This is admittedly a soft question and, to make things worse, the goal of my question might even be controversial. Please bear with me, though.

I'm writing a long math paper with hundreds of math blocks (definitions, theorems, lemmata, remarks, examples, …) and some – though not much – text in between. Let's call that text prose because its only purpose is to serve as rhetorical glue between the math blocks and to provide motivation and guidance. In contrast, the math blocks stand on their own and a reader could in theory only read those, i.e. follow the usual chain of definition – lemma – theorem – proof, and ignore the prose completely. (In particular, variables etc. introduced in the prose are not used in the math blocks and every math block carefully spells out its assumptions.)

Unfortunately, the default amsthm-like style of math papers written in LaTeX doesn't seem to lend itself well to this sort of logical separation since (at least in my opinion) there is little to no optical separation between math blocks and prose. Everything is often just one contiguous text.

For instance, in most papers I read, I find it extremely hard to tell upon first sight where a remark, an example, or a definition ends and where prose starts. In addition, corollaries, remarks and examples – which usually refer to / follow from the math block immediately preceding them – are not typeset in a way which makes this relationship clear. Finally, it is also quite hard sometimes to see which theorems are important and which less so.

I've been thinking about various ways to fix this, i.e. about how to use typography to highlight the overall logical structure and, among others, achieve clear optical separation between math blocks and prose:

  • larger verticals margins between math blocks and prose
  • separation of subsequent math blocks and/or prose by a horizontal line or by *** or some other typographic ornament
  • highlight (some) math blocks with borders / frames
  • different indentation/geometry for prose and math blocks
  • always ending definitions/remarks/examples with an analogue of the QED tombstone to demarcate their end
  • different fonts for math blocks and prose
  • different \parindent for the first line of prose that immediately follows a math block
  • start prose with an initial

None of these solutions seems ideal, though (what's best seems to depend very much on the situation – e.g. how many math blocks vs. how much prose), and I have yet to see a math paper where things are typeset beautifully. I am therefore looking for inspiration.

Hence my question:

Can you give examples of math papers or books that, in your opinion, are typeset in a beautiful, non-standard way (maybe even incorporate some of the above fixes) and thereby alleviate or even solve the above issues? Alternatively, are there any best-practice guides on how to solve the above issues typographically?

In the spirit of this question and that one I really hope my question is not off-topic. If it is, I would very much appreciate any help to make it on-topic.

  • 4
    What you describe as amsthm style is usually intended to be expository. What you want to produce is, instead, what I would term "instructional". These are two different modes of presentation. I can't offhand point to any examples of the latter, but the package ``tcolorbox` has been recommended as providing a mechanism for highlighting examples. Feb 20, 2020 at 16:51
  • There are journals devoted to "formal proof" style papers, which somewhat resemble your requirements (from what little I've seen, not my area/taste at all!), perhaps one of their styles serves you.
    – vonbrand
    Feb 20, 2020 at 18:26
  • @barbarabeeton Thank you for your comment! I have seen tcolorbox before but I have yet to see it being used outside PowerPoint presentations, i.e. as part of a book or paper.
    – balu
    Feb 20, 2020 at 19:08
  • As for the amsthm style being intended to be expository: That's not my experience at all. I've read dozens of books that were typeset using amsthm.
    – balu
    Feb 20, 2020 at 19:13
  • @vonbrand Do you have any specific papers in mind? I've briefly taken a look at some formal-proof papers and those didn't look like anything I'm having in mind. (In fact, they were worse than the papers I'm used to, typographically speaking. :)) Please don't mistake my question for a question on how present proofs in a very formal and dry fashion. Instead, what I'm basically looking for are papers whose typography made them more, say, accessible, i.e. whose typography made them easier to digest than amsthm.
    – balu
    Feb 20, 2020 at 19:14

2 Answers 2


I would second the recommendation for tcolorbox. The documentation has many examples, of which very few look like they would work in a Powerpoint presentation. The package also offers the \tcolorboxenvironment command, so that you can adjust the preamble of an otherwise "regular" document to add the boxes you want throughout the document. For example:





Some text before.

This is my theorem.

Some text between.

This is my proof

Some text after.


Having told tcolorbox to box up theorem and proof (the second argument allows you to adjust the default options), you end up with:

code output

Adjusting the default options allows you to color the boxes, or adjust the vertical spacing before, during, or after, or many other things.


Please think hard on what you are trying to do. "Officially" mathematics is all about definitions and proofs. In reality, all that is completely useless if it doesn't help understanding what is meant, how things hang together, why some definition is just so, how to use theorems, what the important problems are. And that requires a lot of text in between.

Think about it: An association of professional mathematicians offers their members (but more generally the whole world) a set of carefully designed tools for one of their principal tasks (writing up their results), and that tool works in a totally different way than what you want to do exactly that. Maybe they are all wrong, but I somehow doubt it...

  • 2
    No offense, but what you're saying basically amounts to: "This is the way it's always been done, therefore it's the right way to do it." Besides, I have been long enough in mathematics to know that lots of mathematicians are not happy with the default amsthm style of books and papers.
    – balu
    Feb 21, 2020 at 11:28
  • @balu true. But before questioning the way professionals have settled down to do things, you should think hard about why it is doing that way.
    – vonbrand
    Feb 21, 2020 at 11:32
  • 1
    Well, I think one reason for why people have settled down to do things the way they do, is for a lack of better options and the fact that each journal enforces a certain style and these styles seem to have converged over the years (in part due to the proliferation of LaTeX and the ams packages). People want to focus on math, not on how to best typeset math. (I mean, just consider for a second how few people use \text{} for textual subscripts, or \operatorname{} etc. It's simply not something they want to waste a lot of time on.)
    – balu
    Feb 21, 2020 at 12:23
  • 1
    But OP could still have a lot of text in between and still want to have the boxes more visibly noticed. Once I'm familiar with a subject, the prose is far less important than the boxes. It could be nice to be able to find the relevant box quickly.
    – Teepeemm
    Nov 17, 2020 at 1:15
  • 1
    Having been brought back here by a newly-posted answer, I can add a reason for the styles adopted and proliferated by the AMS: they are intended to be printed on paper in books and journals. Production of these publications is expensive; costs include physical materials as well as the time of copyeditors and other persons involved. In order to keep costs (and thus subscription prices) to a minimum, nonprofit publishers (like the AMS) try to pack as much as possible onto a page, while retaining intelligibility and comprehension. Adding a "flag" at the end of a block is a reasonable idea. Nov 17, 2020 at 1:51

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .