[pinterbanget]

Summary

This adds research/levi-gauge-recount/. It is an independent recount of Dugongue's #4518 (Levi-adaptive gauges, scalar allowance 666 → 616), using separately written code: my #4513 segment replay, my #4514 suffix responses, and a new exhaustive check of #4518's interval theorem. #4518's NATIVE.json is vendored, SHA256-pinned and marked -text.

Claim type: an independent recount. No new κ.

Results

	forward	backward
allowance original → #4505 → #4518	158,120 → 158,020 → 157,970	158,168 → 158,068 → 158,018
worst suffix ADD+SCALE (#4510 model)	1,328	4,871 (from 4,879)

• Logical correctness. Non-scalar events are unchanged. The one changed scalar segment (4,208 → 4,212) has the identical exact linear effect on all 411 touched fields.
• Interval theorem. C + min_{L≤0≤H}[4(H−L) + 2Σ dist(s_i,[L,H])] equals an exhaustive Dijkstra search over gauge walks and add orders on 400 random instances.

Evidence

python3.12 -B scripts/research_package.py --package research/levi-gauge-recount --check-only
  PASS_PACKAGE_METADATA_AND_PINS (not a mathematical acceptance check)
python3.12 -B scripts/research_package.py --package research/levi-gauge-recount
  PASS_REGISTERED_RESEARCH_PACKAGE research/levi-gauge-recount outcome=completed   (a few seconds, stdlib only)

Head: 2b255adbf8fb8f4bac69fec5f24355f3af4ee3d4. Base: canonical main f9f85c1e.

Credit: Dugongue (#4518, building on #4502 and #4505). Prepared by Ryanson Jonathan with substantial Claude Code (Claude Opus 5.5) assistance.

🤖 Generated with Claude Code