[book-dp1] Chapter 2 HTML/PDF comparison issues

[Chihiro2000GitHub]

Hi @mmcky @jstac

While comparing Chapter 2 of the MyST/HTML version against the PDF, I noticed the following discrepancies and possible rendering issues.

I would appreciate it if you could take a look when you have time!

1. Extra figure number/caption causes all Chapter 2 figure numbers to shift by one

• PDF: The conjugacy diagram appears without a figure number or caption.
• HTML: The same diagram is rendered as “Figure 2.1” with a caption, but the mathematical notation in the caption appears partly as plain text and is split awkwardly across lines, as shown below.

• Issue: The HTML seems to introduce an extra figure number and caption that are not present in the PDF, and the caption also has a math-rendering issue. This appears to shift all subsequent figure numbers in Chapter 2 by one.

For example:

• PDF: The Newton’s method figure is numbered as “Figure 2.1.”
• HTML: The same figure is numbered as “Figure 2.2.”

2. Minor formatting difference in exercise/theorem-like labels

• PDF: Exercise, proposition, lemma, and theorem labels include a period after the number, e.g. Exercise 2.1.2.
• HTML: The corresponding labels appear without the period after the number, e.g. Exercise 2.1.2
• Issue: The numbering itself seems correct, but the label formatting differs from the PDF. This may be a minor difference, but I wanted to flag it in case strict visual parity with the PDF is desired.

3. The partially ordered set conditions are split incorrectly

• PDF: In Section 2.2.1.1, the three conditions for a partial order are displayed together with their corresponding property names: “reflexivity”, “antisymmetry”, and “transitivity”.
• HTML: The three mathematical conditions are rendered as items 1–3, while the corresponding property names are incorrectly rendered as separate items 4–6, as shown below:

• Issue: The HTML separates the property names from the mathematical conditions they belong to, which changes the structure of the display and makes the list misleading.

This appears to be the same type of rendering issue as item 11 in the Chapter 1 issue (#26).

4. Raw LaTeX source appears in Example 2.2.2

• PDF: In Example 2.2.2, the definition of the pointwise partial order is rendered correctly as a mathematical display as follows:

• HTML: The corresponding display appears as raw LaTeX source: given $u, v$ in $\RR^\Xsf$, set $u \leq v$ if $u(x) \leq v(x)$ for all $x \in \Xsf$..
• Issue: The mathematical display in Example 2.2.2 is not rendered as math in the HTML. Instead, raw LaTeX source is shown directly.

5. The not-equal sign can break incorrectly on narrower screens in Exercise 2.2.8

• PDF: In Exercise 2.2.8, the condition is rendered correctly as u ≠ v.
• HTML: On a wide screen, the same condition is rendered correctly as u ≠ v. However, at some narrower browser widths, the not-equal sign breaks and appears as something like u / = v.
• Issue: The not-equal sign seems to be split incorrectly at certain narrower screen widths.

When the screen is wide:

When the screen is narrowed:

6. List numbering differs from the PDF and becomes inconsistent with references in the text

• PDF: Some theorem / lemma / exercise statements use “(i)”–“(v)” or “(a)”–“(d)”.
• HTML: The corresponding lists are rendered as ordinary numbered list items, such as “1.”–“5.”.
• Issue: The numbering style in the HTML differs from the PDF. More importantly, nearby text sometimes refers back to these items using the original PDF labels, such as “(i)”, “(v)”, or “(a)”–“(d)”. As a result, the list labels in the HTML become inconsistent with the references in the surrounding text.

Examples:

• In Lemma 2.2.1, the PDF numbers the statements as “(i)”–“(v)”, while the HTML renders them as “1.”–“5.”. The following paragraph still refers to “(i)” and “(v)”.
• The same issue also appears in Lemma 2.2.5, where statements numbered as “(i)” and “(ii)” in the PDF are rendered as “1.” and “2.” in the HTML.
• It also appears in Exercise 2.3.2, where the PDF uses roman numerals such as “(i)”, while the HTML renders them as numbered list items such as “1.”.
• In Lemma 2.3.5, the PDF labels the four items as “(a)”–“(d)”, while the HTML renders them as “1.”–“4.”. The text following the lemma refers to these items using “(a)”–“(d)”, so the HTML labels are inconsistent with the subsequent discussion.

This appears to be the same type of issue as item 15 in the Chapter 1 issue (#26).

7. Raw LaTeX source appears at the end of the paragraph before Blackwell’s condition

• PDF: In Section 2.2.3.3, the paragraph before the statement of Blackwell’s condition is rendered correctly.

• HTML: The corresponding text is broken and raw LaTeX source appears. The paragraph ends with something like implies $u, and the remaining part is rendered separately as a bullet-like item, e.g. c ¥in U$, as shown below:

• Issue: The inline math at the end of this paragraph is not rendered correctly in the HTML. It also seems to break the paragraph structure, with part of the expression appearing as a separate bullet item.

8. Proof environments are numbered in the HTML but not in the PDF

• PDF: Proof environments are not given numbers.
• HTML: Proof environments are numbered, e.g. “Proof 2.2.3”. In some cases, this appears immediately after a numbered proposition or lemma, such as “Lemma 2.2.4”.
• Issue: This differs from the PDF and may be slightly confusing, since the proof number does not match the number of the lemma being proved (as shown below). If the goal is strict parity with the PDF, it may be preferable not to number proof environments in the HTML.

9. Proof heading number does not match the proposition being proved

• PDF: The proof of Proposition 2.2.7 is not given a separate proof number.
• HTML: The proof is displayed under the heading “Proof 2.2.4”, while the proof itself begins with “Proof of Proposition 2.2.7.”
• Issue: This is potentially confusing because the proof heading number does not match the proposition number being proved. If proof environments are numbered independently in the HTML, readers may wonder what “Proof 2.2.4” refers to.

This is related to item 8 above, and appears to be similar in spirit to item 17 in the Chapter 1 issue (#26).

10. Some math symbols look different from the PDF

• PDF: Some symbols, such as the distribution symbol D(X) in the stochastic dominance paragraph and the self-map symbol S in Exercise 2.2.37, have a different style from the HTML.
• HTML: The corresponding symbols look visually different from the PDF.
• Issue: The meaning is probably unchanged, but the symbols do not visually match the PDF. This seems related to the symbol rendering issue reported in [book-dp1] Text and rendering discrepancies in Common Symbols table #24.

Kind regards,
Chihiro

[mmcky]

Many thanks @Chihiro2000GitHub — another excellent and precise audit. All 10 items verified against the LaTeX source and the conversion pipeline; every one of your observations checks out. Triage and tracking:

Tracking issue (book-dp1 side): QuantEcon/book-dp1#354

#	Finding	Disposition
1	Extra figure number/caption on the conjugacy diagram, shifting all ch2 figure numbers	✅ Confirmed — a book-side figure override added the number/caption (the PDF's diagram is an unnumbered display); the caption math is also mis-escaped. Will switch it to an unnumbered image, restoring Newton's method to Figure 2.1. → book-dp1#354
2	No period after "Exercise 2.1.2" etc.	📝 Documented as an accepted styling difference. The label template also drives inline cross-references, so adding the period would produce "see Exercise 2.3.4. for…" mid-sentence — worse than the mismatch.
3	Property names (reflexivity/antisymmetry/transitivity) split off as items 4–6	✅ Confirmed — same converter list-structure bug as #26 item 11; your ch2 sites added to QuantEcon/claude-latex-to-myst#111 (still open).
4	Raw LaTeX in Example 2.2.2	✅ Confirmed — \begin{center} leaks through as a fenced div the renderer treats as a code block (one more site found in ch. 6). → QuantEcon/claude-latex-to-myst#140
5	u ≠ v breaking as "u / = v" at narrow widths	✅ Confirmed — the source's \not= compiles to two separately-wrappable pieces in KaTeX; will be normalized to \neq (single glyph, cannot split). → QuantEcon/claude-latex-to-myst#142
6	(i)/(a) list labels rendered as 1./2., inconsistent with in-text references	✅ Confirmed — same as #26 item 15; your point that the prose back-references ("(i)", "(a)–(d)") go dangling is a good argument for prioritizing it. → QuantEcon/claude-latex-to-myst#111
7	Raw LaTeX + stray bullet before Blackwell's condition	✅ Confirmed — a source line wrap inside inline math turns into a Markdown bullet, splitting the expression. → QuantEcon/claude-latex-to-myst#141
8	Proofs numbered in HTML but not PDF	✅ Agreed — proofs will be unnumbered to match the PDF. → QuantEcon/claude-latex-to-myst#143
9	"Proof 2.2.4" heading vs "Proof of Proposition 2.2.7"	✅ Same root cause as 8 — once proofs are unnumbered the heading is plain "Proof" and the mismatch disappears. → QuantEcon/claude-latex-to-myst#143
10	D(X) / script S style differs from PDF	📝 Known limitation — the macros are faithful (\mathscr on both sides); the difference is the web math font vs the print font, same as resolved in #24. Meaning unchanged, no action planned.

Item 1 (the figure-number shift) is book-side and will land first; items 4, 5, 7, 8/9 go through the conversion tool and will reach the site with the next tool update; 3/6 remain with the open list-structure issue. We'll ping here for re-validation as the fixes deploy.

[mmcky]

@Chihiro2000GitHub item 1 is now fixed and live: the conjugacy diagram renders without a figure number or caption (matching the PDF), and all Chapter 2 figure numbers have shifted back — Newton's method is Figure 2.1 again, through Figure 2.14 at the end of the chapter. When you have a moment, a re-check of the figure numbering on https://quantecon.github.io/book-dp1/ch_fps would be much appreciated.

The remaining confirmed items (3, 4, 5, 6, 7, 8/9) go through the conversion tool and will reach the site with its next update — we'll ping here again when they deploy.

[mmcky]

@Chihiro2000GitHub all ten items from your Chapter 2 audit are now resolved on the live site (or documented where we deliberately accept a difference) — this is the consolidated wrap-up, and a re-validation pass of https://quantecon.github.io/book-dp1/ch-fps whenever convenient would be greatly appreciated.

#	Finding	Status
1	Extra figure number/caption on the conjugacy diagram shifting all ch2 figure numbers	✅ Fixed — diagram is unnumbered/uncaptioned as in the PDF; Newton's method is Figure 2.1 again (chapter runs 2.1–2.14).
2	No period after "Exercise 2.1.2"	📝 Accepted styling difference — the label template also drives inline cross-references, so adding the period would produce "see Exercise 2.3.4. for…" mid-sentence.
3	Property names (reflexivity/antisymmetry/transitivity) split off as items 4–6	✅ Fixed — the spurious numbering is gone; the conditions and property names display unnumbered, as in the PDF.
4	Raw LaTeX in Example 2.2.2	✅ Fixed — renders as mathematics.
5	u ≠ v breaking as "u / = v" at narrow widths	✅ Fixed — now a single ≠ glyph that cannot split across lines.
6	(i)/(a) list labels rendered as 1./2., inconsistent with in-text references	✅ Fixed — lists display the PDF's (i), (ii), … and (a)–(d) labels, so "by (i)" / "(a)–(d)" references in the surrounding text are consistent again. (One refinement still queued: the labels currently display without list indentation; proper indented list rendering follows with a pending site-theme upgrade.)
7	Raw LaTeX + stray bullet before Blackwell's condition	✅ Fixed — single intact paragraph.
8	Proofs numbered in HTML	✅ Fixed — all proofs are unnumbered, as in the PDF.
9	"Proof 2.2.4" heading vs "Proof of Proposition 2.2.7"	✅ Fixed with 8 — the heading is now plain "Proof".
10	D(X) / script S style differs	📝 Known web-font limitation (same resolution as #24) — the notation is identical; only the script typeface differs between KaTeX and the print font.

Thank you again for the careful, precise reports — items 1, 4, 5, and 7 in particular were conversion bugs we would not have caught without this comparison, and every one of them led to a fix in the shared conversion tool that now benefits the other QuantEcon book conversions as well.

[mmcky]

@Chihiro2000GitHub when you have time would you please review and verify. Thanks so much.

[Chihiro2000GitHub]

Many thanks for the updates, @mmcky!

I re-checked the items that were fixed on the live site, and overall the fixes look good to me.

I noticed just two small remaining points:

1.

For item 3, the property names still appear on separate lines rather than next to the corresponding mathematical conditions. In Section 2.2.1.1, “reflexivity”, “antisymmetry”, and “transitivity” are still displayed below the three conditions, rather than aligned with them as in the PDF.

2.

For items 8/9, the proof numbering issue looks resolved. In most cases, the heading is now simply “Proof”, which looks good to me.

However, for the proof immediately following Proposition 2.2.7, the proof block heading is “Proof”, and the body begins with “Proof of Proposition 2.2.7.” Since this proof block appears directly after Proposition 2.2.7, it seems clear from the context that it is the proof of that proposition. So I wonder whether the body could simply begin with the proof text, without repeating “Proof of Proposition 2.2.7.”

I wanted to flag it just in case. Please let me know what you think.

Thanks again for taking care of these quickly.