CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

What this is

A Quarto static website (https://www.diiv.io) — a personal learning site for self-directed study. There is no application code: every page is a .qmd (Quarto Markdown) file rendered to HTML. Content is split into course syllabi + per-week lecture notes and worked textbook exercise sets. The heavy lifting is math typesetting (KaTeX), so most “code” is LaTeX inside Markdown.

Commands

Quarto must be installed (quarto --version). All commands run from the repo root.

  • quarto preview — live-reloading local server; the primary dev loop. Renders on save.
  • quarto render — full build to _site/ (the output dir, gitignored).
  • quarto render course1/week01.qmd — render a single page (fast iteration on one file).
  • quarto check — verify the Quarto install / environment.

There are no tests, linters, or build scripts beyond Quarto itself. Deployment is the rendered _site/ (GitHub Pages via CNAME → www.diiv.io). execute: freeze: auto means rendered output is cached; an explicit quarto render refreshes it.

Architecture & conventions

Config: _quarto.yml is the single source of truth for site title, navbar (left: Courses, Books / right: About, GitHub), theme (litera), and math engine (katex). Custom styling lives in styles.css (.diiv-card, KaTeX sizing, container width).

Top-level pages: index.qmd (landing/bio), about.qmd, courses/index.qmd (course directory), books/index.qmd (book directory).

Four courses, each a folder courseN/ with an index.qmd syllabus plus weekNN.qmd lecture notes: - course1/ — Mathematical & Theoretical Foundations (pure theory, 20 weeks, week01week20) - course2/ — Microelectronic Circuits & Signal Processing (bench-based, 10 weeks) - course3/ — Computer Graphics, Vision & Autonomous Robotics (10 weeks) - course4/ — NLP + Deep Learning + LLM (10 weeks)

Courses 1 & 2 are treated as mastered prerequisites by Courses 3 & 4 (their concepts are referenced, never re-taught). See memory/MEMORY.md for the full phase/week breakdown of each course — it is the authoritative map and is kept current.

Books: books/<book-slug>/ holds exercise sets, organized by category (Math, Physics, CS, Computer Vision, Robotics/AI, NLP, EE). Most are stubs (index.qmd only); finished ones have chapter subfolders. Two layout patterns for worked sets: - Axler LADR: per-subsection files — chNN/index.qmd + exercises-1a.qmd, exercises-1b.qmd, … - Ross: per-section files — chNN/index.qmd + exercises-s1.qmd, exercises-s2.qmd, … (Ross numbers exercises §section.exercise)

books/index.qmd lists every book with a **Minimum** line (bare-minimum chapters to read) and marks finished sets **✓ done** with the specific worked problems inline. Course week notes link here for exercises rather than listing them.

Content templates (match these when adding pages)

Week note (courseN/weekNN.qmd): YAML title → ← [Course N syllabus](index.qmd) backlink → ## Overview## Readings## Key Concepts (KaTeX-heavy) → ## Theory Exercises/## Exercises (links to the relevant books/ page; some weeks omit this) → ## Connections## Further Reading. Course-specific variants exist (Course 2 adds Lab-Bench Work / Measurement Methodology; Course 3 adds Implementation / Benchmark).

Exercise file: YAML title → a .callout-note block with anchor links ([Exercise N](#ex-N)) → per-exercise []{#ex-N} anchors, each followed by Exercise N. statement and Proof. with aligned LaTeX (\begin{aligned} with && \text{[justification]} columns). Mirror the formatting of an existing finished set (e.g. books/axler-ladr/ch01/exercises-1a.qmd).

Content authorship rules (important context, not editorial license)

Per the owner: weekly lesson titles and syllabus summaries were AI-assisted, and the site skeleton was AI-scaffolded — but all proofs, exercises, lecture-note prose, and solutions are hand-written by the owner. Proofs are done by hand on paper, then converted to LaTeX. When asked to add content, default to scaffolding/typesetting/formatting help; do not invent mathematical content or worked solutions unless explicitly asked. Worked solutions are intentionally omitted from many syllabus exercise sections (left as student exercise) — don’t “helpfully” fill them in.

Memory

memory/MEMORY.md (auto-loaded each session) holds the detailed, frequently-updated map of every course’s phases/weeks and every book’s status. Consult it before making structural changes, and update it when course/book structure changes — the repo’s organization has been renumbered/merged several times and that history matters.