Mister-Meeseeks · Oct 15, 2026 · Oct 14, 2026
diff --git a/certificates/dyadic-central-factor.json b/certificates/dyadic-central-factor.json
@@ -102,7 +102,7 @@
     "scripts/binary_phase_frames.py": "4148cfa313cf050ed25296e8db36d35871520b8318fd086c7d1facf0c6af67d7",
     "scripts/complex_compression.py": "9b6f758a5985ba5d9fdd726fe6556b3d3b4bd9d98aaf8ba06a7bb58093c4c504",
     "scripts/exclusion_circuit.py": "f9fbcd6e85e442cbd0926ac0887414e20735b2dc923d2d5822c803ceb46726d3",
-    "scripts/experiments/complex_centers.py": "3744396587d729253d80980f6310b56e7bf0ea21b165bd178d0d6d4c49faf2e2",
+    "scripts/experiments/complex_centers.py": "39cfcf869dc4a00dd30a1e91efb986074371621dad14b390cf1053f6aeea79bf",
     "scripts/experiments/disjoint_tensor.py": "085600fc922ae167b5981505e1a5caba4d1c1df15482f61d83d2c086fa89eac1",
     "scripts/general_network_guard.py": "790a2064449afdbffd6dbd4adbcf4a28eaf41931ce7134e4a4e8766f3764270c"
   },

diff --git a/research/central-factor-production/NOTICE.md b/research/central-factor-production/NOTICE.md
@@ -0,0 +1,19 @@
+# Attribution and license
+
+Prepared by Wentao Li ([VictorLiwentao](https://github.com/VictorLiwentao),
+[@Is__Victor](https://x.com/Is__Victor), wentaol4@illinois.edu) with substantial
+AI assistance for orientation, overlap search, profiling, production wiring,
+packaging, benchmarking and writing. Not independent human peer review.
+Apache-2.0; see the repository root `LICENSE`.
+
+The injective packed-row algorithm and its radix argument were introduced by
+**Ross Richardson (@RossRichardson)** in the opt-in adapter
+`research/central-factor-packed/` (see that package's NOTICE.md). This package
+wires an equivalent check into the production
+`scripts/experiments/complex_centers.py:verify_factor` path used by
+`make verify-certificates`.
+
+The original dyadic factor, structural validation, receipt schema, side circuit,
+phase argument and cost ledger are by **Douglas R. Colkitt**, retained at
+`d55bcd62ceb8f322bcf311a5a66e94c3ac7724f0`. Upstream material retains its own
+source notices and license.
diff --git a/research/central-factor-production/README.md b/research/central-factor-production/README.md
@@ -0,0 +1,94 @@
+# Production packed dyadic central-factor check
+
+**Tooling improvement; no new kappa.** At base
+`0fdd7d82de03e43c9e5a1000dfbcd380d1b687d2`, `scripts/experiments/complex_centers.py:verify_factor`
+still used the historical nested entrywise loop over all scatter/gather pairs.
+Ross Richardson's opt-in adapter `research/central-factor-packed/` already proved
+an injective packed-row algorithm with identical receipts. This package wires
+that equivalent check into the **production** source used by
+`scripts/audit_complex_centers.py` / `make verify-certificates`.
+
+Only the source pin for `scripts/experiments/complex_centers.py` changes inside
+`certificates/dyadic-central-factor.json`. Every coefficient is still checked;
+`circuit`/`factor` generators, side/phase obligations and the retained historical
+`2^-31` language are unchanged. Selected frontier untouched.
+
+## Reproduce
+
+Python **3.12**, standard library only, from the repository root:
+
+```sh
+python3.12 -I -B research/central-factor-production/verify.py
+python3.12 -I -O -B research/central-factor-production/verify.py
+python3.12 -B scripts/research_package.py --package research/central-factor-production
+python3.12 -I -B research/central-factor-production/benchmark.py --rounds 5
+```
+
+Expected:
+
+```text
+PASS_CENTRAL_FACTOR_PRODUCTION heights=[8, 10, 12, 18, 26] ... no_new_kappa
+PASS_REGISTERED_RESEARCH_PACKAGE research/central-factor-production outcome=completed
+PASS_IDENTICAL_CENTRAL_FACTOR_PRODUCTION_RECEIPTS
+```
+
+Default verify is read-only. Use `--write-certificate` only after editing the
+production source. Timing is host-descriptive and is not a correctness gate.
+
+On Linux / CPython 3.12.3, five alternating fresh-process pairs gave median
+**factor-only 2.615 → 0.059 s (44.06×)** and **complete audit 5.784 → 3.356 s
+(1.72×)**. Samples are in [`RESULT.json`](RESULT.json).
+
+## Why the packed check is equivalent
+
+For each scatter row `u(S)` and gather column `v(T)` the historical obligation is
+
+```text
+sum_i u(S)[i] v(T)[i] = |S ∩ T| - 1.
+```
+
+Gather entries are in `{-1,0,1}` and appear at most once per column, so
+`|actual - expected| <= ||u(S)||_1 + 2`. Choose
+`bits = (max_S ||u(S)||_1 + 2).bit_length()` and place `B = 2**bits`. Packing
+every gather/incidence column as `sum_j coeff_j B**j` and comparing packed
+actual/expected rows is an injective encoding of the difference vector: the
+packed equality holds if and only if every coefficient matches. This is the
+same argument as `research/central-factor-packed/README.md`; structural and
+scalar-charge gates are unchanged.
+
+## Overlap and credit
+
+Closest priors:
+
+- [Ross Richardson / central-factor-packed](../central-factor-packed/) — opt-in
+  packed adapter; this package is the production wiring of that algorithm.
+- [CrocSwap/integer-mult-bounds#756](https://github.com/CrocSwap/integer-mult-bounds/pull/756)
+  — identical-receipt tooling quality bar.
+- Open leave-alone tooling on other objects (#1922 joint frames, #2054/#2150
+  shared/mixed-point, #2100 ternary, #2117 DisjointTensor.frame_data, #1861
+  parity-side, #2155 exclusion embedding residual) was not reopened.
+
+Douglas R. Colkitt authored the dyadic central factor and original checker.
+Ross Richardson authored the packed algorithm. Prepared by Wentao Li
+([VictorLiwentao](https://github.com/VictorLiwentao),
+[@Is__Victor](https://x.com/Is__Victor)) with substantial AI assistance, under
+Apache-2.0.
+
+## Open next questions
+
+1. **Question:** After #2117's `DisjointTensor.frame_data` lands, what is the
+   residual wall time of `audit_complex_centers.py` with packed `verify_factor`
+   already in production?
+   **Why it matters:** `frame_data` still dominates the complete audit after the
+   coefficient check shrinks.
+   **First step:** Replay `benchmark.py --rounds 5` on a tree that also contains
+   the merged #2117 head.
+   **Blocked by:** #2117 is leave-alone / open.
+
+2. **Question:** Should `verify-certificates` adopt a microbenchmark regression
+   gate for `verify_factor(h=26)`?
+   **Why it matters:** A future edit could restore the nested loop without
+   failing equivalence tests.
+   **First step:** Decide whether `RESULT.json` medians belong in CI or remain
+   descriptive.
+   **Blocked by:** Maintainer CI-budget preference.
diff --git a/research/central-factor-production/RESULT.json b/research/central-factor-production/RESULT.json
@@ -0,0 +1,142 @@
+{
+  "ai_assistance": "Substantial AI assistance for orientation, overlap search, profiling, implementation, packaging, benchmarking and writing. Not independent human peer review.",
+  "base_commit": "0fdd7d82de03e43c9e5a1000dfbcd380d1b687d2",
+  "complete_audit": {
+    "median_baseline_seconds": 5.784228472999985,
+    "median_optimized_seconds": 3.356052378000186,
+    "median_ratio": 1.7235215132270096,
+    "rounds": 5,
+    "samples": [
+      {
+        "baseline_seconds": 5.760912544999883,
+        "optimized_seconds": 3.3874211810002635,
+        "order": [
+          "baseline",
+          "optimized"
+        ],
+        "ratio": 1.7006779603647506,
+        "round": 0
+      },
+      {
+        "baseline_seconds": 5.784228472999985,
+        "optimized_seconds": 3.356052378000186,
+        "order": [
+          "optimized",
+          "baseline"
+        ],
+        "ratio": 1.7235215132270096,
+        "round": 1
+      },
+      {
+        "baseline_seconds": 5.861394950999966,
+        "optimized_seconds": 3.3083781870000166,
+        "order": [
+          "baseline",
+          "optimized"
+        ],
+        "ratio": 1.7716822623337942,
+        "round": 2
+      },
+      {
+        "baseline_seconds": 5.726917460000095,
+        "optimized_seconds": 3.339230219000001,
+        "order": [
+          "optimized",
+          "baseline"
+        ],
+        "ratio": 1.715041217408225,
+        "round": 3
+      },
+      {
+        "baseline_seconds": 6.1704699699998855,
+        "optimized_seconds": 3.517462851000346,
+        "order": [
+          "baseline",
+          "optimized"
+        ],
+        "ratio": 1.7542388452645747,
+        "round": 4
+      }
+    ],
+    "scope": "audit"
+  },
+  "engines": {
+    "baseline": "research/central-factor-production/baseline.py",
+    "optimized": "scripts/experiments/complex_centers.py"
+  },
+  "environment": {
+    "cpu_affinity_set": false,
+    "executable": "/usr/bin/python3.12",
+    "implementation": "CPython",
+    "os_caches_flushed": false,
+    "platform": "Linux-6.12.94+-x86_64-with-glibc2.39",
+    "python": "3.12.3 (main, Aug 31 2026, 10:18:26) [GCC 13.3.0]"
+  },
+  "factor_only": {
+    "median_baseline_seconds": 2.615371209999921,
+    "median_optimized_seconds": 0.05935716699968907,
+    "median_ratio": 44.06159091139948,
+    "rounds": 5,
+    "samples": [
+      {
+        "baseline_seconds": 2.646467106000273,
+        "optimized_seconds": 0.061992203000045265,
+        "order": [
+          "baseline",
+          "optimized"
+        ],
+        "ratio": 42.690321974819,
+        "round": 0
+      },
+      {
+        "baseline_seconds": 2.538598202999765,
+        "optimized_seconds": 0.060868976000165276,
+        "order": [
+          "optimized",
+          "baseline"
+        ],
+        "ratio": 41.70594561986507,
+        "round": 1
+      },
+      {
+        "baseline_seconds": 2.518348651999986,
+        "optimized_seconds": 0.05801914999983637,
+        "order": [
+          "baseline",
+          "optimized"
+        ],
+        "ratio": 43.40547305513935,
+        "round": 2
+      },
+      {
+        "baseline_seconds": 2.615371209999921,
+        "optimized_seconds": 0.05935716699968907,
+        "order": [
+          "optimized",
+          "baseline"
+        ],
+        "ratio": 44.06159091139948,
+        "round": 3
+      },
+      {
+        "baseline_seconds": 2.6376186860002235,
+        "optimized_seconds": 0.05685944800006837,
+        "order": [
+          "baseline",
+          "optimized"
+        ],
+        "ratio": 46.38839768541285,
+        "round": 4
+      }
+    ],
+    "scope": "factor"
+  },
+  "no_new_kappa": true,
+  "predecessor": "research/central-factor-packed/",
+  "source_sha256": {
+    "baseline.py": "3744396587d729253d80980f6310b56e7bf0ea21b165bd178d0d6d4c49faf2e2",
+    "research/central-factor-packed/packed.py": "42f244d6e81dc5ea038a9dea19ccc803872603a94d5e405de14af8a47a11b9f6",
+    "scripts/experiments/complex_centers.py": "39cfcf869dc4a00dd30a1e91efb986074371621dad14b390cf1053f6aeea79bf"
+  },
+  "status": "PASS_IDENTICAL_CENTRAL_FACTOR_PRODUCTION_RECEIPTS"
+}
diff --git a/research/central-factor-production/SOURCES.json b/research/central-factor-production/SOURCES.json
@@ -0,0 +1,54 @@
+{
+  "schema_version": 1,
+  "base_commit": "0fdd7d82de03e43c9e5a1000dfbcd380d1b687d2",
+  "files": [
+    {
+      "path": "docs/research/agent-task-catalog.json",
+      "usage": "provenance",
+      "git_blob": "e8870e5917057d59b730eb14721a778639eca050",
+      "sha256": "d46eb6d5b3e279428753d770653663fbdf1063e55c255ea073d768f344927ef7"
+    },
+    {
+      "path": "certificates/selected-result.json",
+      "usage": "provenance",
+      "git_blob": "ae6a99f28aa468649eb66a361e764d769363bf11",
+      "sha256": "10b9e340fd8e354a37a0c4813c90e5de1dd42044b5f6aca4a3fc5a42b685fa32"
+    },
+    {
+      "path": "certificates/dyadic-central-factor.json",
+      "usage": "provenance",
+      "git_blob": "5bd4e583cd61b0e5183d74799e886afb57690960",
+      "sha256": "3f46000e645c338a583e0f2ab669a7c681980743d03335548eeb838072f1d458"
+    },
+    {
+      "path": "docs/research/dyadic-central-factor.md",
+      "usage": "provenance",
+      "git_blob": "033537f01d3fced8515fac9e2c10b7048d351df3",
+      "sha256": "e97b1dfe11b5eaebacf3f910131e13bcad586ea5fc1af05db23c00ae0eea5c11"
+    },
+    {
+      "path": "scripts/audit_complex_centers.py",
+      "usage": "provenance",
+      "git_blob": "3d985ee3eb1fd43ab5cc07a8570b438bf080e4a4",
+      "sha256": "166b18f560cd0d520d23fa29ee8147e0ea056358c4268a466ec458195a97a960"
+    },
+    {
+      "path": "scripts/experiments/complex_centers.py",
+      "usage": "provenance",
+      "git_blob": "13be871d501b7ebaaaec7e755ff4834c6cc49dc7",
+      "sha256": "3744396587d729253d80980f6310b56e7bf0ea21b165bd178d0d6d4c49faf2e2"
+    },
+    {
+      "path": "research/central-factor-packed/packed.py",
+      "usage": "input",
+      "git_blob": "379b6cc4a5933ec6c0b6cbe73273e5d097fb1335",
+      "sha256": "42f244d6e81dc5ea038a9dea19ccc803872603a94d5e405de14af8a47a11b9f6"
+    },
+    {
+      "path": "research/central-factor-packed/README.md",
+      "usage": "provenance",
+      "git_blob": "baac43252038396da941185b2e49c5a5002d95c5",
+      "sha256": "b91678e94e1fa6828ecbe66a623e6164765599d54f2bcdc9c9e006a5338926a3"
+    }
+  ]
+}
diff --git a/research/central-factor-production/TASK.json b/research/central-factor-production/TASK.json
@@ -0,0 +1,70 @@
+{
+  "schema_version": 1,
+  "task_id": "replay-runtime",
+  "priority_revision": 4,
+  "base_commit": "0fdd7d82de03e43c9e5a1000dfbcd380d1b687d2",
+  "chooser_seed": "23e9e450f1bf96b41368bf1bef28dd8e",
+  "original_card": {
+    "id": "replay-runtime",
+    "question": "Can one measured replay bottleneck be sped up without dropping coverage?",
+    "starting_points": [
+      "scripts/run_isolated_tests.py",
+      "scripts/verify_replay_coverage.py",
+      "certificates/selected-result.json"
+    ],
+    "methods": [
+      "Profile one isolated replay and optimize the measured bottleneck.",
+      "Design source-bound reuse with explicit dependency invalidation and negative controls."
+    ],
+    "acceptance_gate": "Report before/after wall time on the same fixture, retained checks and invalidation behavior.",
+    "first_step": {
+      "command": [
+        "python3",
+        "-B",
+        "scripts/research_baseline.py"
+      ],
+      "expected": "PASS_RESEARCH_BASELINE_BINDINGS_ONLY",
+      "inspect": "Read the selected reproduction command and choose a single replay stage for profiling.",
+      "scope": "Positive control / source-bound orientation only; this is not a new construction or full cold replay."
+    },
+    "first_experiment": "Measure one unchanged fixture, preserve check coverage, and test one optimization with invalidation controls.",
+    "bottleneck_link": "Faster trustworthy replays increase review throughput and let mechanism candidates be checked sooner.",
+    "allowed_variations": [
+      "Profile one stage or design one source-bound cache boundary."
+    ],
+    "scope_change_triggers": [
+      "Skipping coverage or comparing different fixtures to claim speedup."
+    ],
+    "overlap_terms": [
+      "replay runtime",
+      "cache invalidation",
+      "isolated tests"
+    ]
+  },
+  "scope_status": "refined",
+  "actual_scope": "Can the production dyadic central-factor checker (scripts/experiments/complex_centers.py:verify_factor, exercised by scripts/audit_complex_centers.py / certificates/dyadic-central-factor.json) adopt the injective packed-row algorithm from research/central-factor-packed/ so the shipped verify-certificates path is faster while every coefficient check and receipt field stays identical?",
+  "scope_change_reason": "The catalogue card asks for one measured replay bottleneck with before/after wall time and retained checks. Expertise override selected verifier-engineering after profiling make verify-certificates poles. Leave-alone objects (joint frames, shared-point, mixed-point verify, ternary frames, parity-side, exclusion embedding residual, inplace compile-scan) were skipped. Ross Richardson's opt-in packed adapter already proved the algorithm; production still used the nested O(n^2) entrywise loop. This session wires that equivalent check into the shipped source and certificate pin.",
+  "bottleneck_link": "Side result on review throughput: dyadic-central-factor is part of verify-certificates. A faster identical coefficient check does not change kappa or the binding complex supplier.",
+  "budget_minutes": 60,
+  "stop_rule": "Ship an equivalence-preserving production verify_factor change with frozen baseline, identical factor and audit payloads, rejection/temp-corruption controls, certificate pin discipline, registered replay and a measured before/after median. Do not invent kappa or reopen leave-alone objects.",
+  "outcome": "completed",
+  "evidence": "PASS_CENTRAL_FACTOR_PRODUCTION via research/central-factor-production/verify.py (also under -O); PASS_IDENTICAL_CENTRAL_FACTOR_PRODUCTION_RECEIPTS via benchmark.py; certificates/dyadic-central-factor.json differs from base only in the complex_centers.py source pin.",
+  "author": "Wentao Li",
+  "github": "VictorLiwentao",
+  "email": "wentaol4@illinois.edu",
+  "ai_assistance": "Substantial AI assistance for orientation, overlap search, profiling, production wiring, packaging, benchmarking and writing. Not independent human peer review.",
+  "open_questions": [
+    {
+      "question": "After #2117's DisjointTensor.frame_data lands, what is the residual wall time of audit_complex_centers.py with packed verify_factor already in production?",
+      "why_it_matters": "frame_data still dominates the complete audit after the coefficient check shrinks; measuring the stacked gain decides whether further central-audit work is worth another session.",
+      "first_step": "Replay research/central-factor-production/benchmark.py --rounds 5 on a tree that also contains the merged #2117 head and compare complete_audit medians.",
+      "blocked_by": "#2117 is leave-alone / open; do not edit shared frame_data here."
+    },
+    {
+      "question": "Should verify-certificates adopt a microbenchmark regression gate for verify_factor(h=26)?",
+      "why_it_matters": "A future edit could restore the nested loop without failing equivalence tests.",
+      "first_step": "Decide whether RESULT.json medians belong in CI or remain descriptive.",
+      "blocked_by": "Maintainer CI-budget preference."
+    }
+  ]
+}