Repository navigation
Conversation
|
Thank you @chafreaky. This contribution is now recorded on main in the scoped reconciliation review and contributor credits. Contribution: Retired/pending indices, atomic two-cycle changes and contracting multisplits. Review and limits: Exact-head CI evidence; serial lifecycle and ordering obligations reviewed. The complete original source, proof/research notes, notices and relevant replay/CI evidence are retained through the pinned source archive. The review distinguishes fresh runs, exact-head historical CI and prior integration evidence. Closing this historical PR as reconciled; its obsolete headline does not replace the current selected bound, and draft/retracted claims retain their explicit qualifications. Reviewed head: |
Change and mathematical scope
This composes anchored multi-split recursion, balanced coarse sums, exact retired/pending frame indexes, bounded carry two-cycles, PR74 node order, late-use output routing and coordinate flags into one frozen finite conditional witness:
κ = 261356445457/5000000000000000 = 5.22712890914e-5, with bit saving 408390793141/7812500000000000.
The exact value is about 0.8494742% above the reported pinned #74 construction (
3f78d8967c009153f2df5c6a6577b6ab1d37e2ca) and 0.4583459% above the reported #82 value (3410b940aa26e5876202152dfa7c4f66451ee22c), observed 2026-10-08. These are historical comparisons. Complete pinned #74 and #79 recursive child lists are strictly excluded at the new accepted bit saving; #81/#82 comparisons use their reported values only. No foreign producer replay or current-global-frontier claim is implied.A subsequent #74 revision at
428b5c167686b42a83f1abf7003c6d17aec561eareports κ = 5.278214423139e-5, above this frozen witness. That later construction is outside this package and has not been replayed here.The result retains the inherited ordered residual compiler, all-size recursion, finite scalar overhead, routing, prime selection, exact recovery, fixed-tape and analytic transfer interfaces. It does not assert an unconditional or fully formal multiplication theorem, global optimum, or measured practical speedup. This package is stacked on #79 and preserves its complete published construction source and finite evidence at
9a58e62cf9e8954a4dde3c9b1372d733c276a6d1.Construction
Both axes use anchored point order, balanced coarse columns when
i+j<ng-1and rows otherwise, PR74 node order with natural region execution, pending next-use scoring, future unit completion, and up to three single-exchange and three two-cycle passes. h23 uses recursive split vector[4,1,1], baseline outgoing-use order and cyclic coordinate shift one. h25 uses[1,4,1], descending future-use deadlines and the pinned PR74 coordinate permutation. Code 4 splits first and middle eligible pairs with an explicit strict-contraction guard.The exact indexes preserve the prior candidate order and both pending-frame containment tests. Two-cycles preserve each affected output/carrier basis, future-use uniqueness and total cardinality; every accepted swap lowers a fixed integer heuristic score. Coordinate relabeling preserves the scalar map and physical role operations, while actual fixed-coordinate child profiles are freshly recomputed. All physical XORs, raises, copies, dirty inputs and cleanup remain charged.
Attribution
All earlier contributor, OpenAI, manuscript, license and AI-assistance notices remain attached. The 250-file published #79 construction closure, 56-file frozen R5 discovery bundle, and original #74/#78 contributor inputs remain in separate hash-checked archives.
Validation
The portable engine and PR74 node/relabel extracts are byte-identical to the audited frozen R5 source. Both parent physical words and both coordinate-relabeled words regenerate byte-for-byte. Construction
88ca39571907343a49e97f328971ec7bcd26fbfdhas source freezec5b314b4acff73f8fcd31802db2609f848ef6fcc2b63740fd985c353d994a612. The source/finite-input closure contains 580 pins, including the code, configurations, proof, tests, selected words, profiles and additive integration files.The focused gate passed in 87.835 seconds. Independent portable-package review cleared all 580 working/staged/committed pins, all 250 original #79 archive files, unchanged R5/PR74/PR78 source bytes, both regenerated words and the complete exact arithmetic. The final 23 controls passed after two proof terminology corrections and preservation-only whitespace changes. The complete expanded repository run passed all 13 verification groups, 586 tests in 84 isolated modules, and 20 patch checks, with unchanged source freeze and a byte-clean tree (1815.025 seconds). All 42 construction CI checks passed. Receipt-only follow-up
424ef44609eda377250b2985a5210cb336ade19erecords these results without changing the source freeze or numerical witness. All 42 final-head CI checks passed; this PR is ready for review and mergeable. The newindexedgroup retains every inherited verification group, formal package and workflow.See
research/indexed-cycle/PROOF.mdfor algebra and inherited scope,SOURCE.jsonfor exact inputs, andcertificates/indexed-cycle-validation.jsonfor the focused receipt. Arithmetic cutoffs are not full operational thresholds; inherited Lean checks do not formalize the complete multiplication theorem.