Skip to content

Scoped negative result: X4 target-forest floor holds for all dyadic parent coefficients (no new kappa) - #961

Merged
Mister-Meeseeks merged 1 commit into
CrocSwap:mainfrom
woklerx:research/complex-x4-dyadic-forest-seed31
Oct 13, 2026
Merged

Mister-Meeseeks merged 1 commit into
CrocSwap:mainfrom
woklerx:research/complex-x4-dyadic-forest-seed31

Conversation

@woklerx

@woklerx woklerx commented Oct 13, 2026

Copy link
Copy Markdown
Contributor

The pinned PR799 X4 target-forest result (#932) establishes a 10,883-primitives-per-real-component optimum only for signed-unit parent coefficients. This research note extends that exact floor to every nonzero dyadic parent coefficient under the same independent binary-Horner lowering and leaf-before-parent forest schedule. No new kappa or supplier integration is claimed.

Two elementary tail bounds reduce all dyadics to 1,086 coefficients. Two separately implemented exact checkers cover 13,501,152 ordered coefficient/target-pair cases. 970 nonunit edges beat their original root, but every nonunit edge is strictly dominated by a root or signed-unit edge to the same parent. A forest exchange argument transfers the 10,883 floor; independent Prim/Kruskal and 12,544 cycle inequalities recheck the attaining signed-unit graph.

Reproduce offline, standard-library Python 3.10+:

python -I -O -B research/complex-x4-dyadic-forest-seed31/verify.py
python -I -O -B research/complex-x4-dyadic-forest-seed31/test_package.py

The full checker passes with CPython 3.11.0 and 3.14.5; nine package tests pass. The original PR932 full scalar witness and selected-record verifier also pass. Selected-result tests pass; two existing symlink tests skip in the Windows sandbox. Full repository/cold-supplier suites and hosted CI were not run because no existing supplier or generator changes.

Base: 018418f; source/dependency pins, complete coefficient certificate, proof, local sprint claim (seed 31), controls and scope are in the new package. Credit Ross Richardson #932, ToikY #909, the restored-pair/forest predecessors and Jeremy Kothe/Chafik Boukhalfa/icekylinx supplier lineage. Substantial OpenAI Codex assistance, including a separate checker implementation and mathematical review, is disclosed. No independent human review or full multiplication-theorem formalization is asserted.

Multiple parents, helper-basis changes, arbitrary sequential shears, shared coefficient work and other lowerings remain outside scope. The selected frontier and upstream/ remain unchanged.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants