[flyingviper]

Change and mathematical scope

Independently audit the fixed P104 consumer cost screen and apply it separately to PR2725. With width 73, 832 original families, five PR2458 forward calls and four calls of either PR2712 or PR2725, neither profile can reach the selected target under the stated coverage contract. The rejection survives 416 integer rank-deletion units forgiven per dual, ideal fractional rank-73 packing of all remaining coverage, and every finite number of fully paid auxiliary banks.

The package reconstructs the actual source words, separates persistent coverage from private lifetimes, and proves the precise scope of the forgiveness. An abstract same-mass profile 3093 H9 + 17 H1 passes the optimistic screen through a=1/580; it is not a circuit, but demonstrates why unrestricted coarsening/resynthesis is outside the obstruction. Remaining correction/helper/residual/idle coverage is counted collectively, avoiding an idle-only double charge.

This is a conditional obstruction and source/applicability audit, not a new supplier, owner schedule, universal P104 barrier or kappa improvement. The local endpoint identities remain valid. The stopping decision is to avoid owner construction for this unchanged consumer.

Task and overlap

User-assigned p104-side-correction, refined to this complete-cost audit. Scientific base 65e840f55428829ecc75d7c21006cc13ce1268c8, priority revision 5; publication branch starts at refreshed main 7b39515117fe26e75d16c268877391f1bdaa5640. Guidance, selected record and consumed inputs are unchanged.

Jose Manuel's Discussion2619 calculation owns the original PR2712 rejection. This adds independent source/scalar replay, the explicit applicability boundary and rank-nine witness, and the separate PR2725 case. Nick Petalas's concurrent local-only all-pair claim is credited; this package stays at four pairs. PR2729 supplies separate local headroom accounting.

Pinned heads: PR2458 3170afd2fc13a990bfe28f5ab6421bcda4ffd0f6; PR2712 afa5d6649460dcdb93489babd74f0ee9d01a4911; PR2725 71d24b655c2fdddc063b9aaf285e8425ce197499. Final overlap at 19:23 UTC covered the latest 100 updated open PR bodies, relevant Discussion2619 and local notes, not complete remote pagination or unpublished work.

Validation

python3.12 -B scripts/research_package.py --package research/p104-consumer-budget-audit

Passes on CPython3.12.14/Windows, standard library only. Registered commands use -I -O -B. A separately written checker imports no primary/predecessor executable code: three complete scalar/frame words on 358 arbitrary inputs each, 256,328 inverse-transpose coefficient comparisons, 150 dirty helpers restored per word, and six semantic fault controls. Six behavioral/export tests cover contract changes, source drift, typed/duplicate receipt errors and unchanged Git-free replay. Three separate AI agents audited proof, source bindings and failure modes. All 15 published files match the tested export byte-for-byte.

Full predecessor native-interpreter proofs, actual owner schedules, all-size supplier replay, hosted CI, human review and formal-kernel verification are not claimed.

Credit

Jose Manuel, Ross Richardson, ywz65, commdata2338, Nick Petalas and the native-channel, scalar/frame, schedule and coverage lineage retain credit in the package and exact retained sources. Prepared by Vasek (flyingviper) with substantial OpenAI Codex assistance, including separate-agent review. Apache-2.0; predecessor notices retained.

Open next questions

Specify a changed operation profile, coverage law or consumer and screen its complete bill before constructing owners. Extra unpaid coordinates require a replacement coverage theorem; joint retiming or coarsening requires an actual combined word. Passing that relaxation would justify an admission experiment, not establish a supplier.