Tooling: accelerate historical Leech minors with exact byte lanes

research/leech-minor-byte-lanes/BENCHMARK.json

@@ -0,0 +1,88 @@
+{
+  "adapter_sha256": {
+    "PINS.json": "47d4c05d3753e26cb84aeaea12a697174e62a7bf24f7c4699c4a5f9f19afe6d0",
+    "benchmark.py": "0e0d7e9c9f2c6d7f050bf015a5d6fbc892df809bbac5faae9b88a1f004fd1bf2",
+    "lanes.py": "2c242d4284e1c8520cbc0cfbbb20ddf186c662e5f6173bed400bbd4022df34c7",
+    "replay.py": "aa73a3ea405c1b3cdbe20888624b2f3e93a8ddff65476e25171c71975f4b8490"
+  },
+  "base_commit": "f567884aeb8851b6e794af1184119a3bfbe1e631",
+  "certificate_sha256": "f72787ebf4a991e32705c7fb7ef7ad559a9964b27c0f91d1b5a034c05027cfb0",
+  "machine": "x86_64",
+  "median_speedup": 2.107932325412675,
+  "median_wall_seconds": {
+    "baseline": 18.94655950699996,
+    "lanes": 8.988220009999964
+  },
+  "platform": "Linux-6.18.44-x86_64-with-glibc2.41",
+  "python": "3.12.14 (main, Aug 25 2026, 14:00:49) [Clang 22.1.3 ]",
+  "samples": [
+    {
+      "mode": "baseline",
+      "receipt": {
+        "certificate_sha256": "f72787ebf4a991e32705c7fb7ef7ad559a9964b27c0f91d1b5a034c05027cfb0",
+        "executed_geometry_sha256": "756778eeca27f490adc7354a122dc50c40375156259991469c47a90949118ea1",
+        "mode": "baseline",
+        "status": "PASS_AFFINE_VECTOR_REPLAY"
+      },
+      "round": 1,
+      "wall_seconds": 17.051406369999995
+    },
+    {
+      "mode": "lanes",
+      "receipt": {
+        "certificate_sha256": "f72787ebf4a991e32705c7fb7ef7ad559a9964b27c0f91d1b5a034c05027cfb0",
+        "executed_geometry_sha256": "e3b1cfb013afc4ac8d4306836b8840d9776a2ac82eb76b44582fa6bd46a6f957",
+        "mode": "lanes",
+        "status": "PASS_AFFINE_VECTOR_REPLAY"
+      },
+      "round": 1,
+      "wall_seconds": 8.988220009999964
+    },
+    {
+      "mode": "lanes",
+      "receipt": {
+        "certificate_sha256": "f72787ebf4a991e32705c7fb7ef7ad559a9964b27c0f91d1b5a034c05027cfb0",
+        "executed_geometry_sha256": "e3b1cfb013afc4ac8d4306836b8840d9776a2ac82eb76b44582fa6bd46a6f957",
+        "mode": "lanes",
+        "status": "PASS_AFFINE_VECTOR_REPLAY"
+      },
+      "round": 2,
+      "wall_seconds": 7.5365457379999725
+    },
+    {
+      "mode": "baseline",
+      "receipt": {
+        "certificate_sha256": "f72787ebf4a991e32705c7fb7ef7ad559a9964b27c0f91d1b5a034c05027cfb0",
+        "executed_geometry_sha256": "756778eeca27f490adc7354a122dc50c40375156259991469c47a90949118ea1",
+        "mode": "baseline",
+        "status": "PASS_AFFINE_VECTOR_REPLAY"
+      },
+      "round": 2,
+      "wall_seconds": 18.94655950699996
+    },
+    {
+      "mode": "baseline",
+      "receipt": {
+        "certificate_sha256": "f72787ebf4a991e32705c7fb7ef7ad559a9964b27c0f91d1b5a034c05027cfb0",
+        "executed_geometry_sha256": "756778eeca27f490adc7354a122dc50c40375156259991469c47a90949118ea1",
+        "mode": "baseline",
+        "status": "PASS_AFFINE_VECTOR_REPLAY"
+      },
+      "round": 3,
+      "wall_seconds": 22.67777050899997
+    },
+    {
+      "mode": "lanes",
+      "receipt": {
+        "certificate_sha256": "f72787ebf4a991e32705c7fb7ef7ad559a9964b27c0f91d1b5a034c05027cfb0",
+        "executed_geometry_sha256": "e3b1cfb013afc4ac8d4306836b8840d9776a2ac82eb76b44582fa6bd46a6f957",
+        "mode": "lanes",
+        "status": "PASS_AFFINE_VECTOR_REPLAY"
+      },
+      "round": 3,
+      "wall_seconds": 14.988175710000064
+    }
+  ],
+  "scope": "Complete nine-screen historical affine-vector audit; process startup, source admission, all checks and output included. No timing threshold.",
+  "status": "PASS_IDENTICAL_COMPLETE_CERTIFICATES"
+}

research/leech-minor-byte-lanes/CLAIM.md

@@ -0,0 +1,57 @@
+# Seed 86 verification engineering claim
+
+- Assignment: OS randomness (`secrets.randbelow(100)`) returned **86** at
+  2026-10-13 10:29:34 UTC; track 85–94, verification engineering.
+- Exact base: `f567884aeb8851b6e794af1184119a3bfbe1e631` (fetched upstream main).
+- Object: the five historical Leech lattice minor replays in
+  `scripts/experiments/affine_vector_screens.py`, using every recorded row and
+  column in `scripts/experiments/affine_vector_witnesses.json`.
+- Method: replace pairwise Python dot products by carry-safe exact integer
+  lanes; compare full packed row streams, ranks and certificates on identical
+  inputs. Preserve validation and rejection checks; no probabilistic ranks.
+- Deliverable: opt-in source-pinned accelerator, before/after benchmark,
+  complete equivalence checks and negative controls. No new kappa or theorem.
+- Expected finish: 2026-10-13 11:10 UTC (22:10 Australia/Sydney).
+
+Three candidates were (0) memoizing repeated subset-core interval evaluations,
+(1) accelerating the coded-carry exhaustive affine-agreement control, and
+(2) reconstructing the five Leech minors with exact integer lanes. `86 % 3 = 2`
+selects the last after excluding occupied recent tooling targets. Seed 86 also
+fixes the supplemental generated test cases; it does not change the five
+historical fixtures or their complete coverage.
+
+The selected record was followed to the PR799 X4 construction and maintainer
+review (conditional kappa 0.001661634653669567, with PR589/555). The PR ledger,
+contribution guidance and historical affine-vector negative-result scope were
+read. The latest 100 open PRs through #891 and all-state tooling, subset,
+affine, coded-tooling, Leech-tooling and `affine_vector_screens` searches were
+checked. #752 accelerates quadratic-affine minors and #778 single-intersection
+Gram matrices: related integer-packing techniques, distinct matrices and
+checks. #849 concerns dyadic X4 arithmetic and #871 historical joint-frame
+rank checks. No duplicate Leech-minor accelerator was found.
+
+Only issue #650 (request to enable Discussions) was open and had no comments;
+no pinned sprint coordination issue was available. Claim recorded locally per
+the sprint instructions. No delegated workers.
+
+Contributor: CubeJerry, with substantial OpenAI Codex assistance. Apache-2.0.
+
+## Result — 2026-10-13 10:50 UTC
+
+Completed the opt-in accelerator: all 5,927 original Leech rows (7,531,037
+entries) and the complete nine-screen certificate agree exactly. Three final
+alternating fresh-process pairs measured 18.946560s to 8.988220s median,
+**2.107932x** on Linux/CPython 3.12.14. Eleven test groups, all three unchanged
+predecessor tests in both modes, minimal offline relocation, selected-record
+integrity and its five tests pass. No new kappa is claimed.
+
+Result package and reproducible evidence:
+[leech-minor-byte-lanes](https://github.com/CubeJerry/integer-mult-bounds/tree/tooling/seed86-exact-verification/research/leech-minor-byte-lanes).
+
+Final metadata refresh covered new PRs through #894 and all-state Leech and
+affine-checker searches. No overlapping accelerator was found. DominikScholz's
+#598 (head `615ca6f06fdea0c50dbef733370c47349e25c555`) studies different
+Leech/Turyn constructions and fitting bounds; those negative searches are not
+reopened here. The pinned-issues GraphQL query returned no coordination issue.
+Upstream main remains the recorded base. All 41 retained input hashes also
+match the base Git objects. The branch is ready for one focused tooling PR.

research/leech-minor-byte-lanes/EXPECTED.json

@@ -0,0 +1,49 @@
+{
+  "assignment_seed": 86,
+  "cases": [
+    {
+      "columns": 715,
+      "rows": 683,
+      "rows_sha256": "b54f09a244ab154e868979e71fd81a428a6b4a9169f96bc6ded04e45d8fdfa35",
+      "shift": 0,
+      "signed": false
+    },
+    {
+      "columns": 1343,
+      "rows": 1311,
+      "rows_sha256": "f743deb780a127a387d8e086ae1d48b42259e3070cc2c506516f0a51d2a88f33",
+      "shift": -16,
+      "signed": true
+    },
+    {
+      "columns": 1343,
+      "rows": 1311,
+      "rows_sha256": "51e0a2e8c02829aa8cac3dc1bde9d917e6330a7cb3526c35691b94be8cb8a833",
+      "shift": -8,
+      "signed": true
+    },
+    {
+      "columns": 1343,
+      "rows": 1311,
+      "rows_sha256": "10ecb66bffb241e22238f9b6264821f95f5f6924414a007df63ba59db2260fb5",
+      "shift": 8,
+      "signed": true
+    },
+    {
+      "columns": 1343,
+      "rows": 1311,
+      "rows_sha256": "c99ed2b19e1210503de5cba13452b355e351f238e6a4676f6bfe8036bf9de4d0",
+      "shift": 16,
+      "signed": true
+    }
+  ],
+  "complete_certificate_sha256": "f72787ebf4a991e32705c7fb7ef7ad559a9964b27c0f91d1b5a034c05027cfb0",
+  "executed_geometry_sha256": {
+    "baseline": "756778eeca27f490adc7354a122dc50c40375156259991469c47a90949118ea1",
+    "lanes": "e3b1cfb013afc4ac8d4306836b8840d9776a2ac82eb76b44582fa6bd46a6f957"
+  },
+  "full_rows_compared": 5927,
+  "matrix_entries_compared": 7531037,
+  "scope": "Five historical Leech minors and unchanged nine-screen certificate; no new kappa.",
+  "status": "PASS_LEECH_BYTE_LANES"
+}

research/leech-minor-byte-lanes/NOTICE.md

@@ -0,0 +1,34 @@
+# Attribution and source pins
+
+Contribution: Jerry Chen (@CubeJerry), with substantial OpenAI Codex assistance
+in implementation, analysis, tests and documentation. Apache-2.0; the
+repository LICENSE applies. No independent human review is claimed.
+
+The original affine-vector families, Golay/Leech generator, fixed witnesses,
+rank rejections and audit are by **Douglas R. Colkitt**, retained from commit
+`d55bcd62ceb8f322bcf311a5a66e94c3ac7724f0`. The source note credits Brouwer and
+Van Maldeghem, *Strongly Regular Graphs*, section 6.3.1 for the minimal-vector
+construction. The original note, source files and all attribution remain
+unchanged. `PINS.json` gives exact file hashes at base
+`f567884aeb8851b6e794af1184119a3bfbe1e631`.
+
+**Curtis Dransfield (@Curtis-D), PR752**, head
+`3603d8d9184995ae1cc67c0f5e01f02f3a1d634e`, supplies the related carry-safe
+integer-lane approach for quadratic-affine Hermitian minors. Its technique and
+coverage discipline are credited; no code or fixture from that PR is copied.
+This contribution reconstructs a distinct set of five signed real Leech
+minors and decodes bytes directly.
+
+**merramil, PR778**, head `f015b04c74801fa330f818d178e809c760fb5469`, improves
+historical single-intersection Gram construction with symmetric packing.
+It is related concurrent performance work, not a dependency of this package.
+The original `rank_of_rows` imported here remains unchanged.
+
+The retained upstream OpenAI manuscript and sources are pinned at
+`adc7f1241b42e322a6451854ab7e4b4c146bf78a` with their original Apache-2.0
+license and notices. No upstream file is modified.
+
+Selected-result context follows Jeremy Kothe's PR799 X4 construction,
+Chafik Boukhalfa's PR741 and DreamingOfClouds' PR589 on eumemic's PR555.
+Those constructions and their predecessors are not changed or re-certified
+by this historical-checker optimization.

research/leech-minor-byte-lanes/PINS.json

@@ -0,0 +1,46 @@
+{
+  "base_commit": "f567884aeb8851b6e794af1184119a3bfbe1e631",
+  "files": {
+    "certificates/affine-vector-screens.json": "f72787ebf4a991e32705c7fb7ef7ad559a9964b27c0f91d1b5a034c05027cfb0",
+    "docs/research/affine-vector-screens.md": "22e5beab6df8127fc4377142768fc1a0a3317d7e8fba52c089670ecf1aa076b6",
+    "scripts/audit_affine_vectors.py": "a3fccaeadb9a5db72c33787a055a54374504a6aa231907080396b2cf822e94a7",
+    "scripts/audit_cancellation.py": "508934ee4d9514a5519813603b6192a0d53103b6e30d65b8d53a1a774de244b6",
+    "scripts/audit_fused_block.py": "2afb83ca8e5f193ca8a4c242becf267ad1d9dc2578ba238b820e98d5185f6288",
+    "scripts/audit_joint_frames.py": "3df8061392ce0ba42bbde8c02f2a30c8141145e70e9a48c18da73d3aafb14423",
+    "scripts/certify.py": "2348d95008276c5d5764f7e028c3918e484582cd589bd4d9b82919161745e53e",
+    "scripts/exclusion_circuit.py": "b253f9d7318becda1c1f3b24e2287ed44a9c27fcea00cf9d531afb950d0457a0",
+    "scripts/experiments/affine_vector_screens.py": "756778eeca27f490adc7354a122dc50c40375156259991469c47a90949118ea1",
+    "scripts/experiments/affine_vector_witnesses.json": "672a4318c48f5286ce088d34736c38957b7f662a5f753d440dbc638e7da5f62a",
+    "scripts/experiments/cube_core_family.py": "e58f1f4067bda2d4838343f42ec2505316f006561f0702e869847d88b85413a8",
+    "scripts/experiments/rank_product_core.py": "7a91c1188c5d42e864a5d2e42397b09c435b526ae4f4210c44bf615de488708e",
+    "scripts/experiments/single_intersection.py": "08ddae46e5621a5b4c525b504e67ae655abe9b49461d894feaa7abc89a51847d",
+    "scripts/paired_exclusion_circuit.py": "8f05f04cf06217befd6d0ee0f2d7d0c5c5eb32cfab8c9a8ca7917fd6d620b7d2",
+    "scripts/paired_network.py": "991223dc3c1374284c361ed017e42bf732dd971fa0bea2be2282cfb18257b3f7",
+    "scripts/prepare_layers.py": "a6e2adc612f264b1c0e51d18ce6315de9ae5e2ae2420aa22e99a22d8bf304947",
+    "scripts/rational_tensor.py": "629b3f00a1aaab32fe908a73ca8f8412918209a300e6b1e5afe6325b9cfed3c3",
+    "scripts/search_network.py": "3a4faa5037d9b8e7ca5828afac193a96b6966a8b114f14937921795b24ec4f6a",
+    "scripts/shared_point_circuit.py": "c9bb7a5df92d260a4e3ad532b53e26936730d7c93f7facfb0784957c1e81c8ba",
+    "scripts/shared_point_network.py": "a85e7f0d8c42d6a27e3846308571ca324a1824ac5c2d59e4daa5c7673c33c697",
+    "tests/test_affine_vector_screens.py": "ab2c301c10a28438ac3d45a7e052e453b87a1f8057ce3ea897b14573f99f8bed",
+    "upstream/LICENSE": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4",
+    "upstream/README.md": "8d1ceff898a4e0e681679f91c4def485820be9e99f04e4861f2ce601a364723e",
+    "upstream/build/figures/multiplication-flow.tex": "3654d2728b0c98919d61760e1802804ed616546a747f9167a93e17bb8697f320",
+    "upstream/build/figures/proof-map.tex": "76d4deacf07bea355412abcfd798c7cbe6dcccef29a4a420a9e2a7b158f23848",
+    "upstream/build/main.tex": "42f0f506976bc2084374a71e97edc1549689a7accf35dda4db2a18af4f13c1aa",
+    "upstream/build/references.bib": "33cc2df13ccf85330849e8fb34d4ac7613b23f625787457a5b7b9aa6d7369476",
+    "upstream/build/sections/00-introduction.tex": "98721f03682ccf1cc7d0850b07adabaa2d8f3afe552e3861753eba9d284d0da0",
+    "upstream/build/sections/01-history.tex": "798956c8150c31e3e2f5f29c47110d25ce619faad1374f4d3a0f15fa3ee33ead",
+    "upstream/build/sections/02-streams.tex": "606c80db61cad13aa0c3b6060dc4b88dbbe7ac60cdd831aa93f28d908fd09dab",
+    "upstream/build/sections/03-motifs.tex": "ed76761194988a01af77ead0ea6469cb5dcd29690ab985a0b5a4006b0f2b69bb",
+    "upstream/build/sections/04-swap.tex": "412b170ccaab38dba96e95e3cbfae09f4876a9818f8d278776a746b0f8044906",
+    "upstream/build/sections/05-layers.tex": "20cfc8d12099d4e4ed9e3e7eeb6162edd9b6e0e076f24693364cd24377252594",
+    "upstream/build/sections/06-transforms.tex": "924316e4bc905066281d8f9e8746a49d8829aeb393688608deb1e61c3c87853c",
+    "upstream/build/sections/07-resampling.tex": "8a6b91670f9163afb54892489fa2210c77fe1b058f0325ec7d18b44b94be6e2c",
+    "upstream/build/sections/08-assembly.tex": "763d7945b1e1ec4ebae9b799bcefcc45fa9bae6e2ffad799868a9c3d513dbd4a",
+    "upstream/build/sections/09-exact-arithmetic.tex": "5714e5d0656c78c6a1fde8d01fd11f1ba856a7ea94149b283bd9a50fdbf516c3",
+    "upstream/build/sections/10-transposition.tex": "c14ad8d56112f6578678a58211bc8d63c3a4b60e55078ac73bc38f7348e713cb",
+    "upstream/build/sections/11-grouped-rectangles.tex": "6525cd0a01980e56accfeaf06e9d875bc699197ba42c6e22b69435e352275174",
+    "upstream/formalization.yaml": "2dcbd0d6e6a22475f53d49bfdfcac9b6d2b470653c332a9f56522f5be166edb9",
+    "upstream/manifest.json": "16ab13971694cabab09440c47ebf2fc0496a9d35fc3af33950fda3498f8e2392"
+  }
+}