Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions research/p104-a2-cleanup-composition/.gitattributes
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
inputs/** -text
20 changes: 20 additions & 0 deletions research/p104-a2-cleanup-composition/CONCURRENT-SOURCE.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,20 @@
[
{
"commit": "7916490124ac946704339a6a332bb76f06e5db5c",
"path": "research/p104-a2-neutral-cleanup/WORD.json.gz",
"local": "inputs/pr5650/WORD.json.gz",
"sha256": "e94809cecf9e9cba56dd7a8a11ff1edeedc4c34013ff6caebdaf3b021351cf8c"
},
{
"commit": "7916490124ac946704339a6a332bb76f06e5db5c",
"path": "research/p104-a2-neutral-cleanup/FULL-RESULT.json",
"local": "inputs/pr5650/FULL-RESULT.json",
"sha256": "89c5b2bae30e73c626dfdfee3aebd1e8033d1f6d33688ba9d85ea502ef37c018"
},
{
"commit": "7916490124ac946704339a6a332bb76f06e5db5c",
"path": "research/p104-a2-neutral-cleanup/NOTICE",
"local": "inputs/pr5650/NOTICE",
"sha256": "f7cdc54036e36f7b3409d988bcfea371627923bfca5f2fc2c6c7c8c1c7b65331"
}
]
6 changes: 6 additions & 0 deletions research/p104-a2-cleanup-composition/NOTICE
Original file line number Diff line number Diff line change
@@ -0,0 +1,6 @@
P104 A2 and helper-cleanup composition.
New composition and checks: Ali Romman with substantial OpenAI Codex assistance, 2026.
Apache-2.0.
The paid A2 witness is by zxv, with the PR3998/4019/4112/4152 lineage.
The helper cleanup and common PR5627 word are by Ali Romman with their retained source authors.
All source files and notices in inputs/ are retained unchanged. See README.md and SOURCE-CONTEXT.json.
18 changes: 18 additions & 0 deletions research/p104-a2-cleanup-composition/PRICING-EDITS.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,18 @@
[
[
"Q('.00310626335'), Q('.003106263355')",
"Q('.00310644469'), Q('.003106444692')"
],
[
"PREVIOUS = Q('.00310440546')",
"PREVIOUS = Q('.00310626335')"
],
[
"'kappa_decimal':'.00310626335'",
"'kappa_decimal':'.00310644469'"
],
[
"previous_open_pr5627",
"previous_open_pr5644"
]
]
162 changes: 162 additions & 0 deletions research/p104-a2-cleanup-composition/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,162 @@
# Corroborating replay of the concurrent P104 composition

## Concurrent result and contribution status

PR5650 was published while the final checks for this package were running.
The next upstream refresh found the overlap. The two emitted programs differ
only in their schema labels. Every instruction, frame, boundary, native map,
physical caller and complete invoice agrees exactly. The pinned comparison
is in CONCURRENT-SOURCE.json and compare.py. Both constructions were prepared
from PR5643/5644 before this cross-comparison.

This package is retained as a separate construction/replay and exact artifact
comparison. It is not a new mechanism beyond PR5650. The 10^-11 difference in
the chosen claim is not counted as a research improvement. This concurrent
result is excluded from the donor's ten-result quota. Shared frozen check
engines are reused; this is not an independent implementation of every engine,
independent human scientific review or scientific admission of PR5650.

The normal and optimized full replays and the Git-free full replay were
completed at bdf59f7af8049ef226e3c68ee730c2ac12d8a421. The reconciliation adds
only frozen comparison inputs, a comparison checker and explanatory prose.
The scientific constructor, verifier, pricing engine, emitted word and receipt
are unchanged. The comparison runs in normal and optimized mode and includes
a missing-event rejection control. The revised committed export is checked.


This construction supports conditional **kappa = 0.00310644469**. It composes
the paid A2 path of PR5643 with the three helper cleanups of PR5644 in one
literal word. The gain over PR5644 is **0.00000018134**. The raw crossing is
strictly between **0.00310644469581 and 0.00310644469582**. This is a complete
conditional construction under the retained five framework assumptions and
helper endpoint contract. It is not a selected-frontier update or independent
scientific acceptance.

## Exact sources and overlap

- PR5643, zxv: `a3ff5c628bfc8825553d62d913f3e91f75d6825f`.
- PR5644, Ali Romman: `c5871955d89bf73506fd68b0a5bf1cf2b3153c02`.
- Their common source PR5627: `ed4dc88a09a7d201971f6603b8c2074e95dd9c32`.
- Canonical main base: `df0137efd35e11b4445734bf612bdd4c7ebc330c`, priority 7,
selected conditional kappa 0.00310045.

SOURCE-CONTEXT.json authenticates every retained source file before imports.
The PR5644 loader authenticates the complete common PR5627 export and each
of its payloads before use. Frozen files and notices stay unchanged. This
package reconstructs both published source words exactly. It then applies
the four published changes in both orders. Both orders produce the same
literal program, not only equal scalar endpoints.

The live overlap check read the complete open PR/Discussion cache and the
current claims. New internal scalar identities, helper relations, MOVE
scheduling, paid histories, boundary gauges and input quotients remain
outside this scope. This is a four-change integration with a full new bill.
No scope held by another investigation is taken.

## Composition proof

The three PR5644 prefix identities restore helpers 291,321,369 to their
original independent input rows at their final charts. Each deletes only
later writes to that helper after its last read. The input preparation,
cut cancellation, dyadic scaling and exact prefix relation are retained.
They remove contacts on main roles 55,49,80, whose H1+H5 paths become H6.
No helper input is identified, erased or removed from the target invoice.
The complete proof and exact witnesses are frozen under inputs/pr5644.

The PR5643 scalar pair uses A=715, B=388, C=340. Its old operations are
B -= 4A and C += B at chart 825. Its replacement is C -= 4A at 396,
C += B at 825, then B -= 4A at 127. Both give (A,B-4A,C+B-4A) on arbitrary
inputs. The actual nested chart chain is 598<396<825<127, of dimensions
1,3,6,9. Chart 598 has basis [491]. This changes A's two paid ranks from
5,3 to 2,6. The extra ordinary ADD is included in the complete cost.

The constructor uses unique literal source gates and exact helper anchors.
In the reverse application order, the helper cut is relocated by its unique
anchor and the entire prefix row is checked again. It does not transfer an
old row identity merely by moving its index. All transports are reconstructed
from the scalar contacts, with strict public/owned nesting and source-frame
checks. The resulting two complete programs agree byte for byte after the
same deterministic transport reconstruction. All productive boundaries and
the chronological owned-instruction sequence agree with the common source.

The combined local paid change from PR5627 is

```
-3 H1 + H2 - H3 -4 H5 +4 H6.
```

Every productive target is unchanged. The local paid profile is
`{1:2197,2:457,3:493,4:91,5:159,6:22,7:15}`.
The full caller executes the local word 18 times. It retains all 15 thin
query sites, all 3,120 idle jobs and all 12,958 productive targets. It pays
66,492 calls, with rank-mass surplus 760. The gain over PR5644 is the exact
moment change 18*(3^t+5^t-2^t-6^t)>0 for 0<t<1, by strict concavity.
The reported kappa comes from the complete invoice, not from adding the
predecessor kappa gains.

## Complete cost and ceiling

Two rational interval engines check the complete moment. The derivative
certificate proves a unique raw crossing on [0,0.0034] for this fixed invoice.
The retained finite strictification uses power 1024, cutoff 17, target floor 18,
tail offset 1/5, reserve 10^-12 and decay 39/40. The test saving is
0.003106444692. Compiler thresholds, their preceding integers, shifted exits,
complement capacity, initial range and both completion tails are charged.
All 54 strict and four functional outer conditions must pass.

The ordinary scalar allowance is 220,220 per local word. Native normalizers,
main adapters, owned initialization/COPY/CLEAR, routes, queries and idles are
included. The source raw ordinary budget remains 10^9. The BIT contract
remains 0.0034. The source coefficient and native rank bounds supply the same
explicit precision envelope. PRICING-EDITS.json lists every change from the
frozen PR5644 pricing engine: only the claim, test point, previous comparison
and output label change. No predicate or cost charge is removed.

The claim is within 5.82*10^-12 of the proved raw crossing for this invoice.
More digits do not provide a meaningful new mechanism. Further substantial
gains must change a physical paid path, target or complete caller.

## Verification

```
python3.12 -B scripts/research_package.py --package research/p104-a2-cleanup-composition
```

Expected: PASS_P104_A2_CLEANUP_KAPPA_00310644469 and
PASS_REGISTERED_RESEARCH_PACKAGE. Python 3.12 standard library suffices.
The quick mode checks both source words, both application orders, all scalar
rows and all frame/owned conditions. Full replay adds all 752 dense transpose
columns, all distinct signed native maps with exact global phases, the full
physical caller, both moment engines, compiler and outer predicates. Controls
reject missing A2 correction, missing helper preparation, missing owned clear,
unpaid query and extra target credit, plus the retained caller controls.
RESULT.json records the exact counts and WORD.json.gz stores the common word.

Normal and optimized registered replay, committed-byte checks, export
validation and an isolated Git-free replay are separate publication checks.
Full-history ancestry preflight is unavailable in the shallow checkout.
No formal kernel or independent human scientific review is claimed. The
huge uniform word and complete rounded multiplier are not materialized.
Native realization, owned erasure, uniform compiler, precision/fixed-tape
execution and analytic recovery retain the source assumptions.

## Open next questions

The new word is available for independent scientific review. Check the exact
anchor binding and both composition orders first, then the complete native,
caller and cost receipt. A changed scalar motif or boundary gauge requires
a new source-bound construction and live overlap check; the active searches
are not reserved by this result. The fixed-invoice ceiling gives no upper
bound on those changed constructions.

## Credit

zxv supplies the actual paid A2 rank exchange. Chris R/Clear1nsight, ToikY and
the PR3998/4019/4112/4152 lineage retain credit for the A2 identity and boundary
discipline. Ali Romman supplies PR5644 and the common PR5627 construction.
Retain flyingviper's relays, Scratch-net's preparation, Pierre-Loic Doulcet's
commuting cascade, GamingPuzzled, DreamingOfClouds, thu-nmrc, David Arias,
Gareth Davidson, Cinna Davis, Jacob Alan Sussman, Cao Yifan, zxv, and every
native/geometry/BIT/compiler/analytic notice in the authenticated sources.
New composition and checks: Ali Romman with substantial OpenAI Codex
assistance. No subagents were used. Apache-2.0.
Loading
Loading