Repository navigation
Research: PR #3189's eighteenth-position QQ16 error witness survives the local native core - #3216
Merged
Mister-Meeseeks merged 1 commit intoOct 17, 2026
Conversation
…urvives the local native core Lifts PR CrocSwap#3189's B1,2 child-error witness through the literal local native suffix instead of treating non-scalar maps as identities. The Gaussian-integer direction w = T_236 ONE (entries in {-1,0,1}, squared norm 512) is injected after child 13489 on L344, and the 13,529-event suffix is executed exactly. After scalar 26472, L93 is exactly (-81921/262144) ONE: 17 fractional positions cannot represent this channel, while 18 suffice at every completed event for this direction. Final data errors are L75 = -ONE/2 and L93 = ONE/2; six borrowed helpers (L337, L338, L339, L341, L343, L346) retain errors and all private fields are zero, so an endpoint-only grid check misses the intermediate requirement. A separately implemented sparse Fraction engine matches the complete changed-field trace and every endpoint vector, and four reverse adjoints check the attaining and terminal components. This is a finite certificate for one direction at completed events under a completed-child defect model. Actual child rounding, recipe interiors, dirty-input bounds, helper restoration, global embedding and the paid consumer remain open; no supplier, selected-record or kappa change is claimed. Adds research/qq16-native-noise-lift/ only. Submitted by @commdata2338 with substantial AI assistance (OpenAI Codex research and implementation; packaged with Claude Code). Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01X86sCSXdoDd61SDFfC1PKG
This was referenced Oct 17, 2026
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Change and mathematical scope
PR3189's eighteenth-position child-error witness survives the literal B1,2 local native suffix, including changing frames and occupied residual adapters. Insert the explicit Gaussian-integer direction
w = T_236 ONEafter child 13489 on L344. At scalar cut 26472, every real component of L93 is exactly-81921/262144: 17 fractional positions fail, while 18 suffice at completed events for this one direction.Final data errors are
L75=-ONE/2,L93=ONE/2; six borrowed helpers retain errors, and all private error fields are zero. Endpoint-only grid checks miss the intermediate requirement. This is a finite certificate and scoped producer-interface audit, not an actual rounded child, a global precision theorem, an admitted supplier/consumer or a kappa improvement. Only eight files inresearch/qq16-native-noise-lift/are added; frozen inputs and the selected result are unchanged.Reviewer note (from an independent re-run during grading, not checked by this package): after every one of the 13,523 non-phase events the native error equals PR #3189's scalar coefficient times the register's frame vector, and all 2,442 distinct native child maps in the suffix carry frame vector to frame vector, so the survival follows from PR #3017's frame identities. The four residual adapters that touch the error run after the attaining event (events 26796-26799) and change only the final helper vectors, not the 18-bit witness.
Task and overlap
Task card:
joint-query-controller, priority revision 6; catalogue scope key42581d429acc9509d503, refined keyPR3189/B1_2/13489-L344/26472-L93/native-error-lattice. Full original card:TASK.json. The claim is in discussion #2727.463113d675af7410ac91d53a4fed51ec6494eb9a.research/qq16-native-noise-lift/, commands below. Credit and exact sources follow.70537595e46444c8a63455523d9851a5efe5538580fc72d14546088e9fce3c06dc12bc67270da0af0f493f4c264fbc70ee14364b68af260d858c143ce1e28b4070ae52efa5a9d0aa826bc588be7206f2526f2c242202b31b15e09a7502dfb8ff757b2badSOURCES.jsonpins the live QQ16 certificate, selected replay archive, PR3017 primitives and PR3189 receipt. The primary dispatcher adapts PR3017/3067 conventions. The separate checker imports no predecessor executable or primary implementation.Handoffs, current main and overlap were checked. The final offline snapshot covers 251 open PRs including drafts at 2026-10-17 17:37:12 UTC and 2,015 Discussion records at 17:39:04 UTC, with complete pagination.
13489 AND L344finds the predecessor and this claim/package. F0 rounding packages address different faults; concurrent #2727 F0 exact-grid and two-stage norm-envelope claims address different contracts. Cached metadata is non-atomic and does not cover every diff or unpublished result. Source-only archival intake is not scientific admission.Final canonical refresh: 2026-10-17 17:40:15 UTC, still
463113d675af7410ac91d53a4fed51ec6494eb9a, revision 6. No relevant guidance, selected record or scientific input changed.Validation
Linux / CPython 3.12.3, standard library only; all commands passed:
Expected markers:
Both implementations agree on the complete changed-field trace and every endpoint vector. Four reverse adjoints separately check the attaining component and terminal data/residual components. All nine behavioral groups passed in the final run (17.691 seconds). Controls cover native phase, injection frame/timing, scalar sign, residual orientation, source/site changes, unknown events, zero/scaling, nonzero adjoints and typed receipts. Committed-byte preflight checked eight package files plus four live inputs. Whitespace and exported-patch checks passed.
Unrun: full predecessor suites, selected cold replay, all 7,004 native error channels, child-interior finite arithmetic, global owner/embedding execution, formal proof kernel and hosted CI. The upper guard concerns completed events for one direction, not recipe interiors. The rounded-state consequence requires the ideal component to lie on the working grid; no unconditional rounded-state failure is asserted.
Credit
Prepared for commdata2338 (@commdata2338), with substantial OpenAI Codex assistance. Credit Tekalign for PR3189's question/witness and PR3017's reused exact primitives; dcharb78 for the ideal interior/adjoint investigation; Ross Richardson for the physical execution methodology; Nick Petalas for PR2965; Arham Tahir for earlier fault/rounding work; Ali Romman and the joint-query team, Sagar Thakrar, DLamdata, ywz65, starship-s, Douglas Colkitt and the retained selected/native lineage. Source bytes and notices remain unchanged. New work is Apache-2.0.
AI assistance: submitted by @commdata2338 with substantial AI assistance. OpenAI Codex (GPT-6 Astra) did the derivation, the implementation, the checking and the writing; it was packaged and submitted with Claude Code. No independent human review is claimed.
Open next questions
The README and TASK.json retain two handoffs: bind the actual child from frame 226 to 236 to a returned-value/defect lattice and dirty-input bound that excludes or prices this channel; then bind the local core to the actual global caller and prove borrowed-helper restoration while paying for native arithmetic, movement, storage and F17 consumption. A specific child/owner/consumer contract is the blocker. This local certificate does not justify a larger construction search.
Workflow feedback (optional)
The handoff-first procedure exposed a concrete missing implication in newer retained evidence. The card's
existing_checksis empty; linking PR3189's exact projected/native coverage and open implication there would reduce repeated reconstruction of older QQ16 work.🤖 Generated with Claude Code
https://claude.ai/code/session_01X86sCSXdoDd61SDFfC1PKG