Skip to content

Conditional κ ≥ 0.00378751509478 (+0.64782% over #7717) with interleaved helper bases - #7726

Open
aliromman wants to merge 2 commits into
CrocSwap:mainfrom
aliromman:codex/p104-joint-quotient
Open

aliromman wants to merge 2 commits into
CrocSwap:mainfrom
aliromman:codex/p104-joint-quotient

Conversation

@aliromman

@aliromman aliromman commented Nov 18, 2026 •

Copy link
Copy Markdown
Contributor

This construction gives conditional κ ≥ 0.00378751509478. Its certified lower endpoint exceeds the upper endpoint of PR #7717 by 0.00002437823416 (0.64782%). That gain is 15.24% larger than the preceding gain of 0.00002115478655 reported in PR #7713. The new raw crossing lies in [0.00378751509486, 0.00378751509487]. Construction proof · Complete application receipt

Starting from the deposited PR #7713 circuit, 220 restrictions remove 230 paid returns. Twelve causal splits introduce fresh independent helpers. Fourteen paid pivots and nine hub reroutes complete the graph. Choosing different local equations at the same rank permits two independent splits at chart 288 and one at chart 306. Three counterexamples identify the exact input or output coefficients that fail under the alternate equations. Edit journal · Witnesses and counterexamples

Complete 16-stage count PR #7717 This PR
Paid recursive jobs 100,416 96,512
Productive recursive jobs 20,656 17,296
Productive rank mass less paid rank mass 1,328 1,328

The verifier debits every retired helper and prices both sides of the new bill. The resulting margin weakly dominates PR #7717 for every increasing concave rank price that vanishes at zero. The directed power-price and compiler checks establish the strict κ gain.

Verification

Run the five registered checks with CPython 3.12.

python3.12 -B scripts/research_package.py --package research/p104-interleaved-helper-bases

The forward and reverse sweeps check all 1,260 public fields. The catalogue checks cover 2,938 signed chart pairs and all 1,664 port-stage contacts. All 54 strict and four functional compiler predicates pass. The exact graph SHA-256 is ba59d5db17c40185c8608a734ced10135d8c041e432c3840e9034868781c31a2. Manifest and replay instructions

The hosted Ubuntu replay reproduced the bound and all five checked receipts at PR head 925719c8a. The planning suite ran 760 tests with 24 skips, and the historical replay passed. Scientific replay · Complete CI run

All five commands also passed from a 59-file export of scientific commit 3ba9094c5 in a fresh directory without a Git checkout. It regenerated the same graph digest. The repository byte, executable-mode and source-history preflight passed in a separate full-history clone.

One finite cost increases. The sufficient scalar-step allowance per padded cell per service rises from 2²⁹ in PR #7717 to 2³⁰ here. PR #7718 tightened the scale for the older graph. The minimum grid for this graph is open.

The application inherits the five framework assumptions for native all-column application, owned erasure, recursive compilation, precision and tape accounting, and analytic recovery. Independent scientific review and practical multiplication timings remain open. An independent reconstruction of the last three restrictions is a bounded first review task. Canonical finding · Independent-reproduction handoff

The first commit contains the complete scientific contribution. A separate commit carries the existing Clear1nsight/ToikY CI repair from #7257. Canonical main at e54523517 fails tests.test_research_current_status.CurrentCiNavigationTests because the CI guide still points at the older replay. The focused test and the historical replay both pass with this repair. All scientific files are in the first commit.

Ali Romman prepared this contribution with substantial OpenAI Codex assistance. It uses the restrictions in #7713, the circuit export in #7716, Clear1nsight’s causal split and banked relay, DreamingOfClouds’ literal splits and signed-curvature gate, SYLforge’s signed word, the scalar recodings in #7708 and Rohan Arun’s caller lineage. The physical comparison comes from #7717. Jerry Chen / CubeJerry, Douglas Colkitt and the contributors credited in the source notices retain their framework and verification credit. Credit and exact source links

aliromman and others added 2 commits November 18, 2026 13:22
Build the signed 16-call application from the published compact graph. Verify all 220 restrictions, 12 causal splits, both public operators, all chart phases and the complete bill. Retain the five framework assumptions and source attribution.
Reuse Clear1nsight's exact CrocSwap#7257 guide and historical-source snapshot repair, already validated by the hosted checks on CrocSwap#7644. The research package bytes are unchanged.

(cherry picked from commit 694667c)
(cherry picked from commit c58206c)
(cherry picked from commit 75cdd54)
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