Construct no-helper segmented Pauli feedback with exact coordinate ledger

[Clear1nsight]

The actual selected195/521 boundary requires C_f x+B_f y/2 while restoring donor521. This package replaces #4009's whole-register feedback with fixed signed coordinate-pair shears inside the donor. It constructs the complete boundary using one rank2f child on one V-sized target and no additional vector helper.

The exact signed-pair schedule reduces rational coordinate ADD volume (26f+1)V→(4f+1)V, reduces live vector payload3V→2V, and uses3fV unit-phase coordinate maps. The additional ADD work over the original boundary falls26fV→4fV, an exact6.5× reduction; this is an arithmetic-volume statement, not a runtime-speedup claim. Its table-free pair-addressing algorithm and inverse visits are included in the ledger. No enlarged native operand, second recursive call or precomputed mixture oracle is hidden.

The written all-f proof supplies arbitrary donor restoration, exact source/global-phase binding, addressing and scalar-workspace obligations, an exact dyadic grid bound, and a uniform outer-prefix condition bound729/4 with the native child treated atomically. Finite checks cover1024 literal boundary inputs,128 source orbit charts,168 reduced tensor inputs,64 prefix-norm controls, fixed operation counts, precision fixtures and four corrupted schedules. An independent agent and the coordinating agent reviewed the factorization, counts, precision and conditioning argument.

No new κ or sub-fV algorithm. The construction still adds4fV ADD work, which eventually exceeds the available A V [2−2^(1−α)] f^(1−α) recursive saving for each fixedA>0. Blocking/rescheduling this network improves locality options but leaves its arithmetic count unchanged. This is this schedule's explicit cost rejection, not a universal sparse lower bound. The separate masked-volume and variable-native studies have different contracts and artifacts.

Validation: registered and optimized exact replay passed; publication-byte/source-history checks and official export passed; read-only Git-free ordinary/optimized export replay passed with unchanged files. Source base/main20ecf289 and priority6 remain unchanged. Hosted exact-head CI will be checked after publication.

Limits: this is a physical coordinate-shear identity, not an admitted inherited common-frame event word. Full bit/tape routing, scalar primitive stock, approximate-child precision and global/all-size supplier admission remain obligations. No selected profile/frontier changes and no new complete cold replay. Stop expanding this network as a uniform improvement; a different algorithm must supply a sub-fV complete invoice before another construction round.

Credit GonzaloMier/#3975 for the literal candidate, #3989/#4009 for the Pauli boundary and prior feedback, CinnaDavis/#1836, AdamGall's heterogeneous interface and the selected#1094/#679 lineage. Ross Richardson's #4023 fixed-carrier query/scan result and #4027 H6 U9 splice have different boundaries; the new P104 handoff on#3954 uses #4009 full-array owner mixing and explicitly leaves sparse variants open. Relevant heads/scopes and limited overlap coverage are documented in README. No general novelty claim about elementary shears or butterflies is made.

Prepared by Clear1nsight with substantial OpenAI Codex assistance in research, proof, implementation, independent review and packaging. New work Apache-2.0; no independent human or formal-kernel certification claimed. Pending predecessors are credited without runtime dependencies.

Final hosted confirmation: run39243207483 succeeds at the unchanged published head e8c411b0b12f7bc9ed60ad71afd9bd7ade8c9834, completed11:17:36UTC. The literal/tensor/volume/control and registered-package PASS markers and unchanged-input verification are present. Feedback was empty at11:18:46UTC. The standalone export has11 total files including manifest, SHA256 72426a51ce4df45bc4b6a2887fde5c110fe3947b1ba4efdc9d80e30f50a6faad; normal/optimized read-only Git-free replay leaves every file unchanged. Separate same-team reviews are not human or proof-kernel acceptance.