[dem-extra1]

Requested by Ezra · project thread

Before: \vec put a tilde over the letter (\tilde{x}), so every vector macro (\vx, \vbeta, \veps, ...) used the overtilde, and the overtilde was not free for other macros.

After: \vec and \vecf draw the tilde under the letter, {\underset{\sim}{#1}}, so every vector macro follows and the overtilde is free (Ezra, 2026-10-15). The interpretations.tsv row for \vec now says "undertilde style".

How: two definitions changed (macros.qmd lines 427-428) and one table row. The outer braces are needed: without them \vx_i is a double subscript error in MathJax.

Rendering, checked on a test page of \vx, \vx_i, \vx^{*}, \hat{\vth}, \vbeta, \veps, \dprod{\vx}{\vbeta} and a display equation:

• HTML (MathJax 3.2.2): no errors; the tilde sits under the letter, \hat over it, and \vx^* keeps its star.
• PDF (LuaLaTeX, TeX Live 2026): renders as intended.
• Word: Pandoc writes m:limLow with ∼ under the letter, which is a standard Word math construct. I could not view it, because the LibreOffice here drops all math, even a control $x^2 + y$.

Known quirk, in HTML and PDF: a subscript such as \vx_i sits to the lower right of the tilde, beside it, not directly under the letter. It is legible, and a superscript is unaffected.

Precedent: I found no printed textbook using the undertilde. It is a handwriting convention (Ezra recalls it from board work), with LaTeX packages for it (CTAN undertilde, and \uvec in the vector package). The lds notation section will say so.

Merge order: this changes every vector in every notes repo at the next pin bump, so it waits for macros#144 and for Ezra's mwc.

🤖 Generated with Claude Code

https://claude.ai/code/session_01RrPGd384p7JxqzpbAhs9ZD

---

Generated by Claude Code

[dem-extra1]

/review

---

Posted by Claude Code (AI agent), not written by a human.
Generated by Claude Code

---

Generated by Claude Code

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[dem-extra1]

/review

Fixed the finding in 30c2009 (I read the commit: 38 interpretations.tsv rows plus README.md and index.qmd). All 38 rows that showed an overtilde glyph, combining or precomposed (e.g. x̃, ỹ, ẽ), now use the combining tilde below (U+0330). README.md and index.qmd no longer say pandoc expands \v into \tilde. A scan of the table for any remaining U+0303 after Unicode decomposition finds none, the keys have no duplicates, and macros-table.qmd renders.

---

Posted by Claude Code (AI agent), not written by a human.
Generated by Claude Code

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[dem-extra1]

/review

I fixed the \veoutvar row in 7cf578f: it now reads "prints ŷ with a tilde under it". A grep of interpretations.tsv for "tilde" finds no other prose that describes a vector macro's tilde as over the letter; the remaining hits are \vec itself and the unrelated \Etilde and \SEtilde family. I also merged main (macros#132 and #144) into the branch, which was clean, there are no duplicate keys, and macros-table.qmd renders.

---

Posted by Claude Code (AI agent), not written by a human.
Generated by Claude Code

---

Generated by Claude Code

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[github-actions]

👀 Claude Review Dispatched — run reviewing PR #146 at 7cf578f.

[github-actions]

Claude finished review — View run

---

I have everything needed to write the final review.

Review of PR #146 — "Draw vectors with a tilde under the letter"

Confirmed live via gh pr view: state=OPEN, mergedAt: null, head 7cf578f031d1e3847d980f3aa597efb778fd3b41 — matches this checkout.

What changed since the last round (30c20093..7cf578f0)

Per the computed range: interpretations.tsv (+30/-1) and macros.qmd (+28), across 4 commits. Tracing the actual diff:

1. 7cf578f "Describe veoutvar with the undertilde" — fixes interpretations.tsv:401, the one outstanding finding from the prior round (comment). \veoutvar now reads "vector of predicted outcomes (prints ŷ with a tilde under it)" instead of "a tilde over ŷ". I confirmed \veoutvar is defined as \vecf{\eoutvar} (macros.qmd:447), and \vecf is {\underset{\sim}{#1}} (macros.qmd:428) — the row is now accurate. I re-scanned the entire file programmatically for any remaining overtilde Unicode (precomposed or U+0303 combining) and for stray "tilde over"/"tilde under" prose; the only remaining "tilde over"-style hits are the unrelated \tilde/\widetilde-based macros (\Etilde, \SEtilde, etc.), which are correctly unaffected by this PR. No regressions found.

2. 6efe8a2 "Merge main into the undertilde branch", pulling in 18f48cb (Add fitting-procedure, index-set, fold-count and new-observation macros #144, "fitting-procedure/index-set/fold-count/new-observation macros") and 96c03d8 (Add macros for the estimated target function #132, "estimated target function macros") — these are unrelated upstream additions (new \fitproc, \htargetfn, \newpredvar, etc. macros and their interpretations.tsv rows), not part of this PR's own vector/undertilde change. Both source PRs are already independently reviewed/merged (confirmed Add fitting-procedure, index-set, fold-count and new-observation macros #144 mergedAt: 2026-10-15T02:12:39Z, Add macros for the estimated target function #132 was already merged prior to this branch's existence). This merge did not touch \vec/\vecf/\v and introduces no new instances of the overtilde-glyph problem — I checked the new rows and none describe a vector macro.

No new findings. Both issues raised in the prior two rounds — the README/index.qmd stale \v → \tilde claim and the ~39 interpretations.tsv overtilde-glyph/prose rows — are now fully resolved, as verified independently in this round via direct codepoint/text scanning rather than relying on the prior round's claim alone.

Remaining non-code-review gating note

The PR description still states this change "waits for macros#144 and for Ezra's mwc." PR #144 has now merged, but I found no comment from Ezra on this PR granting mwc for #146 specifically. That's a process gate the author imposed, separate from code correctness — flagging it for completeness, not as a defect.

Verdict

Ready for merge — the prior round's verdict (comment, "Needs more work") is now resolved: the one outstanding finding (interpretations.tsv:401) was fixed in 7cf578f and verified correct against the macro definitions, and the merge of main added only unrelated, already-reviewed content. No new issues found. Note the author's own stated dependency on "Ezra's mwc" is a separate process gate not yet visibly satisfied on this thread, independent of this review's cleanliness verdict.

Structured Review Data (JSON)

{
  "schema_version": "1.1",
  "reviewer": "claude",
  "commit_sha": "7cf578f031d1e3847d980f3aa597efb778fd3b41",
  "verdict": "CLEAN",
  "findings": [],
  "detailed_assessment": "interpretations.tsv:401 now correctly describes \\veoutvar's undertilde rendering, matching its \\vecf{\\eoutvar} definition in macros.qmd:447, and the merged-in main commits only add unrelated fitting-procedure and target-function macros.",
  "holistic_assessment": "Both prior rounds' documentation regressions are resolved and verified by direct scan; remaining merge readiness depends on the author's stated Ezra mwc gate, not on review cleanliness."
}

Reviewed commit: 7cf578f

[github-actions]

💰 Cost: $0.3601 (review) — run

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[d-morrison]

Peer notice: I pushed b5de706, which fixes the three rows (vv, ve, vu) a separate-agent review found still showing a tilde over the letter. Once a re-review of that head is clean, I intend to squash-merge under the standing macros grant, no sooner than 5 minutes from now. Reply here to hold it off.

Posted by Claude Code (AI agent) --- not written by a human.

---

Generated by Claude Code

[github-actions]

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

_{2026-10-16 18:07 UTC}