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
36 changes: 36 additions & 0 deletions research/pr1407-complex-moment-audit-seed39/CLAIM.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,36 @@
{
"run": 4,
"assignment_seed": 39,
"track": "complex supplier",
"base_commit": "bce54fb3f895cbe5cd4c0eb477c3afeed100190f",
"started_utc": "2026-10-14T00:18:11Z",
"planned_finish_utc": "2026-10-14T01:08:11Z",
"verification_start_utc": "2026-10-14T00:55:41Z",
"status": "completed",
"candidate_tasks": [
{
"scope": "PR1094 neutral-then-strict two-bin whole-helper path search.",
"overlap": "Dropped after live refresh found PR1407 covering neutral whole-helper hosting paths."
},
{
"scope": "PR1387 operator-preserving subset of simultaneous idle-axis optimum.",
"overlap": "Open PR #1387 and completed discussion #1339 already cover this optimum and counterexample."
},
{
"scope": "Independent arithmetic audit of PR1407 complex profile and kappa receipts.",
"overlap": "Selected as explicitly complementary second audit; no separate arithmetic-audit claim found at refresh."
}
],
"object": {
"pr": 1407,
"head": "d8640f5b7ae99b43f5d56146cb94ca62b8b3835d",
"candidate_profile_delta": "+e1 - 7e7 + 3e16"
},
"method": "Third code-disjoint exact interval implementation using binary-reduced atanh log bounds and positive Taylor exponential with geometric tail; verify primary and conservative complex endpoint signs, exact kappa grid, rank mass/deficit, unchanged bit endpoint and failure controls. Does not replay the construction.",
"expected_deliverable": "Small registered research package and one focused independent-verification PR if the final replay remains green.",
"finish_time_utc": "2026-10-14T01:08:11Z",
"coordination": "No pinned sprint issue exists; local claim recorded. PR #1407 itself has two exact moment engines and pending cross-vendor/fresh-context review; this audit uses an independent third implementation.",
"completed_utc": "2026-10-14T00:42:32Z",
"result": "The exact Fraction replay certified the selected and PR1407 primary and conservative complex root endpoint signs, profile delta and conservation, unchanged bit endpoint, and primary/conservative kappa grids; four deliberate mutation controls rejected.",
"reproduction": "python3.12 scripts/research_package.py --package research/pr1407-complex-moment-audit-seed39"
}
108 changes: 108 additions & 0 deletions research/pr1407-complex-moment-audit-seed39/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,108 @@
# Independent arithmetic audit of PR #1407's complex profile

**Status: completed, scoped independent verification. No new construction or κ claim.**

PR #1407 proposes a neutral whole-helper hosting construction for the selected
PR1094 owner map. I do not repeat its search or replay its owner paths. This
package independently checks the resulting profile arithmetic with a third,
code-disjoint exact interval engine: the primary and conservative complex root
brackets, the unchanged bit ceiling, and the published κ grid. The upstream PR
reports two exact moment engines; this audit adds a separate rational
atanh/Taylor implementation while leaving its construction proof and native
admission as inherited claims.

## Frozen data and exact result

The research/source base is
`bce54fb3f895cbe5cd4c0eb477c3afeed100190f`. Before publication, `main`
advanced to `410026dc5513b87db92ae98571d1d5984c69a983` through three
reconciliation commits. The selected certificate and the seven other pinned repository
inputs/provenance files have identical Git blobs at both commits; the audit
branch is based on the refreshed `main`, while `TASK.json` and `SOURCES.json`
retain the exact research base. The selected record binds at
`κ = 836330098538459/500000000000000000`, with complex endpoint
`67018507072324140917/40000000000000000000000` and bit endpoint
`1733427039589844234749/10^24`. The source construction and maintainer review
are pinned by `SOURCES.json`.

The external candidate snapshot is PR #1407 at exact head
`d8640f5b7ae99b43f5d56146cb94ca62b8b3835d`. `REFERENCE.json` contains the
candidate profile, endpoint receipts and the original SHA256 and Git blob
identifiers for PR #1407's `PROFILE.json`, `RESULT.json` and `ASSEMBLY.json`.
I fetched that exact PR commit through the upstream pull ref and independently
recomputed all three Git blob IDs and content SHA256 values; all six matched
the recorded pins. Its snapshot hash is also pinned in `SOURCES.json`.

The candidate profile changes from the selected
`{1:349604, 2:60192, 3:62585, 4:20448, 5:11808, 7:7, 8:1152, 48:4866}`
to
`{1:349605, 2:60192, 3:62585, 4:20448, 5:11808, 8:1152, 16:3, 48:4866}`.
The exact delta is `+e1 - 7e7 + 3e16`: rank mass and deficit stay fixed,
while the positive-rank call count drops by three. The priced parameters are
`m=73`, `W=14336`; the primary complex moment has no added surcharge. The
conservative check includes the inherited term
`32*m*N/(10^16*W) * m^a`.

| profile and moment | lower endpoint (`M(a)-1` scaled by `10^32`) | upper endpoint (`M(a)-1` scaled by `10^32`) |
|---|---:|---:|
| selected primary | `[-130421533, -130421532]` accepted | `[162472375, 162472376]` rejected |
| PR #1407 primary | `[-159548291, -159548290]` accepted | `[133342011, 133342012]` rejected |
| selected conservative | `[-281245180, -281245179]` accepted | `[11648729, 11648730]` rejected |
| PR #1407 conservative | `[-171561767, -171561766]` accepted | `[121328535, 121328536]` rejected |

Each displayed integer interval is divided by `10^32`. The exact brackets
reproduce the receipts. The candidate's primary and conservative complex
brackets both move above the selected brackets, and both remain below the
unchanged bit endpoint. Exact flooring of `a/(1+a)` to the repository's
`10^-18` κ grid reproduces the candidate and conservative κ values. The
primary κ difference is exactly `161104813/7812500000000000`.

## Independent interval method

`verify.py` uses only Python's standard library and `Fraction` arithmetic. It
computes logarithms after binary range reduction with 48 positive atanh terms
and the explicit geometric tail
`2*z^(2N+1)/((2N+1)*(1-z^2))`, where `0 <= z <= 1/3`. It encloses each
positive exponential with 32 Taylor terms and a geometric bound on the
remaining tail. The verifier does not import either moment engine used by
PR #1407 or the selected assembly.

Failure controls mutate one rank-16 count, swap accepted and rejected endpoint
labels, change κ by one grid unit, and omit the conservative surcharge. Every
mutation is rejected. The saved exact replay is [RESULT.json](RESULT.json).

Reproduce from the repository root with Python 3.12:

```sh
python3.12 scripts/research_package.py --package research/pr1407-complex-moment-audit-seed39
```

Expected: `PASS_REGISTERED_RESEARCH_PACKAGE research/pr1407-complex-moment-audit-seed39 outcome=completed`.

## Scope and limits

This confirms the finite arithmetic receipts derived from the two histograms.
It does not verify PR #1407's 24 owner moves, its all-input dirty-state
argument, strict source-event windows, native T16 execution, controller,
invoice, all 47 outer inequalities, or full ancestry. The audit does not
establish that the published reassignment realizes the candidate histogram.
PR review and hosted-workflow status are mutable. This package pins the
candidate content at the exact head above and makes no claim that its
construction, CI or review gates are approved.

## Open next questions

The next useful step is a fresh, independent replay of PR #1407's native path
and a role-by-role check of its eleven neutral-then-concentrating paths. That
matters because profile arithmetic alone cannot establish physical admission.
The open blocker is the draft PR's pending review and workflow approval.

## Credit

PAI (@pai-scaffolde) supplies PR #1407's candidate profile and receipts. Rohan
Arun's PR #1094 supplies the selected owner map; Jeremy Kothe and Hao Yan / FanPUYen
supply the retained packing and controller lineage. The selected construction
and its maintainer review remain the pinned baseline. Prepared by Cinna Davis
(@1337isnot1337) with substantial OpenAI Codex assistance in review, exact
arithmetic design, implementation and writing. This is a code-disjoint audit,
not independent human review. New material is Apache-2.0.
91 changes: 91 additions & 0 deletions research/pr1407-complex-moment-audit-seed39/REFERENCE.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,91 @@
{
"schema_version": 1,
"source": {
"repository": "CrocSwap/integer-mult-bounds",
"pr": 1407,
"head": "d8640f5b7ae99b43f5d56146cb94ca62b8b3835d",
"files": [
{
"name": "profile",
"path": "research/astra-ultra-hosting-20261013/data/PROFILE.json",
"git_blob": "54c3e1076c6cccbe6c6a90b39b2aba7b800762e6",
"sha256": "e3f778d24ce02c1ddc21b4632c3e0d99e064d8db359b03b019fd8b37baf815b5"
},
{
"name": "result",
"path": "research/astra-ultra-hosting-20261013/RESULT.json",
"git_blob": "6505aafdcafeb1bcd37aa2ccbf0bd9c5eeab99e0",
"sha256": "6be31e95e41bcc83577b364b01a0339d16a8f6f09277b9d8d964c0ea4c1f158c"
},
{
"name": "assembly",
"path": "research/astra-ultra-hosting-20261013/receipts/ASSEMBLY.json",
"git_blob": "997cce16c3362b27250c3d0de0a29192c9dd4376",
"sha256": "fe9dd283cdc5034b74fcca6d2f344da08a876ef1fde6a4db35626a680a12183f"
}
]
},
"candidate_profile": {
"1": 349605,
"2": 60192,
"3": 62585,
"4": 20448,
"5": 11808,
"8": 1152,
"16": 3,
"48": 4866
},
"result": {
"kappa": "836340409246491/500000000000000000",
"kappa_decimal": "0.001672680818492982",
"baseline_kappa": "836330098538459/500000000000000000",
"delta": "161104813/7812500000000000",
"complex_lower": "837741683691654218469/500000000000000000000000",
"complex_decimal": "0.001675483367383308436938",
"bit_lower": "1733427039589844234749/1000000000000000000000000",
"conservative_complex_lower": "837741682260907314779/500000000000000000000000",
"conservative_kappa": "1672680815641053/1000000000000000000",
"conservative_kappa_decimal": "0.001672680815641053",
"profile": {
"1": 349605,
"2": 60192,
"3": 62585,
"4": 20448,
"5": 11808,
"8": 1152,
"16": 3,
"48": 4866
},
"live": 14336,
"rank_mass": 1041408,
"deficit": 5120
},
"complex_moments": [
{
"surcharge": false,
"endpoint": "837741683691654218469/500000000000000000000000",
"accept": true
},
{
"surcharge": false,
"endpoint": "1675483367383308436939/1000000000000000000000000",
"accept": false
}
],
"conservative_complex_moments": [
{
"surcharge": true,
"endpoint": "837741682260907314779/500000000000000000000000",
"accept": true
},
{
"surcharge": true,
"endpoint": "1675483364521814629559/1000000000000000000000000",
"accept": false
}
],
"surcharge_model": {
"complex": false,
"bit": true
}
}
101 changes: 101 additions & 0 deletions research/pr1407-complex-moment-audit-seed39/RESULT.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,101 @@
{
"base_commit": "bce54fb3f895cbe5cd4c0eb477c3afeed100190f",
"base_conservative_bracket": [
{
"accept": true,
"endpoint": "1675462673946628390003/1000000000000000000000000",
"moment_minus_one_interval_scaled_1e32": [
"-281245180",
"-281245179"
]
},
{
"accept": false,
"endpoint": "418865668486657097501/250000000000000000000000",
"moment_minus_one_interval_scaled_1e32": [
"11648729",
"11648730"
]
}
],
"base_primary_bracket": [
{
"accept": true,
"endpoint": "67018507072324140917/40000000000000000000000",
"moment_minus_one_interval_scaled_1e32": [
"-130421533",
"-130421532"
]
},
{
"accept": false,
"endpoint": "837731338404051761463/500000000000000000000000",
"moment_minus_one_interval_scaled_1e32": [
"162472375",
"162472376"
]
}
],
"bit_endpoint_unchanged": "1733427039589844234749/1000000000000000000000000",
"call_delta": -3,
"candidate_conservative_bracket": [
{
"accept": true,
"endpoint": "837741682260907314779/500000000000000000000000",
"moment_minus_one_interval_scaled_1e32": [
"-171561767",
"-171561766"
]
},
{
"accept": false,
"endpoint": "1675483364521814629559/1000000000000000000000000",
"moment_minus_one_interval_scaled_1e32": [
"121328535",
"121328536"
]
}
],
"candidate_primary_bracket": [
{
"accept": true,
"endpoint": "837741683691654218469/500000000000000000000000",
"moment_minus_one_interval_scaled_1e32": [
"-159548291",
"-159548290"
]
},
{
"accept": false,
"endpoint": "1675483367383308436939/1000000000000000000000000",
"moment_minus_one_interval_scaled_1e32": [
"133342011",
"133342012"
]
}
],
"conservative_kappa_grid": "1672680815641053/1000000000000000000",
"construction_scope": "Arithmetic only; no owner/window/path, native compiler, controller, all-47-outer-constraint or full ancestry replay.",
"exact_kappa_gain": "161104813/7812500000000000",
"external_head": "d8640f5b7ae99b43f5d56146cb94ca62b8b3835d",
"external_source_sha256": {
"assembly": "fe9dd283cdc5034b74fcca6d2f344da08a876ef1fde6a4db35626a680a12183f",
"profile": "e3f778d24ce02c1ddc21b4632c3e0d99e064d8db359b03b019fd8b37baf815b5",
"result": "6be31e95e41bcc83577b364b01a0339d16a8f6f09277b9d8d964c0ea4c1f158c"
},
"failure_controls": [
"added rank-16 child",
"swap accepted/rejected endpoint labels",
"one-grid-unit kappa mutation",
"omit conservative surcharge"
],
"interval_method": "standard-library Fraction; 48-term binary-reduced atanh logarithm bounds and 32-term positive Taylor exponential with geometric tail",
"primary_kappa_grid": "836340409246491/500000000000000000",
"profile_delta": {
"1": 1,
"7": -7,
"16": 3
},
"rank_mass_delta": 0,
"status": "PASS_PR1407_COMPLEX_PROFILE_ARITHMETIC_INDEPENDENT_AUDIT"
}
Loading
Loading