Tooling: 1.70x faster exact historical ternary frame checks (no new kappa)

[bbMisk]

Change and mathematical scope

Tooling improvement; no new kappa.

This opt-in tool speeds up the historical ternary frame checker by caching results for exact ordered frame pairs. It retains both orientations, every physical edge, and the existing rational arithmetic. Complete matrix-and-edge transcripts and published small-control fields for h=8 and h=9 match the retained checker. Timings use alternating fresh processes. The h=29 construction is not replayed, and kappa is unchanged.

Task and overlap

• Base: 1b59b624d21924f8e3537cdc1ac52d3be6eda2fe. Seed: 93.
• Scope: ternary-h8-h9-physical-edge-pair-memoization; task card, 20 exact source pins.
• Nearest work: Faster verification: concurrent checks, make verify-parallel, faster paired-cube frames, no joint-dual re-pins #154 (scheduling/paired-cube frames), Speed up exact frame algebra with pinned replay evidence #367 (PR364 integer elimination). Object here: historical ternary_side.verify_invocation_frames; original rational routines and full schedule retained.

Validation

python3 -B research/ternary-frame-cache/verify.py --out /tmp/ternary-frame-cache-93 --pairs 3

Expected: PASS_IDENTICAL_TERNARY_FRAME_REPLAY 84672 edges; ... ratio of medians. Fresh output directory required.

Local Python 3.14.6 / macOS arm64, three alternating pairs	Reference	Cached
Median wall seconds	11.525	6.775

Ratio of medians: 1.70. Peak RSS: 96.30 → 105.17 MiB. Shared host; OS caches not flushed.

• Exact comparison: all 84,672 ordered edge records, full rational matrices, graph and scalar receipts.
• 18 tests on Python 3.11 and 3.14; full replay on both; 5 original ternary tests; selected-source binding passed.
• Controls: singular Gram, missing common pair, non-idempotence, wrong target/role, rank-only and hash-collision keys, reversed transitions, all 20 source mutations, transcript omissions/duplicates/reordering, disabled assertions, reused output.
• Excluded: h=29 DAG/full historical assembly, selected cold replay, all-size multiplication proof.
• Retained source, upstream/, selected certificate and default checker: unchanged.

Adapter diff · Timing and exact receipts · Validation

Credit

Douglas Colkitt wrote the retained checker. Zhihao Chen's PR7 supplied the earlier five-subset motif and common-pair geometry. Hanson Sun prepared this contribution with substantial OpenAI Codex assistance. See CLAIM.json for scope, SOURCES.json for pins, and evidence/ for checks and measurements. Apache-2.0.

[Mister-Meeseeks]

100-PR batch review at cb5a79f12531022c6fcd2fe592b5030d4e3ae9e7.

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

Reviewed scope: Opt-in exact ordered-frame cache: h8/h9 full edge/matrix transcripts and controls compared with retained checker; no default replacement, h29 replay or selected κ change.

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.