Repository navigation
Construct a target-aligned phase repair of the block-3 two-call prefix (no new kappa) - #4710
Merged
Mister-Meeseeks merged 4 commits intoOct 22, 2026
Conversation
This was referenced Oct 22, 2026
Closed
This was referenced Oct 22, 2026
This was referenced Oct 22, 2026
This was referenced Oct 22, 2026
This was referenced Oct 22, 2026
This was referenced Oct 23, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The invertible two-call prefix in #4688 attains the hook rank bound, but on three cuts its quotient rows are misaligned with the desired target. For its literal use as the first two calls on one fresh input, the remaining target deficit is 1016, greater than twelve further calls can remove (960). This is a scoped obstruction to that beginning, not to five-in-fourteen generally.
This follow-up constructs an explicit phase repair,
P=SUS,Q=SVS,S e_x=i^popcount(x)e_x, using the same two calls and one zero scratch. All eight prefix ranks remain 160, and all joint target-prefix ranks become 224. The package proves these equalities for every natural lifted hook, gives a uniform inverse, and supplies an exact target-plus-one-call residual identity for continuation work.This is a constructive local interface and a necessary-condition repair. No completed five-in-fourteen unit, improved kappa, or selected-frontier change is claimed. Other invariant tests and the actual multi-input continuation remain open.
Validation: exact Gaussian-integer ranks and four-point pencils; predecessor calibration and seven controls; independent Gaussian-rational 8-by-8 residual identities; normal and optimized Python replay; registered Python 3.12 package replay; committed-byte and export integrity checks. A separate agent reviewed the proof. No human, Lean, or full multiplication-theorem review is claimed. The optional full-history source preflight declined this shallow checkout; source bytes and original blob equality were verified separately, and retained sources are in this branch's history.
Reproduce:
Credit: lee-phys/#4688 (
264c4b3cb13c82d171de352ecf60486dfb8d2bd4) for the unchanged rank engine, two-image prefix and controls; Jacob Sussman/#4628 (ee4fd87f064017c23e5a86fc1dcfe0a5237caed3) for the model and hook bound. Prepared by Yiyang (@Lyy-iiis) with substantial OpenAI Codex and parallel-agent assistance; Apache-2.0.Based on canonical main
b58ad9ef4d4e861b0eaa5dba41742bdc072dca51, priority revision 7. Overlap coverage and its limitations are recorded inOVERLAP.json(473 open PR bodies plus 210 recent Discussion records, not complete Discussion history).Canonical finding and continuation handoff: #1166 (reply in thread)
Hosted Ubuntu/Python 3.12 replay now passes at
b0d5192390af493d626054b060921895df3c94d2: https://github.com/CrocSwap/integer-mult-bounds/actions/runs/39824365361 . The initial 240-second limit expired after six successful cuts on the slower hosted CPU; the same unchanged exact checker passed with a 600-second bound.