knowledge: D-MORTON-KA + D-SF-BASIN-1 GREEN — gridlake + episodic-basin measured ([H]→[G] ×2)

[AdaWorldAPI]

Summary

Records two parallel probes against the perturbation-cascade synthesis, both measured GREEN, both reusing shipped workspace primitives (the corrected-codec lesson applied by construction). Doc + board hygiene only; the measurements are the stockfish-rs probes.

D-MORTON-KA (spatial, D-SF-V3-3) — "NNUE 4096 = gridlake 4096" [H]→[G]

Re-projected HalfKA's piece-square axis onto the certified FacetTier::morton (2bit×2bit Z-order) and measured nearest-in-Morton vs nearest-in-feature over 24 (king × piece) FT-column configs on nn-1b6a82263149:

• feature locality IS spatial — board recall@8 = 0.440 (3.5× chance), ρ 0.407;
• Morton preserves a majority — Morton recall@8 = 0.347 (2.7× chance), ρ 0.328.

The NNUE HalfKA square addressing is a Morton-preserved gridlake. Morton ≤ board is expected (~79% of board's signal). Honest: locality-preserving, not a perfect isometry.

D-SF-BASIN-1 (episodic) — opening variants ARE le-contract basins

On the shipped FacetCascade/NodeGuid types + real chess rules: a 1.d4/1.c4 transposition to the same QGD position collapses to ONE NodeGuid (G1), which is memberof both basins {Queen's-Pawn, English} — the L2 many-basins-one-node round-trip (G2); the L1 part_of:is_a rails cluster lines by opening (hi_distance 5 < 6, G3). Promotes the "Opening variants" and "Transpositions" rows [H]→[G]. Honest: the facet lane isn't yet a persisted ValueTenant — encoding proven, persistence future.

Changes

• .claude/knowledge/stockfish-nnue-as-perturbation-cascade.md — D-SF-V3-3 → GREEN; the two temporal table rows → [G]; D-SF-BASIN-1 deliverable → GREEN, each with the measured result + honest scope.
• .claude/board/EPIPHANIES.md — E-CHESS-GRIDLAKE-BASIN-GREEN-1 prepended (both probes; the substrate-thesis status roll-up).

Substrate thesis after this pair: temporal (D-SF-EPISODIC-1), palette-tenant (D-PALETTE-NNUE), Morton-gridlake (D-MORTON-KA), episodic-basin (D-SF-BASIN-1) all GREEN. Doc stays SYNTHESIS-not-canon — four rows promoted, none overclaimed; D-SF-RUNG-1 / D-SF-ARIGRAPH-1 / D-SF-V3-4 / D-PALETTE-NNUE-VEC remain [H].

Testing

Doc + board only. Measurements are the stockfish-rs probes (morton_ka.rs, basin_transposition.rs), both fmt/clippy clean and net-/CI-safe.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Awg6TXocHcwTtc6eGsHcdD

---

Generated by Claude Code

Summary by CodeRabbit

• Documentation
  • Added measured findings on NNUE square addressing and Morton-preserved grid locality.
  • Documented opening variants and transpositions as confirmed episodic basin behaviors.
  • Clarified node consolidation, taxonomy clustering, and current persistence limitations.
  • Updated synthesis and verification status for Stockfish integration findings.

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: serverGenReqId_9f20746c-1e2c-4718-825d-334cf8ba0888)

[coderabbitai]

📝 Walkthrough

Walkthrough

The PR updates project knowledge documents with measured results for Morton-preserved NNUE locality and opening-variant basin/transposition behavior, and records these findings in the epiphanies synthesis board.

Changes

Substrate findings

Layer / File(s)	Summary
NNUE locality measurement .claude/knowledge/stockfish-nnue-as-perturbation-cascade.md	D-SF-V3-3 now records HalfKA reprojection details, measurements across 24 configurations, recall/correlation metrics, and locality-preservation wording.
Opening basin and transposition measurements .claude/knowledge/stockfish-nnue-as-perturbation-cascade.md	Opening variants and transpositions are promoted to measured confirmations, while D-SF-BASIN-1 documents transposition identity, basin membership, taxonomy clustering, and the in-memory persistence scope.
Synthesis status update .claude/board/EPIPHANIES.md	Adds the dated findings and updates promoted GREEN rows and remaining H items.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• AdaWorldAPI/lance-graph#639: Updates overlapping epiphany findings and GREEN status entries.
• AdaWorldAPI/lance-graph#679: Introduces the related Stockfish perturbation-cascade knowledge document.

Poem

I hop through Morton squares so neat,
While opening basins gently meet.
One node gathers paths anew,
GREEN findings bloom in morning dew.
Documentation dances—thump, thump, thump!

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately reflects the two measured GREEN probes and the [H]→[G] promotions described in the changeset.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In @.claude/knowledge/stockfish-nnue-as-perturbation-cascade.md:
- Around line 200-201: Clarify the D-SF-BASIN-1 notes around the transposition
result to distinguish the probe-local HashMap<NodeGuid, …> multimap from the
shipped graph contract. State that mailbox_scan.rs provides only one
Option<BasinOf> parent per row, and avoid presenting one NodeGuid as
production-shipped membership in both basins; retain the explicit HashMap scope
where the many-basins-one-node observation is described.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: e26531f7-a040-46e5-a9d7-c46e7d28f509

📥 Commits

Reviewing files that changed from the base of the PR and between 6153293 and 6c0b038.

📒 Files selected for processing (2)

• .claude/board/EPIPHANIES.md
• .claude/knowledge/stockfish-nnue-as-perturbation-cascade.md