Tooling: 2.29× faster production HPE transport via integer rows with identical receipts

[VictorLiwentao]

Change and scope

At base 0fdd7d82de03e43c9e5a1000dfbcd380d1b687d2, production research/complex-k492-owner-repair/code/hpe_audit_and_transport.py still stored every scalar row as a dictionary of Fraction values. Wronskode’s opt-in adapter research/hpe-audit-fast-seed93/ (incorporated #1137) already proved shared-denominator sparse integer rows with identical complete transport receipts. This PR wires that equivalent rewrite into the production k492 transport auditor used by changed-word tooling.

SELECTED-SOURCE-bound research/complex-sw4-frame8/local/hpe_audit_and_transport.py copies are intentionally left pairwise so the selected inventory stays untouched. Dependent packages that previously live-pinned the k492 path retarget to that sw4 transport. Distinct from #2357 (hpe_audit_local.py) and from #2409 (PR635-path production integer rows).

No certificate κ pins change.

Tooling improvement; no new kappa. Selected frontier untouched.

Result

On Linux / CPython 3.12.3 with taskset -c 0, five alternating fresh-process pairs over the five honest frontier1042 layer-end words gave median wall time 6.452 → 2.816 s (2.29×). Timing is host-descriptive and is not a correctness gate. Every honest and fuzz outcome matches the pairwise baseline.

Package: research/hpe-transport-integer-production/.
Tip: 7452a6ce9b79dee2e2b0d2a8819c4d4099d302e2.

Validation

python3.12 -I -B research/hpe-transport-integer-production/verify.py
python3.12 -I -O -B research/hpe-transport-integer-production/verify.py
python3.12 -B scripts/research_package.py --package research/hpe-transport-integer-production
python3.12 -B -m unittest tests.test_hpe_transport_integer_production
taskset -c 0 python3.12 -I -B research/hpe-transport-integer-production/benchmark.py --rounds 5

Expected:

PASS_HPE_TRANSPORT_INTEGER_PRODUCTION assertions=26 honest=5 fuzz=20 corruption=receipt_divergence certs=0 no_new_kappa
PASS_REGISTERED_RESEARCH_PACKAGE research/hpe-transport-integer-production outcome=completed
PASS_IDENTICAL_HPE_TRANSPORT_INTEGER_PRODUCTION_RECEIPTS

Default verify is read-only. Temp-copy ADD-drop foil rejects via audit failure or receipt divergence. Package gates under -O still load checkers with compile(..., optimize=0). 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 delivery): Wronskode’s research/hpe-audit-fast-seed93/ / #1137 (opt-in integer rows; this is the production wiring), CrocSwap/integer-mult-bounds#756, CrocSwap/integer-mult-bounds#2357 (integer rows for hpe_audit_local.py, different object). Leave-alone tooling through #2409 was not reopened. No competing open PR on production k492 hpe_audit_and_transport.py at refresh.

Credit Wronskode, icekylinx, and the retained transport-auditor lineage. Prepared by Wentao Li (VictorLiwentao, @Is__Victor) with substantial AI assistance, under Apache-2.0.