 0.003484629984 - #5647
Merged
Mister-Meeseeks merged 2 commits intoOct 26, 2026
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
A five-value native history reconstructs the missing SOURCE77 function at FRAME509. It lets helper444 replace H3+H1 by H4 while preserving every independent target and reusing the old zero centre. This is a new source reconstruction, distinct from PR5639's helper465/467 snapshots and PR5629's four retirements.
The complete ordinary BIT saving is 0.003484629984, versus the pinned PR5512 value 0.003483969976. The signed moment improves by
18*(1+3^t-4^t)for every0<t<1; the full caller has 56,979 paid requests, 11,238 productive targets and eleven owned banks. No original input is erased and all five gathers, dirty clear, injection, routes and finite fees are charged. The old zero-centre work is removed from the literal program.This is an independently usable component improvement, not a new best whole multiplication kappa. PR5641's separate four-retirement component is already numerically stronger. Combining them belongs to the integration lane; no such composition is claimed here. The selected frontier is unchanged.
Read
research/g20-affine-history/PROOF.mdandRESULT.json. Source base: canonical maindf0137efd35e11b4445734bf612bdd4c7ebc330c; priority7. All 152 source-closure files are pinned and declared. Coordination.Validation: Python3.12 standard-library registered normal and optimized replay pass. They check all 660 local RAW/DIRTY columns, exact rational H3/H1/H4 composition, full physical controls at 101 and 101 squared, the literal eighteen-call tape, an independent 11,249-column event replay, all unchanged target descriptors, two exact interval engines and fresh ordinary compiler constants. Missing-gather, missing-dirty-clear and underpriced-move controls reject. Committed-byte and export-integrity checks pass. Git-free replay is recorded in VALIDATION.md. Full-history ancestry, independent human/formal proof and hosted CI are separate.
Expected marker:
PASS_G20_AFFINE_HISTORY_ORDINARY_003484629984.Credit ai-nanachi PR5512, Ali Romman PR5421, Dugongue PR5629/PR5639, DreamingOfClouds, eumemic, Cinna Davis, rohanarun, zxv, Jacob Alan Sussman and all retained notices. Prepared by Ali Romman with substantial OpenAI Codex assistance; no subagents. New work is Apache-2.0. The retained native/all-width, ROW, compiler, owned storage, precision, fixed-tape, completed-backend and analytic assumptions remain.