You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
feat(exactTriangle): split modular queries and recover zero-triangle witnesses #1447
Mathematical implementation milestone of #1394, stacked after modular hashing #1445 and the lopsided interface #1441. Preserve the full native 3SUM endpoint, including exponent 1.9992.
Adapt the complete residue/chunk/piece reduction and witness recovery from pinned anthropics/formal-math@16a5ce4, ThreeSumApsp/Sec3/Theorem17/{Instances,Witnesses}.lean and only required Sec3/{Chunks,Parameters}.lean notions. Reuse the three supplied weight matrices, existing ExactTriangle.HasZeroTriangle, #1445 counts and LopsidedGraph. Do not recreate TriangleInstance or a broad problem universe. Prove unique coverage and size of pieces, residue/query chunks and total count bound, middle-size and adjacency-label characterization, count of instances and degree bound, and sound/complete witness recovery connecting actual lopsided answers to zero-weight triangles (with false-positive accounting where required). Extract only used parameter definitions.
Acceptance: complete finite reduction/recovery statements with exact source hypotheses/quantifiers, nonvacuous positive/negative/boundary examples, all six gates and hygiene, checked pinned source quotes/mapping, source credit, abstract assessment and outside-coordinator review request. No native program/time bound or fast solver is claimed. Tensor/sparse-query solvers, native programs, complete composition and improved Section4 route remain #1394. A partial interface alone is not completion of this milestone.
Delegator: codex-three-sum-continuation-t3-3dda8289. A distinct isolated delegated implementation worker will claim with its own session and actual branch before editing. Its mutable output ownership excludes native Chan–He work and the existing hashing implementation; there is no competing worker.
active — implementation worker codex-three-sum-splitting-t3-0a4d90c0 claims #1447 under codex-three-sum-continuation-t3-3dda8289 on branch t3/0a4d90c0, starting 8e182b0. Scope: complete finite residue/chunk/piece splitting, degree and instance bounds, and witness recovery/false-positive accounting. Native Chan–He and native solver/time work are excluded; #1394 remains open at its original 1.9992 endpoint.
codex-three-sum-splitting-t3-0a4d90c0 — branch is now t3/three-sum-splitting-witness-recovery (T3 assigned name). Finite construction and recovery adaptation are in progress; establishing isolated dependency/build caches before elaboration. Scope and #1394 gaps are unchanged.
active — codex-three-sum-splitting-t3-0a4d90c0: complete finite splitting and recovery modules compile, with exact integer/real parameter bridges and concrete boundary/false-positive/recovery examples. Full validation is in progress; the first aggregate build hit shared-host committed-memory pressure, so I released only my own completed MCP processes and requested resource coordination. No scope narrowing: #1447 still requires all six gates and review; #1394 remains open at 1.9992.
awaiting review — codex-three-sum-splitting-t3-0a4d90c0: complete finite splitting/recovery implementation is in #1455, stacked on #1445. All six gates and static checks pass; source mapping, nonvacuous tests, credit and abstract no-impact assessment are recorded in the PR. Non-authoring mathematical/API agent review is in progress; SamuelSchlesinger has been requested for outside coordinator review. This delivers #1447 only; the native 1.9992 endpoint and remaining work stay open in #1394.
awaiting review — codex-three-sum-splitting-t3-0a4d90c0: #1455 now includes the non-authoring math/API review record (no blockers) and its sole documentation nit is fixed. The final follow-up changes only ROADMAP/UPSTREAM; the tested Lean payload is unchanged. Outside coordinator review remains requested from SamuelSchlesinger. Evidence and exact revisions are kept in the PR; #1394 remains open.
active — delegator codex-three-sum-continuation-t3-3dda8289 resumes #1455 for the ancestor CI fix/restack on work/three-sum-splitting-restack-3dda8289. The completed finite implementation is preserved; I will validate the new head and request a delta review before submitting the later thin-product layer.
active — codex-three-sum-continuation-t3-3dda8289 is validating the #1455 restack above the fixed hashing layer. The small residue-API reuse fix and quotation clarification are implemented; focused builds and the full build/import/test gates pass so far. Next: finish lints, push and request delta review.
awaiting review — codex-three-sum-continuation-t3-3dda8289: restacked #1455 at dc4206a, inheriting the repaired header; reused Mathlib residue bounds and corrected source/cost wording. All six gates pass; final validation/review evidence is in #1455. Outside delta review requested. Native endpoint remains #1394.
triage-claude-e6cf7b0d-1 (Claude Code, Opus), issue triage for coordinator-claude-e6cf7b0d.
Closing as completed. This milestone landed in #1455 on 2026-10-29, and that PR records the validation and review against the acceptance criteria above. The broader native ordinary-3SUM endpoint (exponent 1.9992) stays open in #1394.
Activity
edemaine commentedon Oct 28, 2026
active — implementation worker codex-three-sum-splitting-t3-0a4d90c0 claims #1447 under codex-three-sum-continuation-t3-3dda8289 on branch t3/0a4d90c0, starting 8e182b0. Scope: complete finite residue/chunk/piece splitting, degree and instance bounds, and witness recovery/false-positive accounting. Native Chan–He and native solver/time work are excluded; #1394 remains open at its original 1.9992 endpoint.
edemaine commentedon Oct 28, 2026
codex-three-sum-splitting-t3-0a4d90c0 — branch is now t3/three-sum-splitting-witness-recovery (T3 assigned name). Finite construction and recovery adaptation are in progress; establishing isolated dependency/build caches before elaboration. Scope and #1394 gaps are unchanged.
edemaine commentedon Oct 28, 2026
active — codex-three-sum-splitting-t3-0a4d90c0: complete finite splitting and recovery modules compile, with exact integer/real parameter bridges and concrete boundary/false-positive/recovery examples. Full validation is in progress; the first aggregate build hit shared-host committed-memory pressure, so I released only my own completed MCP processes and requested resource coordination. No scope narrowing: #1447 still requires all six gates and review; #1394 remains open at 1.9992.
edemaine commentedon Oct 28, 2026
awaiting review — codex-three-sum-splitting-t3-0a4d90c0: complete finite splitting/recovery implementation is in #1455, stacked on #1445. All six gates and static checks pass; source mapping, nonvacuous tests, credit and abstract no-impact assessment are recorded in the PR. Non-authoring mathematical/API agent review is in progress; SamuelSchlesinger has been requested for outside coordinator review. This delivers #1447 only; the native 1.9992 endpoint and remaining work stay open in #1394.
edemaine commentedon Oct 28, 2026
awaiting review — codex-three-sum-splitting-t3-0a4d90c0: #1455 now includes the non-authoring math/API review record (no blockers) and its sole documentation nit is fixed. The final follow-up changes only ROADMAP/UPSTREAM; the tested Lean payload is unchanged. Outside coordinator review remains requested from SamuelSchlesinger. Evidence and exact revisions are kept in the PR; #1394 remains open.
edemaine commentedon Oct 28, 2026
active — delegator
codex-three-sum-continuation-t3-3dda8289resumes #1455 for the ancestor CI fix/restack onwork/three-sum-splitting-restack-3dda8289. The completed finite implementation is preserved; I will validate the new head and request a delta review before submitting the later thin-product layer.edemaine commentedon Oct 28, 2026
active —
codex-three-sum-continuation-t3-3dda8289is validating the #1455 restack above the fixed hashing layer. The small residue-API reuse fix and quotation clarification are implemented; focused builds and the full build/import/test gates pass so far. Next: finish lints, push and request delta review.edemaine commentedon Oct 28, 2026
awaiting review —
codex-three-sum-continuation-t3-3dda8289: restacked #1455 at dc4206a, inheriting the repaired header; reused Mathlib residue bounds and corrected source/cost wording. All six gates pass; final validation/review evidence is in #1455. Outside delta review requested. Native endpoint remains #1394.SamuelSchlesinger commentedon Oct 30, 2026
triage-claude-e6cf7b0d-1(Claude Code, Opus), issue triage forcoordinator-claude-e6cf7b0d.Closing as completed. This milestone landed in #1455 on 2026-10-29, and that PR records the validation and review against the acceptance criteria above. The broader native ordinary-3SUM endpoint (exponent 1.9992) stays open in #1394.