[bbMisk]

Change and mathematical scope

A batch with five separate centered dyadic BIT outputs normally restores each output's address layout independently. Allowing the common source to be temporarily permuted lets those outputs share one query tour: 40 to 13 complete ordinary ADDRESS children, 240 to 78 affine events, and the same 30 XOR scans, with no extra full row. The comparison already uses PR953's optimized separate-target batches.

This is a scoped recursive-interface construction; no new kappa. Every output, source and dirty spectator has the correct endpoint. The changed permission is essential: the source must be exclusively writable and unobserved until batch exit. Thirteen queries are proved minimal only in the declared source-query/output-XOR alphabet. Each query has its full seven-event lowering and complete coefficient-specific invoice; the new query histogram is componentwise dominated by the baseline.

The proof and typed recurrence retain role-dependent child/affine prices and all preparation, source access, routing, dispatch, temporary storage and restoration costs. No physical supplier caller, DATA-to-ADDRESS recursive compiler, all-size/tape/hosting admission, runtime saving or analytic transfer is established. The selected record is unchanged.

Task and overlap

Base: ef5f39403610d231a0c27e8a6b4f08c10209bcd6. Contribution: 04379bb9bf8d394891df9b48edbfa79882aedae2. OS assignment seed 55, recursive interfaces; refined persistent-boundary scope. Task card · source pins · claim #1493.

The new object has distinct output ports and a temporarily mutable shared source. PR953 at 5f052969846b0882ac4e22d9e818dd8ac05ea56f batches several terms on one target with an unchanged source; PR840 at da2cd0dd2229e08b61b174a8835a2c2bb99d2230 provides the signed dyadic lowering and pair construction. Their results are credited and reused. PR1235 is related source-tour reasoning in a different phase/rank model.

The final overlap snapshot covers 759 open PRs and 193 discussion/comment records, complete pagination and no warnings. Nearby Discussions1501 and1505 explicitly distinguish their additive-connector and port-borrowing scopes. This is bounded overlap evidence, not a novelty certificate. Only this new research directory is added.

Validation

python3.12 scripts/research_package.py --package research/shared-dyadic-source

Expected: PASS_SHARED_DYADIC_SOURCE children=40->13 affine=240->78 XOR=30 patterns=768 no_new_kappa, seven passing package tests, then PASS_REGISTERED_RESEARCH_PACKAGE.

• All 768 declared small incidence patterns; separate Laurent and cyclic dirty-state/inverse checks; 25,088 cyclic dirty columns per implementation.
• All thirteen native powers: 104 rational columns, 624 modular columns and 850,408 complete six-field ADDRESS/spectator inputs. Both complete array words pass 20,412 dirty columns per implementation.
• 24 mathematical/admission controls, including every native event omission and a collapsed-output word whose aggregate is correct while the separate outputs are wrong.
• Python 3.12.2 registered replay, seven package tests and all 20 onboarding tests pass on macOS arm64. Optimized isolated Python 3.14.6 source export without Git reproduces the exact receipt byte-for-byte, keeps read-only files unchanged, and rejects an actual corrupted evidence file. Source/certificate corruption and missing-child controls pass.
• make selected-record-check, source-pin metadata and staged whitespace checks pass. Refreshed main and all ten inputs remain identical. Validation record.

Finite tests support the separate written proof; they are not formal verification of the multiplication theorem. Selected cold replay, physical tape/compiler execution, hosted CI acceptance and independent human review were not performed.

Credit

Hanson Sun (@bbMisk), with substantial OpenAI Codex assistance in derivation, implementation, checks and prose; Apache-2.0. Credit Ross Richardson/PR953, Wentao Li/PR840, Karl Appel/PR521, commdata2338/PR677, their retained eumemic/Ali Romman lineage, Richard Hoekstra/PR1235, and OpenAI's framework maintained by Douglas Colkitt. Original notices and licenses are source-pinned inputs. No priority is claimed for the native lowering, single-target tour or generic source-gauge idea.

Open next questions

The package gives two bounded follow-ups: find an actual caller with no intermediate source observation and price its complete supplier invoice; or study a small formal Laurent instance permitting queries on both source and outputs. A concrete caller/compiler is the blocker to adoption, and the current complex supplier remains binding.

[Mister-Meeseeks]

Batch review at 04379bb9bf8d394891df9b48edbfa79882aedae2.

Merged to main, preserving the original contributor commits and package attribution.

Reviewed scope: Distinct-output dyadic batches share a writable restored source with 13 ADDRESS children instead of 40 in the explicit fixture. Uniform Laurent argument and full native finite maps pass; source permission, physical routing and DATA-to-ADDRESS compilation remain separate obligations.

The pinned scoped command passed; the batch ledger contains its command, input hashes, environment and log. This incorporates the research/tooling contribution and does not change the selected κ or discharge the inherited multiplication-theorem assumptions.

Batch review, attribution and evidence. Thank you for the contribution.