feat: assimilate statmath package macros (#90)

[dem-extra1]

Closes #90

Assimilates all 137 commands and notations from the CTAN statmath package (https://ctan.org/pkg/statmath) into macros.qmd and interpretations.tsv, adding bold Roman/Greek letters, blackboard bold symbols, convergence arrows, matrix operations, distribution families, expectation variants, operators, and estimator pairs.

[dem-extra1]

/review

[github-actions]

👀 Claude Review Dispatched — run reviewing PR #95 at ad980e9.

[github-actions]

PR Preview Action v1.8.1
Preview removed because the pull request was closed.

_{2026-09-26 21:28 UTC}

[github-actions]

Claude finished review — View run

---

Review: PR #95 — feat: assimilate statmath package macros (#90)

Scope: Adds 186 new macro definitions to macros.qmd (in the corresponding sections) and 186 matching entries to interpretations.tsv, porting the CTAN statmath package's notation (bold letters, blackboard-bold sets, convergence arrows, matrix operators, distribution operators, expectation variants) plus repo-specific \h.../\e... estimator-pair wrappers for Corr/MSE/Bias.

Verification performed

• Symmetry check: extracted all 186 macro names newly defined in macros.qmd and all 186 names newly added to interpretations.tsv — they match exactly 1:1 in both directions (no macro without an interpretation, no orphan interpretation entry). Satisfies the CONTRIBUTING.md rule "Add an entry for every new macro to interpretations.tsv — the macros-table.qmd build fails if any macro is missing an interpretation."
• Collision check: scanned every newly-defined macro name against the pre-existing macros.qmd (origin/main) for duplicate \def/\providecommand definitions. Found none — the pre-existing duplicate defs of \resid/\stdresid/\modresid at lines 559–561 predate this PR and are untouched.
• Correctly avoided a real collision: statmath's vectorization operator is actually named \vc (verified against the upstream statmath.sty: \DeclareMathOperator{\vc}{vec}), but this repo already defines \vc for something else (\providecommand{\vc}[1]{\boldsymbol{#1}}, bold vector notation). The PR correctly ported it under the alias \vct instead, avoiding a silent macro clobber. Good catch by the author — worth a one-line callout in the PR description since "assimilates all 137 commands" is otherwise a literal claim and this is the one renamed exception.
• Anti-hallucination check: fetched the real statmath.sty source and cross-checked \dBeta/\dBern/\dBin/\dF/\dGam/\dInvGam/\dInvW/\dLaplace/\dMN/\dN/\dPo/\dt/\dW/\dWeib, \Corr/\MSE/\Bias/\sign/\tr/\diag/\rank, \bbN/\bbZ/\bbQ/\bbR/\bbC, \inas/\indist/\inprob, \Ebar/\Ehat/\Etilde, \vcs/\vch, \bfzero — all match the upstream operator text exactly. Nothing appears fabricated.
• Convention compliance (per CONTRIBUTING.md):
  • No comments introduced into macros.qmd (required, since no comment syntax is safe across all four consumption paths).
  • \h.../\e... estimator pairs correctly mirror \hat/\widehat width choice: \hCorrt/\hMSEt/\hBiast use \widehat directly (multi-letter operatorname symbols), and the matching \eCorrt/\eMSEt/\eBiast correctly use \Est (the \widehat-composing indirection), exactly as the "mirror whichever width the original \h... macro chose" rule requires.
  • The \Ebarf/\Ehatf/\Etildef argument-forms use \sb{#1} (square brackets) rather than \paren{#1} like the other new "f"-suffixed forms — verified this matches the pre-existing \E/\Ef expectation convention (\providecommand{\E}[1]{\distop{E}\sb{#1}}), so this is correct, not an inconsistency.
  • The "duplicate expansion rather than delegate" pattern for aliases (e.g. \hCorr and \hCorrf both independently expand to \hCorrt\paren{#1} instead of one calling the other) matches an existing precedent in the file (\eSD/\eSDf at lines 696–697), so it's consistent with established style, not a new DRY violation.
• Syntax check: wrote a brace-balance scanner over every \def/\providecommand line in the file; the only imbalanced lines are pre-existing (\cbl by design, and the pre-existing multi-line \thmHessij), neither touched by this diff.
• TSV integrity: confirmed every line has exactly one tab (two fields), no trailing whitespace, and a trailing newline at EOF.
• CI: Quarto Preview Build, CodeQL, and Analyze all passed for this PR's head commit (ad980e9). The downstream workflow_run-triggered Quarto Check Equation Renders / Quarto Preview Deploy checks had not yet produced a run against this PR's branch as of this review (only against main's latest push), so I could not confirm the equation-render check specifically passes for this PR. Note that the most recent run of that check on main did fail with Missing argument for \operatorname at macros-table.html — but I traced every \operatorname{...} call this PR adds and all have well-formed, non-empty single-word arguments, so that failure looks pre-existing/unrelated to this PR's content. Worth a human double-check once that workflow does run against this commit, but nothing in the diff itself points to a rendering break.

Non-blocking nit

• interpretations.tsv:818,834,836 (\dBern, \dN, \dPo) append "(statmath prefix)" while the other 11 new d-prefixed distribution operators (\dBeta, \dBin, \dF, \dGam, \dInvGam, \dInvW, \dLaplace, \dMN, \dt, \dW, \dWeib) don't get the same annotation. Purely cosmetic — either drop it or apply it uniformly for consistency.

No correctness bugs, no CLAUDE.md-equivalent (CONTRIBUTING.md) violations, no naming collisions, and no hallucinated symbols found.

Verdict

Ready for merge (minor optional doc nit noted above; recommend confirming the Quarto Check Equation Renders workflow completes cleanly against this PR's own commit before merging, since that hadn't fired yet at review time).

Structured Review Data (JSON)

{
  "schema_version": "1.1",
  "reviewer": "claude",
  "commit_sha": "ad980e9c40496c8fbad3d621fe38ae190a3244d6",
  "verdict": "CLEAN",
  "findings": [],
  "detailed_assessment": "Checked interpretations.tsv and macros.qmd for symmetry, duplicate macro names, brace balance, and TSV formatting; found no defects in the 186 newly added macro definitions.",
  "holistic_assessment": "Verified statmath fidelity against upstream statmath.sty, confirmed CONTRIBUTING.md estimator-pair conventions are followed, and found no regression risk to existing macros aside from a correctly-handled vc/vct rename."
}

CI/fully-clean instrument correction

Per the check-pr-fully-clean.py gate (the authoritative instrument for a terminal clean claim, not gh pr checks):

python3 scripts/check-pr-fully-clean.py 95 -R Morrison-Lab/macros
→ exit 1, NOT fully clean:
  - Check run 'review / claude-review' is still in status 'in_progress'
  - No automated review comments or reviews found on PR #95

Both findings are self-referential to this very review run (the in-progress check run is this Claude Code Review workflow invocation, and no review has posted yet because this message is that review). I could not independently corroborate the check-runs endpoint directly (gh api .../check-runs) since gh api is disabled in this review-only session — only the wrapper script's read is available to me here. I have no basis to declare a code-defect-driven "Ready for merge"/"Needs more work" split beyond what's in the review body above; the content-level findings stand as written (no blocking defects found in the diff itself), but I'm not asserting the PR is externally/CI-fully-clean, since the one instrument that can authorize that claim currently returns NOT_CLEAN for reasons intrinsic to this review's own lifecycle rather than the code.

Verdict

Content review: no blocking findings (see full analysis above) — but the automated fully-clean gate currently reports NOT_CLEAN, solely because this review/check run had not yet completed/posted at the time of the gate check (self-referential), not due to any defect identified in interpretations.tsv or macros.qmd. Treat this as a progress report on the code content, not a terminal "Ready for merge" claim; re-run the gate after this review posts and the check run resolves.

Structured Review Data (JSON)

{
  "schema_version": "1.1",
  "reviewer": "claude",
  "commit_sha": "ad980e9c40496c8fbad3d621fe38ae190a3244d6",
  "verdict": "CLEAN",
  "findings": [],
  "detailed_assessment": "Checked interpretations.tsv and macros.qmd for symmetry, duplicate macro names, brace balance, and TSV formatting; found no defects in the 186 newly added macro definitions.",
  "holistic_assessment": "Verified statmath fidelity against upstream statmath.sty, confirmed CONTRIBUTING.md estimator-pair conventions are followed, and found no regression risk to existing macros aside from a correctly-handled vc/vct rename."
}

Stopping Point: Clean stopping point reached — full content review completed and posted; the fully-clean CI gate is self-referentially pending this review's own completion, which is expected and requires no further action from me in this turn.

Reviewed commit: ad980e9

[github-actions]

💰 Cost: $1.8324 (review) — run