perf(mmle): use matrix-vector quadrature reductions

[seonghobae]

Summary

Optimizes the tested NumPy reference fallback for unidimensional 2PL MMLE by replacing three broadcast-multiply reductions over quadrature nodes with mathematically equivalent matrix-vector products:

• resid @ nodes
• w @ nodes_sq
• w @ nodes

This avoids materializing temporary active-item × quadrature-node arrays for those reductions. Runtime effects are workload-, layout-, BLAS-, and threading-dependent; this PR makes no universal percentage-speedup claim.

Architecture boundary

The compiled Rust MMLE kernel remains the primary production path. This change only improves the existing NumPy reference/fallback path used when the extension is unavailable and retained for parity and diagnostic coverage. It adds no new Python-only likelihood or estimator.

Validation

• Exact-head CI, Security Scan, and SAST must pass.
• Existing MMLE recovery and Rust/NumPy parity contracts remain authoritative.
• CodeRabbit reported no actionable findings on the previous head; the documentation-only follow-up bounds the performance claim and corrects the record date.

No version bump or release is warranted for this isolated internal optimization; it can be included in the next release-ready vertical slice.

[google-labs-jules]

👋 Jules, reporting for duty! I'm here to lend a hand with this pull request.

When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down.

I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job!

For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with @jules. You can find this option in the Pull Request section of your global Jules UI settings. You can always switch back!

New to Jules? Learn more at jules.google/docs.

---

For security, I will only act on instructions from the user who triggered this task.

[coderabbitai]

📝 Walkthrough

Walkthrough

Changes

MMLE optimization

Layer / File(s)	Summary
MMLE gradient and Hessian reductions python/fast_mlsirm/estimators/mmle.py, .jules/bolt.md	MMLE gradient and Hessian calculations use dense matrix multiplication against quadrature nodes. Ridge penalties and optimization terms remain unchanged. The optimization note records the same replacement.

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

Possibly related PRs

• ContextualWisdomLab/fast-mlsirm#162: Optimizes related MMLE calculations through vectorized item-level Newton–Raphson updates.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
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.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the MMLE performance change to use matrix-vector quadrature reductions.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch bolt-mmle-dense-matmul-13930014294294156352

---

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