5

Suppose that paper A has a lemma called Lemma A, along with its proof.

I want to use this lemma in another paper B. However, in the scope of my paper, I have to change it slightly, but without losing the general idea.

For instance, the original lemma might read as follows:

Lemma A: The intersection of two straight lines in the plane is either empty, a single point, or a straight line.

Suppose I need the following variant in paper B:

Lemma B: The intersection of two planes in 3-space is either empty, a single line, or a plane.

Also, the proofs for both lemmas are very similar. Therefore, I have two questions:

  1. Can I use the same methodology and same terminology with similar words to prove my own lemmas?
  2. If I can, is citing paper A in my lemma confusing? If I cannot, can I just specify the lemma in paper A and say that this lemma and proof can also be used etc.?
10

You should certainly cite paper A in any case.

One way this is commonly handled: state your Lemma B.1 and give the complete proof. At the beginning of the proof, write something like "This closely follows the proof of Lemma A.1 from [A]." Now your paper is self-contained and you have given appropriate credit. It is fine if your proof is similar in structure to theirs; in some ways this is better, because a reader who looks at both will more easily be able to see the similarities and differences. But do not simply copy and paste their proof and change the necessary words. Your proof should be your words, even if it is from their ideas.

Or, state your Lemma B.1, but instead of giving a complete proof, say "The proof is very similar to that of Lemma A.1 from [A]". This saves space but will be more annoying to the reader, who in order to check your result will have to find the paper [A] and read through the proof, adapting it to prove B.1 instead of A.1. (The referee may be similarly annoyed.)

Some people would omit the statement of Lemma B.1 altogether, and when they need to use it, would say "By a slight modification of the proof of Lemma A.1 from [A], we have blah blah blah...". This is even more annoying.

Worst of all is to just say "By Lemma A.1 from [A], we have blah blah blah" where Lemma A.1 claims something different from (and not obviously implying) the statement you want.

| improve this answer | |
  • Would you find it annoying to not restate the proof if the proof is not interesting? This happens to me often when working with programming languages, where there are often proofs that are necessary to do but mostly just handling lots of slightly different cases... – jakebeal Jan 18 '15 at 19:54
  • 2
    This is a good answer. I would say that whether to give the proof is always a judgment call and not always a clear decision. Important factors include (i) the length of the argument you'd be revisiting, (ii) the amount of change you're making -- the given (toy) example is a very small change -- and (iii) the audience of the paper. As a reader, I certainly like it when authors err a bit on the side of being self-contained: it is a lot easier to skip the text that is there than insert the text that isn't there. [Added in edit: my use of "judgment call" was independent from Nate's!] – Pete L. Clark Jan 18 '15 at 20:01
  • 2
    @jakebeal: In my world, there are (essentially) no conference papers. But, aha: the OP talked about lemmas and proofs and without conscious thought I assumed he was in mathematics. Maybe not [OK, I checked: not!], and it would be good to have an answer which speaks to the non-mathematical community which includes theorems and proofs in their papers. – Pete L. Clark Jan 18 '15 at 21:10
  • 1
    In other news: you may be amused to learn that I'm now writing up a paper for the one of the few math journals which imposes hard page limits: Comptes Rendus. The whole paper has to fit on six pages! That's tough, and -- you'll love this -- my collaborator and I actually tried to use the journal's style file to see whether their formatting would still fit. This turned out to be a waste of our time: what we found online was not close enough to give us peace of mind. But holy cow, I've spent a lot of time over the past few days trying to save a few lines here and there. What a pain! – Pete L. Clark Jan 18 '15 at 21:17
  • 2
    @PeteL.Clark Welcome to my world of paper-length optimization. In CS, though, the LaTeX template support is usually excellent and its quite easy to follow formatting. – jakebeal Jan 18 '15 at 21:57
4

In your example, those are not equivalent mathematical assertions, and so they are not identical lemmas. What I have done in similar cases is to say something along the lines of:

Lemma X is closely based on Lemma Y in [cite], and follows a similar proof structure.

This way you give appropriate credit to the original source, while still making your new assertion as you need.

| improve this answer | |
3

Although many people do it, it is bad style and confusing to cite a lemma and restate it in a way that is not equivalent to the original one. I would suggest to state the lemma you need, and, instead of proving the whole lemma, explain in the proof that your lemma is very similar to the lemma A.1 and that the proof can be reused making the changes ... .

If, though, you just steal an idea, it might be better to completely prove your lemma in your version.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.