Research: PR589 whole-call hosting in the exact stream model (no new kappa)

[alephnan]

Change and mathematical scope

Selected PR589 can place five helpers of one call in the same physical family, while PR540's concrete source provider covers an older word with rank-eight COPY transport. This package gives a written, source-bound whole-call hosting proof in the exact stream model, with independent exact certificates for the selected word's owner maps and nine rank-seven COPY/ERASE pairs.

The five-helper fixture uses distinct cover classes and disjoint pending supports in family 666. The argument then covers both complete 27,234-instruction programs, all 661 symbolic input columns, 636 frame projectors, 288 stage/replica descriptors, and chronological composition through the actual idle windows. Arbitrarily correlated dirty helper contents and prior outputs are retained between calls. The cover-sweep argument assumes enumeration of every element of the finite orthogonal group.

Only the new research package is added. The recursive profile, selected frontier and frozen sources are unchanged. No new kappa, PR589 Lean provider, all-size compiler, or admission of the explicit charts to the routing/finite-invoice budget is claimed. The executable finite certificates support the written proof; they do not replace those remaining obligations.

Task and overlap

Task: bit-helper-contract, priority revision 2; scope changed under the explicit assignment from semantic H2/K replacement to selected-source physical hosting. PR1194/1311 already address semantic helper preconditioning. This contribution investigates a separate retained hosting premise; TASK.json records the boundary and stopping rule.

Research base: ef5f39403610d231a0c27e8a6b4f08c10209bcd6. Selected PR589 source head: a5afc8c6169ea61ba7c51dfd3cade02cc7b13b46. Exact retained input blobs and hashes are in SOURCES.json.

Nearest predecessor	Exact inspected head	Added scope
#540	acc1917034d10f464271ab115fda314fd4eee7a3	PR589's literal rank-seven pairs and chronological composition, beyond the older PR442 concrete call
#1307	4cb255ef445e2abc3ff8845fa01210b974e2cf6c	Actual common-cover owners and whole-call execution beyond ambient-axis disjointness
#1351	6060e62e9d42a16eff493aba74c8a617a12d8041	Full calls and all windows beyond one physical-family operator audit
#1371	d47c4c300a53bb206becd13db89cea8c287c929b	Retained-memory execution using the predecessor partial-swap identity
#1421	a79ffd1571fff90f0d1dfbbed97e8ae7483e3e42	Physical owner and chronological composition arguments beyond frame/scalar accounting

The submission overlap refresh at 2026-10-14 04:14-04:15 UTC covered 733 open PRs and 259 discussion records with complete pagination. All five heads above were rechecked. Coverage includes PR/discussion text and local package metadata, not every diff or unpublished result; it is not a novelty certificate. Current upstream main 0a201997097b8a89b638f3a2afcb35db4184c8b2 retains the same pinned inputs and package workflow.

Validation

Passed on Windows CPython 3.12.2:

python -X utf8 -B scripts/research_package.py --package research/pr589-five-helper-hosting --check-only
python -X utf8 -B scripts/research_package.py --package research/pr589-five-helper-hosting
python -X utf8 -B research/e8-bit-p111-tuned/code/xcheck_p.py research/e8-bit-p111-tuned/tune/bitE8mut-p111-tuned.json.gz

• The registered independent standard-library checker passes normally and under -O; all ten malformed controls reject. Controls include coherent owner collisions, wrong supports/windows, stale MOVE frames, rank-eight substitution, wrong rank-seven frame, premature ERASE and missing cleanup.
• The predecessor checker passes all 661 columns and H1-H5 (SymPy 1.13.3).
• All 1,579 selected-source bindings pass; the Windows long-path invocation is recorded in VALIDATION.md.
• The unchanged native PR589 supplier cold replay exits zero in 377 seconds, including native legality/column checks, source reconstruction, hosted mutations and historical cited assemblies. Its WSL environment, exact command, versions and hashes are retained in NATIVE-REPLAY.json.
• Fresh versus archived proof-relevant outputs match. Geometry differs only in runtime seconds, with the corresponding raw geometry-hash reference; the narrowly normalized complete-template digest matches.
• Diff whitespace and package metadata/source-pin checks pass.

Not run: a new Lean source-provider build or the broader make frontier1094-verify, which also replays unchanged complex/composition ancestry. Native replay verifies the frozen supplier's own scope; it does not admit the new explicit charts to its cost model. No independent human review is claimed.

Credit

Substantial OpenAI Codex assistance; Apache-2.0. Credit DaysSky, Zhihao Chen, chafreaky, eumemic, DreamingOfClouds, cheesegrader, commdata2338 and CinnaDavis for the source supplier, formal framework and predecessor audits. Original files and attribution are retained.

Open next questions

Next questions and concrete first steps: instantiate the nine literal rank-seven pairs and 661-role source in PR540's Lean framework; then lower the explicit slot/port charts to the actual compiler and bind routes, inverses, exceptional cover handling and costs to the retained invoice. These are the blockers to a stronger implementation-level claim.

[Mister-Meeseeks]

Batch review at 7943acd4b430dd4286e32954237136ee9657be3b.

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

Reviewed scope: PR589 whole-call hosting equality in the specified ideal finite-group stream model: explicit owner charts, dirty-helper/private-zero simulation, rank-seven COPY transport and chronological residual completion. All-size compiler, tape, invoice and analytic admission remain separate.

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.