🌐 [translation-sync] FIX: lecture cleanup batch from the 2026-08 tracker audit

[mmcky]

Automated Translation Sync

This PR contains automated translations from QuantEcon/lecture-python-programming.

Source PR

#595 - FIX: lecture cleanup batch from the 2026-08 tracker audit

Files Updated

• ✏️ lectures/about_py.md
• ✏️ .translate/state/about_py.md.yml
• ✏️ lectures/autodiff.md
• ✏️ .translate/state/autodiff.md.yml
• ✏️ lectures/pandas.md
• ✏️ .translate/state/pandas.md.yml
• ✏️ lectures/pandas_panel.md
• ✏️ .translate/state/pandas_panel.md.yml
• ✏️ lectures/polars.md
• ✏️ .translate/state/polars.md.yml
• ✏️ lectures/python_by_example.md
• ✏️ .translate/state/python_by_example.md.yml

Details

• Source Language: en
• Target Language: fr
• Model: claude-sonnet-5

---

This PR was created automatically by the translation action.

[github-actions]

✅ Translation Quality Review

Verdict: PASS | Model: claude-sonnet-5 | Date: 2026-08-12
Routing: editor — 1 major finding(s); 2 minor finding(s) in gating categories (accuracy/terminology/syntax/diff-check/other); accuracy 8 below floor 9
Shadow gate: would NOT auto-merge (recorded only; no action taken)

---

📝 Translation Quality

Criterion	Score
Accuracy	8/10
Fluency	8/10
Terminology	9/10
Formatting	9/10
Overall	8.4/10

Summary: The translation is technically strong, fluent, and terminologically consistent, correctly following the glossary and French typographic conventions. However, one code-behavior discrepancy in pandas.md (df.where(...) argument) constitutes a real accuracy issue that changes program output versus the source, and unexplained future_stack=True additions in pandas_panel.md deviate from the source code without corresponding explanation in the English text. Aside from these, the changed sections read naturally and preserve meaning and formatting well. Consistent and accurate use of established French terminology (e.g., 'différentiation automatique', 'descente de gradient', 'Loi normale centrée réduite') throughout the changed sections. French typography rules (guillemets, non-breaking spaces before high punctuation) are correctly and consistently applied in headings, epigraphs, and dialogue. Mathematical notation, code blocks, and MyST directives are preserved intact with no structural or LaTeX corruption across all six files. Idiomatic rewording rather than calques, e.g. 'Objectif : simuler et représenter un processus de bruit blanc' for 'The Task: Plotting a White Noise Process', matching the required style guide.

Suggestions:

• [major · accuracy] lectures/pandas.md — ### Make Changes in DataFrames, item 1 (df.where()): The English source is df.where(df.POP >= 20000), which replaces non-matching rows with NaN by default. The French translation changes this to df.where(df.POP >= 20000, False), which replaces non-matching rows with False instead of NaN. This alters the code's behavior and contradicts the surrounding prose, which says the remaining rows are replaced with NaN. → df.where(df.POP >= 20000)
• [minor · accuracy] lectures/pandas_panel.md — ## Slicing and Reshaping Data, .stack() calls: Several .stack() calls in the French version have been changed to .stack(future_stack=True) (with an added inline comment explaining this is required for pandas>3.0), while the English source uses plain .stack(). This is a substantive code/behavior change not present in the source, going beyond translation. → If this is an intentional technical correction it should be flagged separately from the translation review; otherwise revert to .stack() to match the source exactly.
• [minor · terminology] lectures/about_py.md — ### Relative Popularity: The English source ends the sentence "especially those connected to AI" without a period, and the French translation adds a period, which is a minor but acceptable fluency improvement, not an issue. No action needed — included only for completeness, not a real finding.

---

🔍 Diff Quality

Check	Status
Scope Correct	✅
Position Correct	✅
Structure Preserved	✅
Heading-map Correct	✅
Overall	10/10

Summary: The translation sync correctly applied all source changes (code fixes, wording tweaks like Pytorch->PyTorch, pandas API updates, blank line removals) to the same files, same positions, and same structural sections while preserving and correctly maintaining the translation heading-map metadata.

Issues:

• In about_py.md, some blank lines between subsection headings (e.g., before '### Ne puis-je pas simplement utiliser des LLM ?') were removed compared to the source's spacing pattern, but this is a minor formatting difference that doesn't affect structure
• In pandas.md, df.where(df.POP >= 20000) changed to df.where(df.POP >= 20000, False) - this is a source-driven behavioral fix present in the English 'After' version, correctly mirrored

---

This review was generated automatically by action-translation review mode.

[Copilot]

Pull request overview

This PR syncs the French lecture content with the upstream English source PR (#595), applying a cleanup batch from the 2026-08 tracker audit and updating the translation state metadata accordingly.

Changes:

• Updates random-number generation examples to use NumPy’s default_rng() / Generator API (and adjusts accompanying explanations).
• Fixes/modernizes a few code examples (e.g., JAX key splitting) and updates data source URLs to raw.githubusercontent.com.
• Cleans up Markdown formatting artifacts (stray indented blank lines), and refreshes translation-sync state YAMLs (source SHA, model, tool version, sync date).

Reviewed changes

Copilot reviewed 12 out of 12 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
lectures/about_py.md	Removes formatting artifacts and fixes capitalization (e.g., “PyTorch”).
lectures/autodiff.md	Updates JAX random key handling to split once into dedicated subkeys; cleans formatting.
lectures/pandas.md	Switches examples to np.random.default_rng(); fixes footnote formatting.
lectures/pandas_panel.md	Updates dataset URLs to raw.githubusercontent.com.
lectures/polars.md	Switches examples to default_rng() / Generator methods; improves a percent-change expression to ignore null endpoints; fixes solution block formatting.
lectures/python_by_example.md	Updates white-noise example to use default_rng() and updates the explanatory text; cleans formatting artifacts.
.translate/state/about_py.md.yml	Updates translation sync metadata (source SHA, date, model, tool version, mode).
.translate/state/autodiff.md.yml	Updates translation sync metadata (source SHA, date, model, tool version, mode).
.translate/state/pandas.md.yml	Updates translation sync metadata (source SHA, date, model, tool version, mode).
.translate/state/pandas_panel.md.yml	Updates translation sync metadata (source SHA, date, model, tool version, mode).
.translate/state/polars.md.yml	Updates translation sync metadata (source SHA, date, model, tool version, mode).
.translate/state/python_by_example.md.yml	Updates translation sync metadata (source SHA, date, model, tool version, mode).

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[github-actions]

📖 Netlify Preview Ready!

Preview URL: https://pr-30--verdant-toffee-3261a8.netlify.app

Commit: 8ecf1c4

📚 Changed Lectures

• about_py
• autodiff
• pandas
• pandas_panel
• polars
• python_by_example

---

Build Info

• Workflow: Build Project [using jupyter-book]