Skip to content

Conditional κ = 0.001661960547713947: T37 owner map on PR890's U8 word (independent cross-check of #942) - #962

Merged
Mister-Meeseeks merged 1 commit into
CrocSwap:mainfrom
GamingPuzzled:research/t37-owner-replicas
Oct 16, 2026
Merged

Mister-Meeseeks merged 1 commit into
CrocSwap:mainfrom
GamingPuzzled:research/t37-owner-replicas

Conversation

@GamingPuzzled

Copy link
Copy Markdown
Contributor

Independent, concurrent; same κ as #942 (jon314159), posted 20 min earlier. This construction was developed independently and reaches the identical κ, idle profile {48:11086,1:6} and 1024-bit charged-prefix premise; credit for first publication goes to jon314159 (#942). It is offered as an independent cross-check, not a new record.

Raises the conditional exponent from PR890's 0.001661957774647029 to 0.001661960547713947 (+2.773066918e-9). The U8 local word, its widths, the PR589 bit supplier, both moment engines and the 47-inequality outer recipe are unchanged.

Idea. κ depends only on the global paid rank profile, through the complex moment root b. Here the bit side does not bind, and finite invoice counts enter only as strict admission inequalities. PR890's fixed-family bound stops at T ≤ 22 only because of the charged 512-bit prefix (113+18T < 512). That payload is paid in three places and nowhere else: the prefix reserve, the row reserve (2*payload) and the bootstrap cutoffs (2*(4+payload)). Charging 1024 bits gives prefix 779 < 1024, row 17495 < 20161, and recomputed cutoffs. Without the T cap, PR890's own signed dual (PR795 prices) is tight at T37 with idle {48:11086, 1:6}: six rank-1 remainders give credit 5550 = 150·37 and mass ≡ 30·37 (mod 48). This is the best normalized profile at any replica count.

Construction.

  • PR890's T21, plus PR858's T16 relabelled into replicas 21..36, both rebuilt by their unchanged constructors.
  • The two call sequences are merged proportionally: key (i+1)/189 against (j+1)/144.
  • Eight frozen whole-role edits give a credit-neutral (44,3,2) → (48,1) trade: a two-step width-4 chain, then bins (20,14,111) and (27,15,0) are repacked to {8,4×6}=32 and {9,9,9,4}=31. Every move is cross-component.

All 167,832 roles in 333 calls pass the independent window/family/chart/slice audit. All eight omitted edits are rejected, and so is the sequential (non-interleaved) schedule. The 8+8+8+8+5 bank envelope passes a literal audit with eight mutations. The full PR870 controller chain runs with explicit cardinality and payload edits.

Validation.

  • Strict replay PASS_STRICT_U8_T37_PR589_COMPOSITION 0.001661960547713947. It reused a PR799 baseline that was cold-rebuilt in this session, with ancestry re-bound.
  • A fresh git archive cold replay of commit 50c6ae8 passed: PASS_STRICT_U8_T37_PR589_COMPOSITION 0.001661960547713947 (cold PR589/PR741/PR799 rebuild plus all 19 stages, macOS, Python 3.11).
  • PR890's own cold replay also passed on the same machine (macOS, Python 3.11).
python -B research/complex-t37-owners/verify.py --work /tmp/t37-cold

Credits.

Scope. This is a finite conditional composition. All-size compiler, realization, hosting, precision/tape, analytic transfer and bit H1–H5 remain inherited. The new explicit premise is the 1024-bit charged prefix.

AI disclosure: prepared with Claude Opus 5.5 (Anthropic) via Claude Code, directed by Joel Pulikkan (GamingPuzzled). No independent human review is claimed.

🤖 Generated with Claude Code

Compose PR890's T21 map with a second PR858 T16 map (replicas 21..36),
merge their calls proportionally and apply an eight-role (44,3,2)->(48,1)
trade, reaching idle {48:11086,1:6}, the relaxed optimum of PR890's own
signed-price bound once the replica cap is lifted. The charged global
prefix payload is raised from 512 to 1024 bits (prefix 779, row 17495 <
20161; bootstrap cutoffs recomputed). Gain 2.773066918e-9 over PR890.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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