Repository navigation
Research: exact persistent-work floors for the contracted complex factory - #1032
Merged
Merged
Conversation
Contributor
|
100-PR batch review at Merged to Reviewed scope: Exact joint factory gauge/cache census; forward cache gain does not improve the four-orientation maximum. 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. |
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This finite research result extends PR949's contracted PR922/924 factory gauge class to a just-in-time persistent coefficient-work policy. Complete independent cost censuses cover all39,424 magnitude placements in each of four orientations, with an exact eightfold sign quotient.
Minima per real component are46,46,48,48 for forward/inverse/inverse-transpose/transpose. The strong boundary-clear-only comparator is48 in every orientation, so cache reuse saves2 only forward/inverse and does not improve the maximum48. Earlier clear-elision savings are reported separately. The original PR924 gauge already attains the minima; no new gauge priority or kappa is claimed.
Reproduce:
python -I -O -B research/factory-persistent-work-seed23/verify.py.Expected:
PASS_FACTORY_PERSISTENT_WORK placements=39424x4 minima=46,46,48,48 controls=11 no_new_kappa.The verifier checks exact source inventory/hashes, complete histogram equality against a separate version-label oracle, the credited PR949 histogram, and144 winning mode/sign/placement streams on all nine arbitrary data/work input columns. The unchanged PR696 emitter confirms paid overwrite-COPY semantics. Local prefix peaks/depths match, eleven controls pass, and the whole certificate is compared canonically. Native, whole-provider physical/frame/precision admission and global invoices are not claimed.
Credits PR949/Ross Richardson, PR922/924/CubeJerry, PR696/Sagar Thakrar and the retained historical lineage; known-zero clear and coefficient sharing predecessors are identified separately. Prepared for @woklerx with substantial OpenAI Codex assistance, Apache-2.0. Only research/factory-persistent-work-seed23/ is added.