Skip to content

Finite bit construction: phase-B block programs on the R439 helper-star D, 11,783 -> 11,500 ADDs (no new kappa) - #1062

Merged
Mister-Meeseeks merged 1 commit into
CrocSwap:mainfrom
thakrarsagar:research/r439-bsection-stars-seed01
Oct 13, 2026
Merged

Mister-Meeseeks merged 1 commit into
CrocSwap:mainfrom
thakrarsagar:research/r439-bsection-stars-seed01

Conversation

@thakrarsagar

Copy link
Copy Markdown
Contributor

Status: finite bit-circuit construction; no new κ. The selected result, the selected bit supplier and upstream/ are unchanged; the bit side does not bind.

Track 00–19 (bit supplier), OS-random seed 1, base 2921874911956718fcaaaf9396fe8267b3cfa9f0. Only research/r439-bsection-stars-seed01/ is added.

What changed

The incorporated helper-star word (PR967, research/r439-helper-stars-seed19, 11,783 ADDs) rebuilt D only and kept PR589's phase-B blocks. The shortest recorded block programs (#1015, #923) were built on the 24k-ADD PR589 word. #1015 and the batch-952 review both say they were not composed with the D-substitution line. This PR composes them:

complete local word ADDs D F B local / nine-stage bits (caps 104/192)
helper stars (main) 11,783 5,563 72,847 10,394,084 102 / 192
+ ten spliced phase-B blocks 11,515 5,563 24,625 5,846,566 95 / 181
+ unchanged seed-19 star pass, rerun 11,500 5,548 25,717 5,911,840 95 / 182
pair control (rerun) 11,715 5,763 24,563 5,892,602 95 / 181
  1. Composition (−268 ADDs). Nine Finite bit construction: multi-write shear programs for all ten R439 phase-B blocks (24,305 -> 24,053 ADDs, no new kappa) #1015 multi-write programs plus Finite bit construction: R439 frame520 uses 29 ADDs, four fewer complete ADDs than PR844 (no new kappa) #923's frame-520 program replace all ten blocks (388 → 254 gates), and C is rebuilt by the literal-cleanup rule. Each replacement has the original F₂ block map on all 2^n inputs, an equal integer transfer and a smaller reverse transfer.
  2. Freed guard (−15 more). The helper-star word sat at bit-length 192 under the 2^192 nine-stage cap, and its seed-19 pass rejected 24 nominees on that guard. The spliced blocks lower both positive bounds. The unchanged builder, called as a library and rerun on the same PR915 parent with the same blocks, then makes 0 guard rejections and admits 236 edges instead of 221.

For both words, PR589's A, scatter, copied totals, register frame paths, nine COPY/ERASE pairs and paid profile (rank mass 5,129) are unchanged. The pinned owner checker reports the same H1–H5 pattern as the incorporated helper-star word: only its known syntactic-H2 and K_equals_D flags are false.

Reproduce

python3 -I -B research/r439-bsection-stars-seed01/verify.py
python3 -I -O -B research/r439-bsection-stars-seed01/verify.py

Expected (about 20 s each, stdlib only):

PASS_R439_BSECTION_STARS blocks=10 gates=388->254 ADDs=11783->11515->11500 guard_rejections=24->0 columns=661 controls=8 no_new_kappa

The verifier rebuilds all payloads and search receipts byte-for-byte. It replays all 661 formal columns through the literal COPY/ERASE stream in both directions for four words and checks the D dirty boundary against PR589. It runs the pinned xcheck_p.py and recomputes the profile. It bills the pinned export_records.py records, whose bounds equal the literal-stream bounds, and checks both guards. It also confirms that the original pass still gives 11,783 with 24 rejections.

Eight controls are rejected: mutated block source, dropped block gate, block gate at frame 0, stale cleanup (F₂-exact but rejected by the literal-cleanup rule), omitted cleanup gate, omitted star restoration, a 105-bit local bound, and the 236-edge D without the spliced blocks, which fails the nine-stage guard. That last control shows the extra stars depend on the composition. The pushed branch was also replayed from a clean sparse clone.

Scope

Overlap and credit

Before submission I checked open PRs through #1060. #1015 and #989 compose blocks only on the 24k PR589 word. #1040 and #1047 are single D edits. None combines phase-B blocks with the helper-star D.

Credit: block programs by Aaron Wang (#1015, open; inputs/PR1015-CANDIDATE.json copied unchanged from 7588c797) and Ross Richardson (#923, frame 520). The helper-star builder/D is from PR967 (prepared for @theon199), generalizing Gareth Davidson's #926 on the #915 target basis. The underlying word is DreamingOfClouds' #589 on eumemic's #555, with the retained checker lineage.

Prepared by Sagar Thakrar with substantial assistance from Anthropic Claude (Claude Code).

…ar D (no new kappa)

Splice PR1015's nine multi-write block programs and PR923's frame-520 program into
the incorporated PR967 helper-star word (11,783 -> 11,515 ADDs). The smaller block
transfers lower both positive bounds and clear the 192-bit nine-stage guard, so the
unchanged seed-19 star pass rerun on the same parent admits 15 more stars
(D 5,563 -> 5,548; 11,500 ADDs). Exact F2, frame-path, owner-checker and exporter
verification with eight controls.
@Mister-Meeseeks
Mister-Meeseeks merged commit 2894d40 into CrocSwap:main Oct 13, 2026
61 checks passed
@thakrarsagar
thakrarsagar deleted the research/r439-bsection-stars-seed01 branch October 13, 2026 19:00
Mister-Meeseeks pushed a commit that referenced this pull request Oct 14, 2026
…ks retained via #1062 (no new kappa)

Main retains open #1015's CANDIDATE.json verbatim (blob 16532f2) through merged #1062,
where its nine phase-B block programs supply 244 of #1062's 283 saved ADDs, but #1015 had
no ledger row, CONTRIBUTORS credit or retained NOTICE. Add a hash-checked receipt, retain
#1015's NOTICE.md byte for byte, and add one CONTRIBUTORS credit line. No ledger,
disposition or construction change; #1015 stays open.
Mister-Meeseeks pushed a commit that referenced this pull request Oct 15, 2026
Takes discussion #1795's enlarged-boundary handoff. Moving the left
boundary before helper 466's A[1991] promotion does not remove the B508
frame requirement: in the retained PR #1062 interval A[1991:2075], with
the nine COPY/scatter/ERASE observations and B[0:195], the original
source lines alone force all eleven B508 wires to contain F508. Their
sharp contribution before this right cut is 8H2 + 3H3 (H_r = r^(1-a),
rank mass 25), attained by the baseline's eleven single climbs. The
source-only rank-four conclusion holds at all 1,992 left cuts A[k],
0 <= k <= 1991 (21,912 exact wire/cut checks); a separate checker
agrees. Not a whole-interval optimum; no supplier or kappa change.

Adds research/r439-donor466-enlarged-seam/ 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
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