Repository navigation
Realize the coupled-carrier T740 packing cap: κ ≈ 0.001729777494678597 - #3148
Merged
Mister-Meeseeks merged 2 commits intoOct 17, 2026
Merged
Mister-Meeseeks merged 2 commits into
Mister-Meeseeks merged 2 commits into
Conversation
This was referenced Oct 17, 2026
Mister-Meeseeks
pushed a commit
that referenced
this pull request
Oct 17, 2026
Answers a bounded part of PR #3153's smaller-replica handoff, with PR #3148's coupled inventory fixed. In the inherited source-window model a cap-attaining packing needs one width-eight role in every capacity-eight bin, taken from a call inside that bin's window. At T20, stage-major order has 11,583 eligible roles for 13,440 such bins and replica-major order 5,940, forcing at least 1,857 and 7,500 non-tight bins. Written Hall-count proofs give 78T+297 and 375T for every positive T, also under replica permutations within complete stage or replica blocks. In the inherited packing dual each non-tight capacity-eight bin costs more than 3/250, so both orders push the complete moment above one at the cap bracket's left endpoint. The shipped interleaved T74 order passes the same necessary matching for all 49,728 bins. Conditional on the pinned inventory, window model, additive bin pricing and inherited local paid profile; arbitrary interleaved T20 orders, a full allocation and native or physical admission remain open. No new kappa. Adds research/coupled-t20-window-audit/ only. Submitted by @commdata2338 with substantial AI assistance (OpenAI Codex research and implementation; packaged with Claude Code). Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01X86sCSXdoDd61SDFfC1PKG
This was referenced Oct 18, 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.
This adds a new 145-row allocation exchange for PR3133's changed helper inventory, attaining its fixed-word packing cap at T740:
κ = 0.001729777494678597011886 − 10⁻⁴⁶ − 10⁻⁷⁰, about 1.4052734766776475 × 10⁻⁹ above PR3143's precision-40 T74 claim.
Ten interleaved copies of the literal T74 fill contain twenty
48:[4,8,9,9,9,9]and ten48:[9]non-tight bins. A deterministic exchange using twenty-five32:[9,9,9]donors replaces these with tight patterns. The 145-row preimage-bound journal changes 55 bins; the unchanged owner engine checks all 2,717,280 roles. The exact idle delta is{1:+20,3:+21,5:-25,39:-10,48:+9}. Both local words and their ordered width arrays remain byte-identical to PR3133.The contribution uses Nick Petalas's existing PR3024 lift/dual proof and PR3030 integration mechanism. The new exchange is specific to dcharb78's PR3133 inventory; the previous cap exchange cannot simply be reused. The exact cap proof checks all 2,551 patterns and derives
20*N_(48:[9,9,9,9,9]) = 1323*T. The resulting profile equals the fractional cap profile exactly.This answers the new-journal alternative in handoff #3145, following Kumazaki's stopped attempt to transport the old journals. The pre-submission refresh found that earlier work after this journal was constructed. No priority claim for the T740 objective is made; the old-journal preimage failure remains valid.
Validation: all 11 complete source/native/admission/outer stages pass; the public finite and semantic checks pass. The recorded run rebuilds both current native providers and explicitly reuses an authenticated retained parent plus two retained native providers. A fresh retained bootstrap is the public reproducer's default but was not rerun in this session. The complete physical stock is 663,549, including 91 private zero buffers, versus 66,813 at T74; this small exponent gain costs roughly ten times the fixed stock. The strict physical invoice needs 1,479 prefix bits within the paid 2,048-bit allowance. Exact source/receipt/log bindings, the generator, all-role finite check, all-pattern cap proof, semantic rejection controls and full reproduction commands are in
research/coupled-carrier-cap740/.This is a conditional finite construction and outer certificate. Inherited all-size, compiler, physical/native and analytic obligations remain; no selected record is changed and no independent human or proof-kernel review is claimed. Further packing rearrangements in this fixed-word model cannot improve κ because the cap is attained.
Credit dcharb78/PR3133 and PR3042; Marius-Paul Dumitrean/PR2979 and the relaxation lineage; Nick Petalas/PR3024 and PR3030; Jeremy Kothe, Mason Hale, Yiyang Lu, Dugongue, Tekalign, icekylinx, Jeremy Diviney and the full retained supplier lineage. Gonzalo Mier's PR3143 provides the precise comparison, and Arham Tahir's code-disjoint PR3136 audit was replayed before expansion. New exchange and integration by Jeremy Diviney with substantial OpenAI Codex assistance. Full notices and source pins are retained.