Repository navigation
feat(wordram): compute the exact Chan–He two-prime modulus natively #1501
Description
Complete bounded composition component of #1446, above #1497, under the open #1394 endpoint. Owner codex-chanhe-native-t3-28da2261; implementation is in the owner's existing isolated checkout, with a distinct component branch recorded before submission. This scope was implemented under the existing #1446 claim.
Given the shared prime bitmap/count produced by PrimeTable and three injective signed-encoded input arrays, compute every first threshold 3Λnⱼ², run the complete native least-prime search, save p, compute all three baseline collision counts at p, compute second thresholds 3Λ coll(Sⱼ,p), run the native second search at moduli pq, and return exactly existing ChanHe.modulus. Every setup, copy, encoding, arithmetic operation, query and loop is charged; fixed 64-register code does not read the word width. Prove exact output, caller/input frame and a polynomial scalar/address budget deriving all unsigned selector fits, with explicit signed-input/maximum-modulus fits. Execute an empty-set case and check fully admissible native contracts for lengths 0/1/2 and rejected-prime/default behavior with negative inputs. The output product is unsigned; downstream encoded-modulus delivery remains explicit.
Acceptance: complete concrete composition, not assumed prime-choice/counting procedures; all six gates/static checks, checked source comparator/notation/model/encoding/resource mapping, reuse/source credit, independent nonauthoring math/API review and outside coordinator request. No curated headline impact is expected.
This does not finish #1446: shared-table/State.init integration, exact frontend, heavy/packed-node integration, bounded recursion, (n+2)^κ size/asymptotic bridges and global logarithmic width/subquadratic time remain. Native Exact Triangle/Convolution host and final ordinary distinct-position 1.9992 composition remain #1394.
Activity
edemaine commentedon Oct 28, 2026
active —
codex-chanhe-native-t3-28da2261, brancht3/native-chan-he-modulusabove #1497 0e05605. The complete concrete two-search/count/threshold/product theorem and word-width independence elaborate. An executed full empty-set case and fully admissible native contracts for lengths 0/1/2 plus a negative rejected-prime/default case now pass diagnostics. Source bytes are being frozen for independent review and full validation; this component does not finish #1446.edemaine commentedon Oct 28, 2026
active — final complete six-gate run passed after the independent review's two small cleanup suggestions. No program, public statement or definition changed in that cleanup. Global audit window RELEASED to #1481; remaining work here is static metadata checks and submission. Final evidence will live in the PR.
edemaine commentedon Oct 28, 2026
awaiting review —
codex-chanhe-native-t3-28da2261: #1506 is submitted above #1497 with the complete native two-prime composition. Final source mapping, all six/static gates, nonauthoring review and remaining local/global conventions are recorded once in its description; outside coordinator review is formally requested. This bounded component does not close #1446/#1394.edemaine commentedon Oct 28, 2026
codex-chanhe-native-t3-28da2261 active: read the exact a5a5391 outside FINAL/no-blockers review. I am taking the permitted documentation-only option on existing t3/native-chan-he-modulus: explain the intended nonempty Chan–He prime-window interface behind m≥2 and reflow UPSTREAM.md. The public hypothesis, native code and theorem statements stay unchanged; no ancestor edits/restack. Validation will be serial after the currently held #1496 global-audit window.
edemaine commentedon Oct 28, 2026
codex-chanhe-native-t3-28da2261 awaiting review: documentation-only #1506 cleanup pushed normally at 1085648. All six gates/full hygiene/current-base changed-file/name/diff audits pass, with native/public statements unchanged. Focused outside documentation review requested. Global audit window released; full #1446/#1394 scope remains open.
SamuelSchlesinger commentedon Oct 29, 2026
Completed: concrete native two-prime modulus computation with charged execution and frames. Landed through #1570; verified on main with accepted source, contributor credit and API follow-ups retained. The broader #1394 objective, concrete fast ET backend and complete ordinary-3SUM/1.9992 endpoint remain open.