[Matarener]

Requested by Mark · project thread

Change and mathematical scope

Audit, no new κ. An independent check of #5551 (coupled P104 contacts, conditional κ .00304153, package research/p104-joint-contact on main) along the lines of its intake note. That note asks for the producer/caller and the coupled-contact delta to be reviewed together, with the native paid-map inventory, all productive targets and complete funding, in a serial bounded replay. All four layers pass:

1. Own exact replay of the Improve complete conditional κ to 0.00304153 via coupled P104 contacts #5551 and Conditional κ = 0.00304117 (+6.0e-7 over #5544) #5545 words, importing no source code. It checks strict span nesting for every MOVE, a common chart for every ADD/COPY/SCALE, all 743 endpoint rows and the owned lifetimes. Delta: the scalar program is identical up to frame labels, with exactly 6 ADD + 15 SCALE chart edits and local paid delta 2H1 − 3H2 + H3 − H4 + H5. All public endpoint spans and the helper profile are unchanged.
2. Independently verify the P104 forward caller and target bindings (no new κ) #5550 caller/target checker on Improve complete conditional κ to 0.00304153 via coupled P104 contacts #5551. Chris R's byte-pinned checker is used with four input lines changed in a temporary copy, as in Audit #5561 free helper endpoints against the #5550 P104 endpoint contract (no new κ) #5567. Positive control: it reproduces the complete Independently verify the P104 forward caller and target bindings (no new κ) #5550 output on Conditional κ = 0.00304117 (+6.0e-7 over #5544) #5545 byte for byte (b51a7d19…4009). Result on Improve complete conditional κ to 0.00304153 via coupled P104 contacts #5551: PASS. The checker recounts the invoice as 69,912 paid / 12,814 productive and runs 288 exact signed-Pauli/phase-anchor checks and 9,486 fresh-helper checks. Ordinary passes are 127,292,680 ≤ 184,647,552. All ten controls reject, and so does a changed invoice.
3. Dense native inventory of the whole word. This covers 2,940 distinct MOVE maps, 40 helper endpoint pairs and 208 intrinsic main adapters, 3,188 operators in all. Independently verify P104 contact phases and bound repeated H5 retiming (no new κ) #5548's credited exact Gaussian-dyadic interpreter compares each one on all 512 columns with absolute phase, norm and support size 2^rank. Every operator passes. Earlier dense audits covered only the maps new to each word (3 for Conditional κ = 0.00304117 (+6.0e-7 over #5544) #5545, 4 in Improve complete conditional κ to 0.00304153 via coupled P104 contacts #5551's native peer), and Independently verify the P104 forward caller and target bindings (no new κ) #5550 records paid_native_maps_replayed: false. Two separately written drivers produced the same 3,187 distinct digests. A phase-only fault with an unchanged signed tableau is rejected, and so is a flipped child sign.
4. Own funding engine. It uses 80-digit decimal ln/exp with 10⁻⁷⁰ padding and shares no code with Improve complete conditional κ to 0.00304153 via coupled P104 contacts #5551's two engines or its funding peer. Raw crossing (.00304153116714, .00304153116715), equal to the author interval. The signs at .00304117, .00304153 and .0030415305 are positive; at .00304154 the sign is negative.

Remaining obligations: the uniform compiler for all recursion levels (the finite word is checked, the uniform word is not materialized), owned erasure, precision/tape, analytic recovery, #5551's own outer conjunction and certified constant, and maintainer acceptance.

Task and overlap

Base 5e0052ca0aa3036a3afa8b0438927377c8d97d53. All inputs are canonical files on main, pinned by blob and SHA-256. Card scalar-to-moment, scope refined: this audits a changed paid itinerary that #5551 already proposed. I checked main and open PRs through #5570 and found no full-inventory check of #5551. Nearest work: #5548 (dense interpreter, 3 maps of #5545), #5550 (caller checker), #5557, #5567 (same checker patch), and #5551's own native/funding peers. #5553, #5561 and #5569 build on #5551.

Validation

python3.12 -B scripts/research_package.py --package research/p104-joint-contact-native-caller-audit

Result: PASS_REGISTERED_RESEARCH_PACKAGE ... outcome=completed, serial, about 5 min 7 s, standard library only. The default run checks layers 1, 2 and 4 completely. For layer 3 it checks a fixed subset (all 4 new maps, all 40 helper pairs, all 208 adapters and every 32nd MOVE map) whose digests must equal the committed FULL-CHECK.json. verify.py --full --jobs 4 recomputes all 3,188 digests in about 11 min on four cores. That run was performed and its result is FULL-CHECK.json. It is not part of the bounded CI command.

Credit

thu-nmrc (#5551). Chris R / Clear1nsight (#5550 checker, used unchanged except four input lines; #5548 dense interpreter, imported unchanged). Ali Romman (#5544/#5545, P104 caller). The P104 lineage and all native, Clifford, compiler and analytic contributors are credited in the cited packages. Prepared for Matarener with Claude (Anthropic), via Claude Code. Apache-2.0.

Open next questions

• Bind Independently verify the P104 forward caller and target bindings (no new κ) #5550's uniform-compiler contract to the FULL-CHECK.json digests, so the finite native check carries to every recursion level.
• Maintainer admission of Improve complete conditional κ to 0.00304153 via coupled P104 contacts #5551, after which Two successor hand-offs on the P104 word: conditional κ 0.0030428242 #5553/P104 free helper endpoints: conditional κ .0030814243 (retime of #5551's word) #5561/Conditional κ .00308181 (+3.94e-7 over #5561): priced successor hand-off of helper 706 on the P104 word #5569 rest on an admitted parent.

🤖 Generated with Claude Code

https://claude.ai/code/session_01NGucs6E49BNE3ev5xk8Afu