Is it acceptable for a math paper to have no concluding section? The structure is currently as follows:

  1. Introduction (known results + the paper's main results as 3 theorems)
  2. Preliminaries (proper definitions of all the things and recalling of results used in the proofs)
  3. Proofs (actually, several lemmas and proofs of them, and of the theorems)
  4. Open problems

It's quite a short paper. I thought that since the Introduction is actually containing all the important stuff, making a Conclusions section would be only repeating the same stuff again. However, I'm not sure if this is considered a poor style in maths.

  • 22
    This seems to be one of the cultural differences between math and computer science. "Conclusions" sections are rare in math and quite common in computer science, even in very theoretical computer science whose content is very close to math. In some cases, the conclusions section of a computer science paper is little more than the introduction with the tenses of verbs changed: "we will show" in the introduction becomes "we have shown" in the conclusion. Mar 23, 2015 at 1:36
  • 3
    @AndreasBlass This must vary even within theoretical CS. Very few of the algorithms and complexity papers that I read have a "conclusions" section. Mar 23, 2015 at 17:06

3 Answers 3


The majority of papers in pure math have no concluding section. In fact, such papers most commonly end with the last line of the proof of the last main result (or the last lemma needed for the last main result) of the paper.

I think that I have never seen a pure math paper that has a concluding section in the sense of other academic papers, i.e., whose sole purpose is to summarize what has already happened. Some math papers have a final section which is more forward-looking. Your current structure has as a final section "Open Problems". That is as close to a concluding section as I've ever seen.

In fact though even this much is often not really appreciated or desired (by editors and referees anyway): many of my earlier papers contained a substantial "open problems / further work" section. Once I got the complaint that it seemed that I was trying to stake out territory (and there was some truth to that, I suppose). More than once I've been told that including too much speculation can be "embarrassing" when the truth comes to light. (In one case, I recorded computations that suggested a surprising conjecture. The referee said that I could keep that in if I wanted, but strongly implied that if I were more experienced I would know not to do this. I left it in, and a few years later my first student proved a theorem confirming these calculations.) In general if you include too much material in a math paper which is not directly used to prove the theorems of the paper, then referees start to wonder whether there is enough content in the paper to justify taking up all that valuable journal space: they could after all publish another theorem and proof instead. (And let's be fair: "they" have a point.)

If you're relatively young and inexperienced and hoping for best results on the rapid publication of your work in strong journals, I would stick pretty mercilessly to the format: (i) strong introduction motivating your work and explaining clearly the value added both in the results themselves and the techniques of proof and (ii) the rest of the paper contains careful proofs of all the results, in a very clear, linear, easy to follow fashion, e.g. "Section A.B: Proof of Lemma C".

  • Thanks for this confirmation. Yes, the paper is "linear" (with the exception of the theorems being in front) and clearly organized I think (I'll see what the referees are gonna think :-) ) and not too much auxiliary information is provided (3 figures, 2 examples and 2 short open problems).
    – yo'
    Mar 22, 2015 at 23:36
  • 5
    "I think that I have never seen a pure math paper that has a concluding section in the sense of other academic papers." I've seen papers that could be called "pure math papers" with such a concluding section, but most of them were published in applied math journals which mainly print papers by people not in math departments. (E.g., IEEE journals.) Mar 23, 2015 at 17:26

Plenty of papers in math don't have a conclusion, or even a separate "open problems" section at the end - in fact, I think this is the norm.

Some specific examples: http://arxiv.org/pdf/1503.05803v1.pdf, http://arxiv.org/pdf/1503.05884v1.pdf, and http://arxiv.org/pdf/1503.05880v1.pdf - currently the first papers listed on the arxiv in logic, number theory, and algebraic geometry - have no conclusions.

Although, it is also by no means unheard of: http://arxiv.org/pdf/1503.05799v1.pdf and http://arxiv.org/abs/1503.05902 - the first papers in the "commutative algebra" and "algebraic topology" sections - do have (very short) conclusion-y sections.

  • 3
    It's interesting that you found a paper in which the last section is called "conclusion". One comment about that: this is the first (or second; they were submitted on the same day) by the author. And it has not been published yet. In my opinion it would be more standard to include those seven lines of "further work" in the introduction (though of course there's nothing inherently wrong with the format; it's all cultural). I wonder whether the published version will retain this section? Mar 22, 2015 at 23:37
  • 3
    @PeteL.Clark As a general comment (I didn't look at the paper in question) not only would it be more standard to put this in the intro, it is advantageous if you want it to be read. Even people who are interested in the paper will often read no further than the introduction, if they get that far.
    – Kimball
    Mar 23, 2015 at 3:54

In applied math (at least in my subfield), most papers do have a conclusion, and it's almost always a useless repetition of things already stated in the introduction (sometimes word-for-word). Please don't follow that pattern!

I try to use a conclusion only if there are general observations or discussion that I wish to include and that would not make sense in the introduction. This is rarely the case.

You must log in to answer this question.

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