[book-dp1] Chapter 8 HTML/PDF comparison issues

[Chihiro2000GitHub]

Hi @mmcky @jstac

Below are the differences I found between the PDF and the HTML in Chapter 8.

1. Double parentheses in equation references in the proof of Proposition 8.2.3

PDF: In the proof of Proposition 8.2.3, equation references appear with a single pair of parentheses, such as (8.28), (8.29), (8.30), and (8.31).

HTML: In the corresponding proof block, the same equation references appear with double parentheses, such as ((8.28)), ((8.29)), ((8.30)), and ((8.31)).

Issue: This appears to be a cross-reference rendering issue. The HTML should probably match the PDF and use a single pair of parentheses for these equation references.

2. Lettered condition labels rendered as bullet points in Section 8.3.2.1

PDF: In Section 8.3.2.1, the four assumptions are listed with lettered labels (a), (b), (c), and (d). The following paragraph then refers to them as “Condition (a)”, “Conditions (b) and (c)”, and “Condition (d)”.

HTML: In the corresponding part of the HTML, the four assumptions are rendered as ordinary bullet points rather than as (a)–(d).

Issue: Since the following text refers explicitly to Conditions (a)–(d), the HTML should preserve the lettered labels used in the PDF.

Additional note

This is not a major discrepancy between the PDF and the HTML, but I noticed that the equation number for (8.50) appears slightly lower than equation numbers usually do in the HTML, as shown below. The formula itself is still readable, so this may be only a minor layout issue, but it might be worth checking whether something unusual is happening in the rendering of this equation.

[jstac]

Thanks @Chihiro2000GitHub , great work.

With the lower equation numbers, does that only occur when there's a long equation, like this one?

In my browser it appears centered:

[Chihiro2000GitHub]

Thanks, @jstac.

I just checked a few more places, and it seems that this is not limited to long equations or to Chapter 8. I found similar lower-positioned equation numbers in some chapters as well.

I don’t remember noticing this when I reviewed those chapters earlier, so it is possible that this appeared after a recent update.

But since it looks centered in your browser, it may well be an issue specific to my browser.

I’ll attach a couple of screenshots below just for reference.

Example 1: A long equation in another chapter

Example 2 A shorter equation

[jstac]

Thanks @Chihiro2000GitHub . I'll leave this to @mmcky who is traveling but should be back on deck soon.

[mmcky]

Thanks @Chihiro2000GitHub — these all reproduce against the committed source, and I've triaged each. Two are tracked fixes; the third is an expected rendering characteristic. Details below.

1. Double parentheses ((8.28))–((8.31)) in the proof of Proposition 8.2.3 — confirmed, tracked

Root cause: these four references are written in the source as literal parens wrapping a bare \ref (e.g. (\ref{eq:safv})) rather than \eqref. In LaTeX the two are identical, but the HTML converter routes every equation reference to a role that supplies its own parentheses, so (\ref{eq:X}) comes out as ((N)). These are the only (\ref{eq:…}) instances in the book; everywhere else uses \eqref. The fix is a PDF-identical source change ((\ref{eq:X}) → \eqref{eq:X}, 6 occurrences in Chapter 8), tracked in QuantEcon/book-dp1#367 (needs @jstac, since it edits the .tex).

2. (a)–(d) condition labels render as bullets in §8.3.2.1 — confirmed, tracked

Root cause: in the source this is a manually labelled list (\item[(a)] … \item[(d)]). The converter has a pass that preserves such custom labels, but it currently only fires for enumerate, not itemize — so these labels are dropped to plain bullets, which orphans the "Condition (a)…/(b) and (c)…/(d)…" sentence that follows. This is a converter gap, tracked as a follow-up in QuantEcon/claude-latex-to-myst#178. (The same shape also affects one (A)/(B) list in Chapter 7, so the converter fix clears both.)

3. Equation number for (8.50) sits slightly low — confirmed, expected behaviour

Good eye, and thanks for flagging it as minor. (8.50) is a single-line but unusually tall equation (an inf with a subscript, a stretchy \left\{…\right\}, and a \sum). The HTML theme vertically centres the equation number against the full height of the equation block — so the taller the equation, the lower the centred number sits. This matches the PDF, where the equation tag is likewise vertically centred (which is why it looks centred/normal in @jstac's browser). It isn't tied to a specific chapter or to equation length per se — any sufficiently tall display equation shows the same thing.

So this is an expected rendering characteristic rather than a defect, and it's consistent with the PDF. The only alternative would be a theme change to pin the number to the equation's first line instead of its centre — but that would diverge from the PDF and could look worse on genuinely multi-line aligned equations, so I'd recommend leaving it as-is. I'm recording it as a known characteristic and not filing a change; happy to revisit if you'd prefer first-line alignment.

---

Summary: item 1 → QuantEcon/book-dp1#367 (source fix, @jstac), item 2 → QuantEcon/claude-latex-to-myst#178 (converter), item 3 → accepted as expected rendering. I'll follow up here once 1 and 2 are deployed.

[Chihiro2000GitHub]

Thanks, @mmcky

That all makes sense to me 😊