Tooling: 2.25× faster mixed-point scalar digests with identical receipts

[VictorLiwentao]

Change and scope

At base 0b1cccf86e2f5132ed286a26ef0a681a495740ea, the historical mixed-point audit (scripts/audit_mixed_point.py) and the early-sharing screen both call scalar_check on large cancellation-free circuits. Profiling shows a large share of that helper’s time in per-node json.dumps((node, args), …) and per-output json.dumps((i, node), …) while building circuit_sha256.

This PR writes the same UTF-8 digest lines with an equivalent hand-rolled formatter ([node,null], [node,[a,b]], [i,node]), keeps every support and topology require, and leaves circuit_sha256 unchanged. No certificate source_sha256 pin lists audit_mixed_point.py; the mixed-point and early-sharing mathematical payloads stay fieldwise identical to the committed files.

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 scalar_check on the fixed mixed-point circuit set gave median wall time 10.004 → 4.444 seconds (2.25×). Timing is host-descriptive and is not a correctness gate. Every scalar receipt matches the baseline checker fieldwise (roles, circuit_sha256, flags).

Package: research/mixed-point-scalar-digest/.

Validation

python3.12 -I -B research/mixed-point-scalar-digest/verify.py
python3.12 -I -O -B research/mixed-point-scalar-digest/verify.py
python3.12 -B scripts/research_package.py --package research/mixed-point-scalar-digest
python3.12 -I -B research/mixed-point-scalar-digest/benchmark.py --rounds 5

Expected:

PASS_MIXED_POINT_SCALAR_DIGEST scalar_cases=7 encoding_cases=5 certs=0 no_new_kappa
PASS_REGISTERED_RESEARCH_PACKAGE research/mixed-point-scalar-digest outcome=completed
PASS_IDENTICAL_MIXED_POINT_SCALAR_RECEIPTS

Default verify is read-only. 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): CrocSwap/integer-mult-bounds#756, CrocSwap/integer-mult-bounds#1793, CrocSwap/integer-mult-bounds#1861, CrocSwap/integer-mult-bounds#1922, CrocSwap/integer-mult-bounds#1217, and research/rational-containment-performance/ (mixed-point contained adapter). Open CrocSwap/integer-mult-bounds#1874 optimizes a different hosted-audit digest path and does not touch scalar_check. No competing open PR on the scalar_check digest object at refresh.

Douglas R. Colkitt authored the mixed-point and early-sharing screens and original checkers. Prepared by Wentao Li (VictorLiwentao, @Is__Victor) with substantial AI assistance, under Apache-2.0.

[Mister-Meeseeks]

Batch review at 3221e1066aacf2f96d18d9989fcd02b93ff337f2.

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

Reviewed scope: Shared-point scalar digest optimization preserves canonical bytes for the complete declared scalar and encoding fixtures against the pinned prior implementation. Assertion-enabled replay, optimized replay and regression tests pass. Author timing ratios are not independently benchmarked; no mathematical claim.

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.