sigker: wire signature_kernel_pde to ndarray's SIMD signature_pde_sweep, W1.5 gate closed

[AdaWorldAPI]

Summary

Consumer side of ndarray PR #293 (hpc::signature_pde::signature_pde_sweep, merged). sigker::signature_kernel_pde now delegates to the general-dimension SIMD wavefront instead of its hand-rolled row-major scalar loop (TD-PILLAR11-SCIENTIFIC-LOOPS-BYPASS-NDARRAY-SIMD-1). sigker gains a plain, mandatory ndarray path dep — the same operator ruling ("ndarray is mandatory everywhere numeric/computational code runs") already applied to jc's Cargo.toml earlier this arc, now applied here too.

• All 40 sigker lib tests pass unchanged, including every kernel::tests::pde_kernel_* numerical-convergence test.
• All 144 jc lib tests pass (feature hambly-lyons), including hambly_lyons::tests::pillar_passes — the certificate that exercises signature_kernel_pde through the Goursat depth-∞ leg.

Doc corrections (two stale claims, both said Pillar 11 "activates once benchmarked" when it already had)

• crates/sigker/src/lib.rs — jc's own src/lib.rs has recorded Pillar 11 activated since PR feat(jc): pillar 10 (Pflug-Pichler) + sigker crate + pillar 11 stub #348.
• .claude/knowledge/ndarray-vertical-simd-alien-magic.md's W1.5 section — item feat(graph): add SPO triple store with bitmap ANN, TruthGate, semirin… #6 marked SHIPPED, Arrow 57 / DataFusion 51 / Lance 2 + BlasGraph Algebra + SPO Triple Store #7/feat: add Belichtungsmesser HDR popcount-stacking early-exit cascade #8 marked gate-open-not-yet-built.

Board hygiene

• TECH_DEBT.md: TD-NDARRAY-SIMD-SIGNATURE-PDE-SWEEP marked SHIPPED (append-only — original scoped f32/F32x16 API surface kept alongside what actually landed, a deliberate f64/Vec<Vec<f64>> deviation once it was clear that's sigker's real carrier type).
• EPIPHANIES.md: new entry E-SIGNATURE-PDE-SWEEP-SHIPPED-W1.5-GATE-WAS-QUIETLY-OPEN-1 — both W1.5 activation clauses were independently true before this session touched anything; they'd just never been checked together, and one of them (the "benchmarked at production carrier widths" clause) meant an existing benchmark had simply never been run.

Probe fallout, fixed

crates/jc/examples/goursat_substrate_probe.rs's own A0 arm (signature_kernel_pde) is now itself SIMD-backed by construction, so the probe's original "A0 = A1 bit-exact" assertion no longer holds — A0 has effectively become a second, independent implementation of the wavefront. Restated the contract: A1 is the untouched scalar oracle; A0 (shipped, general-dimension) and A2 (the probe's own from-scratch dim=2 copy) each agree with A1 within the same predeclared tolerance used throughout this arc. Rerun confirms A0 and A2 produce numerically identical relative error against A1 at every size — cross-validating ndarray's shipped primitive against this probe's independent implementation.

Test plan

• cargo test --release --manifest-path crates/sigker/Cargo.toml --lib — 40/40 pass.
• cargo test --release --manifest-path crates/jc/Cargo.toml --features hambly-lyons --lib — 144/144 pass.
• cargo fmt --check / targeted cargo clippy -- -D warnings on the touched files (kernel.rs, the probe) — clean; confirmed the few pre-existing clippy hits elsewhere in sigker/jc are unrelated to this diff.
• cargo run --release --manifest-path crates/jc/Cargo.toml --features hambly-lyons --example goursat_substrate_probe — passes with the corrected contract.
• cargo run --release --manifest-path crates/jc/Cargo.toml --features hambly-lyons --example w5_trigger_check — W5 correctly stays HOLD (memory trigger unchanged at ~11585; time trigger re-derived higher since per-cell cost dropped, exactly as the file's own "never hardcode, always re-derive" design intends).

Branch note

This branch's prior PR (#1140) merged to main before this commit was ready. Per this repo's branch-reuse convention, the branch was restarted from fresh main and this one commit cherry-picked on top (force-with-lease push, no data loss — the discarded history was already fully merged).

🤖 Generated with Claude Code

https://claude.ai/code/session_01KCGhDYoQBXs3poaR7sFuqp

---

Generated by Claude Code

[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_fbfe830d-e154-47a3-92e0-bed60f6e9407)

[coderabbitai]

Warning

Review limit reached

Next included review available in 11 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available. Your 81 included PR review attempts over the past 7 days set your current allowance at 1 review per hour.

Your organization has reached its usage spending cap. Adjust your spending cap in the billing tab.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Essentials

Run ID: b82bc706-75c8-45b7-ad82-cf6b39a05ee7

📥 Commits

Reviewing files that changed from the base of the PR and between 6a7c985 and d69ea53.

⛔ Files ignored due to path filters (1)

• crates/jc/Cargo.lock is excluded by !**/*.lock

📒 Files selected for processing (7)

• .claude/board/EPIPHANIES.md
• .claude/board/TECH_DEBT.md
• .claude/knowledge/ndarray-vertical-simd-alien-magic.md
• crates/jc/examples/goursat_substrate_probe.rs
• crates/sigker/Cargo.toml
• crates/sigker/src/kernel.rs
• crates/sigker/src/lib.rs

---

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