Help me learn to typeset equations like it was my job. (Why? Because it is.)

November 4, 2009 12:56 PM

I am looking for a math typesetting style guide. By this I don't mean the kind of stylesheet for journal submissions that says "Be sure to use the blah-blah-blah LaTeX package and the XYZ equation environment, and our army of editorial assistants will tie up the loose ends and knock off the rough edges." (Why not? Because my advisor is involved in starting a new journal, and suddenly my labmates and I are that army of editorial assistants.)

I am less interested in the technical details of mathematical typesetting. We've got our fonts chosen already, we're committed to using LaTeX and AMSMath which I speak pretty fluently, and we're distributing online so anything having to do with print is Not An Issue. In particular, I am not looking for another LaTeX user's manual — although if the advice I need happens to be buried in one, I'm okay with that. I'm also not particularly interested in simple questions of usage ("bigger parentheses or square brackets?" "~ or ¬ for negation?"), especially since a lot of those boil down to taste and convenience anyway.

I'm more interested in what you might call the visual semantics of it. (F'rinstance: How do you set a long equation so as to reveal its structure quickly and easily to the reader? How can spacing, line breaks, alignment and so on be used to produce that sort of clarity, and what other tricks are there that I'm not thinking of? What about a sequence of equations? A derivation or proof? How do you set a nonstandard symbol — an operator, function, etc. defined by the author;we get this a lot in my field — so that it's clear what its role in the equation is? This isn't a complete list of questions, but it's questions like that that I want to learn how to answer.) Aesthetic details — good spacing, good line breaks and page breaks, all-around symmetry and tidiness — are also important. The goal is to make these thorny and technical articles as easy and joyful to read as I possibly can.

Bonus points for a guide with good advice on the odd situations that come up in formal semantics and mathematical logic. (For instance, I've been unable to find any advice for laying out expressions in lambda calculus, or ones containing multiple quantifiers, and both of those are frequent sources of difficulty here.) But if that's asking too much, then I'm looking for general best practices that I can apply to the edge cases when they come up.

posted by nebulawindphone to media & arts (11 comments total)

I'm looking forward to the answer, if there is one. I'd love to have a general reference for mathematical typesetting. It's been practiced for hundreds of years prior to the arrival of LaTeX, and I'm sure there's a classic text out there.

As an AMS LaTeX user you're probably already aware of the standard LaTeX-specific reference Math into LaTeX. Also, AJ Hildebrand's LaTex Tips are probably beneath someone of your experience, but are worth linking for any neophytes reading.
posted by rlk at 1:17 PM on November 4, 2009

It's been practiced for hundreds of years prior to the arrival of LaTeX, and I'm sure there's a classic text out there.

Yeah. In fact, I probably coulda done a lot less rambling in my question if I'd just asked for that classic text that I'm also pretty sure must exist....
posted by nebulawindphone at 1:37 PM on November 4, 2009

I haven't seen the latest versions, but I think the Scientific WorkPlace software might come with manuals of what you're looking for.
posted by Melismata at 1:47 PM on November 4, 2009

Math into Type: The Next Generation

I'm actually not familiar with this unfortunately now out-of-print edition, but the previous and likewise out-of-print edition was my bible when I was doing composition and layout for math-heavy college textbooks in the early 1990s. For what you're interested in, I think it'd be worth tracking down a copy.
posted by drlith at 7:10 PM on November 4, 2009

Ah, I just found "Mathematics into Type: Updated Edition" after a bit of digging, which I'm now convinced is the revised edition of the text I used to use.
posted by drlith at 7:24 PM on November 4, 2009

Check out Stefan M. Moser's How to Typeset Equations in LaTeX [PDF]. It assumes the typesetter's using the IEEEeqnarray environment, but the guidelines, splitting rules, and what to do in x situation might be what you're looking for.
posted by jenh at 7:29 PM on November 4, 2009

One of the problems is that there doesn't seem to be a standard way to write lambda expressions. For example, Heim and Kratzer define [[love]] as λx . [λy . y loves x] (with dots) whereas Chierchia and McConnell-Ginet write λxλy[y loves x] (without the dots; different bracketing). And then there are the argument positions. Some papers seem to hold that you apply the arguments outside in, and some go in a strict left-to-right correspondence. For example, the expression λxλy[y loves x](John)(Mary) sometimes means "Mary loves John" and sometimes means "John loves Mary." This is usually figure-out-able in the context of a given paper, but it is frustrating that there doesn't seem to be a standard. These are decisions that you should make early on and have guidelines for so that your authors know how to format their expressions.
posted by tractorfeed at 10:16 PM on November 4, 2009

Parts of these notes, from Donald Knuth's Mathematical Writing course, may be helpful.
posted by James Scott-Brown at 9:43 AM on November 5, 2009

(Sadly, that link seems to be broken.)
posted by nebulawindphone at 10:22 AM on November 5, 2009

Perhaps it was this.
posted by crapmatic at 10:34 AM on November 5, 2009 [2 favorites]

Looks good, crapmatic. Thanks.
posted by nebulawindphone at 4:40 PM on November 8, 2009