Tooling: 4.80× faster DisjointTensor.frame_data with identical frame receipts

[VictorLiwentao]

Change and scope

At base 0fdd7d82de03e43c9e5a1000dfbcd380d1b687d2, merged CrocSwap/integer-mult-bounds#2001 left DisjointTensor.verify_frames / frame_data as an explicit next question after shipping closed-form verify support masks. Profiling on the historical full-size circuit (h=26, 60372 active nodes) shows most of verify_frames wall time inside the per-node reachable-target bit enumeration in frame_data.

CrocSwap/integer-mult-bounds#1044 already proved exact target-union propagation as an opt-in adapter under research/disjoint-frame-unions/. Production scripts/experiments/disjoint_tensor.py still used the historical bit loop. This PR promotes the same algebra into the default path, keeps every descendant bitset, source union and rectangle require, and leaves verify_frames receipts fieldwise identical. Only the source pin for scripts/experiments/disjoint_tensor.py changes inside the four certificates that bind that file.

Tooling improvement; no new kappa. Selected frontier untouched.

Result

On Linux / CPython 3.12.3 (Linux-6.12.94+-x86_64-with-glibc2.39), five alternating fresh-process pairs timing only DisjointTensor(26).verify_frames (construction excluded) gave median wall time 2.232 → 0.465 seconds (4.80×). Timing is host-descriptive and is not a correctness gate. Every frame receipt matches the baseline checker fieldwise (distinct_nontrivial_transitions, frame_types, flags).

Package: research/disjoint-tensor-frame-unions/.

Validation

python3.12 -I -B research/disjoint-tensor-frame-unions/verify.py
python3.12 -I -O -B research/disjoint-tensor-frame-unions/verify.py
python3.12 -B scripts/research_package.py --package research/disjoint-tensor-frame-unions
python3.12 -I -B research/disjoint-tensor-frame-unions/benchmark.py --rounds 5

Expected:

PASS_DISJOINT_TENSOR_FRAME_UNIONS frame_cases=6 map_cases=6 certs=4 no_new_kappa
PASS_REGISTERED_RESEARCH_PACKAGE research/disjoint-tensor-frame-unions outcome=completed
PASS_IDENTICAL_DISJOINT_TENSOR_FRAME_UNION_RECEIPTS

Default verify is read-only. XOR-of-mask nonequivalence and corrupted-rectangle ValueError controls reject. Full make verify, selected cold replay, Lean, and multiplication-theorem obligations were not rerun for this isolated historical-checker change.

Overlap and credit

Closest priors (different objects or delivery): CrocSwap/integer-mult-bounds#756, CrocSwap/integer-mult-bounds#1044 (opt-in exact union propagation; this promotes it to production), CrocSwap/integer-mult-bounds#1793, CrocSwap/integer-mult-bounds#1861 (open; already conflicting with main; different primary pin binary_phase_frames.py), CrocSwap/integer-mult-bounds#1922, CrocSwap/integer-mult-bounds#1984, CrocSwap/integer-mult-bounds#1994, CrocSwap/integer-mult-bounds#2001 (support masks + open verify_frames / frame_data question this answers), CrocSwap/integer-mult-bounds#2100 (same ternary-side.json file, different pin key ternary_side.py). No competing open PR on production frame_data at refresh.

Douglas R. Colkitt authored the disjoint-tensor / exclusion circuits. Prepared by Wentao Li (VictorLiwentao, @Is__Victor) with substantial AI assistance, under Apache-2.0.