Repository navigation
Conversation
…ed-cube word (κ 7.65146115390441e-4)
PR315's package with PR287's descent stage shape-generalized to h = 20
(m = 100, h centres of rank h-2, 3v/h scatter reads): the 480 unborrowed
source-pair mixes run at their rank-18 common delivery frame (PR244's
pattern), local delta {1:-960, 2:+480, 16:-480, 17:+960, 18:-480}, 480 calls
removed at unchanged rank mass; pins regenerated with discovery/repin.py.
+0.120% over PR315; bit side binds.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The "Verify research artifacts" workflow (make joint-dual-verify) fails on main and on every open pull request since the root Makefile and README changed (AssertionError: Makefile in joint_dual_compose.check_sources). Regenerated with scripts/experiments/pin_joint_dual_sources.py; verify_joint_dual.py and tests/test_joint_reclaim.py pass (16 tests) under Python 3.11. The joint-dual certificates only change in their recorded source-manifest hash. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> (cherry picked from commit a45747d)
|
CI note: the only failing check here was the repo-wide 'Verify research artifacts' job ( |
|
Thank you @rohanarun. This contribution is now retained on main in the scoped 100-PR review and contributor credits. Contribution: Concave-descent retiming at p=10. Review and limits: Retiming preserves the scalar word but changes frame bills; exact geometry and revised paid profiles remain checked. Disclose unrelated root CI differences. The complete original source, notices and relevant execution evidence are retained through the pinned source archive. The review distinguishes fresh local runs, source-bound historical Linux CI and prior exact-head review; it does not claim 100 fresh full mathematical audits. Closing this historical PR as reconciled. This preserves parallel and cumulative credit without replacing the selected frontier or promoting priced research targets to completed constructions. Reviewed head: |
2 participants
Conditional κ =
765146115390441/10^18= 7.65146115390441e-4, +0.120 % over #315's p = 10 word (7.64230861320245e-4). The bit supplier binds (bit root 7.65732e-4 against the centre-sharing complex supplier's 7.72714e-4).One self-contained package,
research/p10-descent/(#315's package with one added transcript stage), with its own workflow; branched frommain, no existing file changed.Full replay: PASS (strict
verify.py --output, ~4 minutes): all nine stages fresh (virtual, raw, bit with the descent stage, scalar, primes, banks, complex incl. gx.check1 and the gxcore mirror, math, finite), package integrity unchanged (127 files, manifest06dd8ddd…),PASS_IMMUTABLE_PARITY_FUSED_P10_FIVE_STAGE_BANKED_CONSTRUCTION, κ765146115390441/1000000000000000000.Change
PR244's common-delivery-frame retiming on the p = 10 word, as the concave-descent stage of #287 (
descent_transform.py, shape-generalized from h = 24 to h = 20, m = 100, h copied centres of rank h − 2 with 3v/h scatter reads). The p = 10 word has no frame retiming; local descent on Σ r·ln(m/r) finds the same pattern at this size: each of the 480 unborrowed source-pair mixes runs at its rank-18 common delivery frame, so per pair the paid children of ranks 1, 1, 16, 18 become 17, 17, 2. Local histogram delta {1: −960, 2: +480, 16: −480, 17: +960, 18: −480}; 480 local recursive calls disappear at unchanged rank mass 181,050 (five-stage calls 248,360 → 245,960); no new frames. The transform rebuilds every MOVE, checks the byte-identical scalar/COPY projection, nested chains, fixed endpoints, COPY lifetimes, nondegenerate endpoint bases, both reflected ledgers and the exact integer source span of every non-target operand;raw_ledger.rebind_descentrecounts the profile. Literal stock 663,030 → 660,630; all changed counts re-pinned inword-pins.jsonby #315'sdiscovery/repin.py;verify.pyruns strict.Verify
Scope and credits
Exactly #315's hypotheses and conditional interfaces; no Lean build. DreamingOfClouds (#315: the p = 10 word, package and complex supplier; #285/#276), eumemic (#244 retiming pattern, #234 lineage), hcg890, Jacob Sussman, Evan McKinney and the retained lineage. The remaining gen5-lineage stages (squares, kernels, restorations, sinks, reorder, constructed-frame descent) are not yet re-derived at p = 10. Prepared by Rohan Arun with Anthropic Claude assistance; Apache-2.0;
NOTICE.md.🤖 Generated with Claude Code