Ask YC: Contributing to Mathematics Journal?

[ComputerGuru]

While working on a (coding) project I had need of a particular numerical method/algorithm which is pretty popular and used in many places. While working with it, I noticed that with a small change to the fundamental idea of the method would result in a pretty decent improvement in terms of efficiency and accuracy, and rewrote it. A friend recommended that I try to publish the revised algorithm in an academic journal as it would be good for my career, etc.

What's the general opinion about coders contributing to scientific journals in the community? Keeping in mind that I've never even read a mathematics journal (though I have read many individual articles here and there), what would be a good journal to approach and how does one go about submitting an article for peer review with them (process, steps, requirements, etc.)?

[cperciva]

The process varies from journal to journal, but often you can just submit a paper online (see http://ams.org/mcom/mcomsubmit.html for example).

That said, if you're not familiar with the journal your odds of getting a paper accepted on your own are rather slim. A better approach would be to find someone with publication experience in the field and get them to look over your work before you submit it -- if you don't have anyone else, I'd be happy to help with this.

As for which journal to approach -- it depends on the material, of course. A good place to start is with whichever journal is most frequently named in your list of references. :-)

[ComputerGuru → cperciva]

Thanks. I have a pretty good relationship with some university professors who I could ask for a hand - but I've heard scary stories about professors taking all the credit?

[cperciva → ComputerGuru]

I've heard scary stories about professors taking all the credit?

Don't worry about that -- it happens very rarely, but I suspect that even from those few cases the vast majority are dealing with graduate students where there's a real question as to who had which ideas.

[hugh → ComputerGuru]

Only a small minority of jerks would bother to try, so just make sure that the professor you're approaching isn't a jerk. Frankly most professors don't have time to bother trying to steal credit, especially for something which isn't close to their core research.

Ultimately you're just asking them for help in refining it into publishable form, so you should be first author, they should be second author.

Also, note that the first professor you go to is likely to think that what you've done isn't sufficient for a publication. That's OK. But if you try three professors and they all think that, you might want to give up trying. Based on your description (small change, decent improvement), what you've done may or may not warrant a publication all on its own.

[cperciva → hugh]

Ultimately you're just asking them for help in refining it into publishable form, so you should be first author, they should be second author.

Only if you've feeling very generous. I would acknowledge help received in a section at the end of the paper, but I would never list a co-author who didn't contribute any of the ideas.

[brent → cperciva]

If I had to guess I would presume the OP has not done a formal analysis of the algorithm. This would undoubtedly be the contribution of the professor/researcher assuming the analysis proves it is an improvement over prior algorithms. This analysis/proof is certainly worth being a second author even if it is not a contribution to the algorithm itself.

[hugh → cperciva]

Perhaps we're estimating differently the amount of work which will be needed to turn this into an actual paper.

But I've never published in a maths journal (only physics) so you probably know better than I do on this point.

[donal → ComputerGuru]

Having someone who has successfully been published by the journal you are submitting to review your article will greatly help your chances. It will also probably smooth out the process some since each journal has its own little quirks.

[lliiffee]

The possibility of it being accepted depends on the nature of the improvement. If your improvement, for example, takes it from 100n^3 to30 n^3 operations, your odds of being accepted are very low. If you go from 100n^3 to 100n^2 (or even 10000n^2), and that has never been discovered before, it is definitely publishable. However, even so, it is important that you write in the correct style, because (unfortunately) there are certain conventions, and many journals aren't that friendly to "outsiders". So the advice of getting someone who has published to look at it is very good.

Nevertheless, it depends on the nature of the improvement, as I said above. If you have that, you can be sure that it can eventually be published.

[cperciva → lliiffee]

If your improvement, for example, takes it from 100 n^3 to 30 n^3 operations, your odds of being accepted are very low.

I'm not sure I'd go that far. There have been many papers published which make improvements in the constant factors in FFTs, and a few others in areas like integer factorization and primality proving.

What matters is whether the work is exciting to other people in the field -- an improvement which is inconsequential in one field could be groundbreaking in another.

[bdittmer → lliiffee]

100n^3 and 30n^3 are both O(n^3)...no improvement, hence it won't be published. Depending on the problem the algorithm is solving O(n^3) to O(n^2) can be a massive improvement.

[rw → bdittmer]

Constant-time factors matter to everyone working with finite resources.

[vegashacker]

I don't think you have to limit yourself to journals--conference papers are another alternative. At least in Computer Science, a good conference is a totally respectable way to go. (It's been awhile since I've been in grad school, so my info may be stale.) STOC and FOCS come to mind. SODA, as I recall, has some sort of "short paper" conference which might be good.

[brent → vegashacker]

I think this is dead on. The OP should look towards publishing in a conference proceedings. In CS this is much more common than journals and his odds of getting it accepted are probably much higher. You nailed the top conferences in this area. However, these are still quite competitive. Perhaps the OP could be more specific about the area where he made an improvement and experts on the forum can try to suggest the best places to submit it.

[mark-t]

Journals are really picky. The whole process takes ages. I'd recommend just writing it up and posting it to the arXiv.

[wheels → mark-t]

The time argument is the thing that really turned me off. In my one published paper we wrote the draft a full two and a half years before it saw the light of day. One referee didn't like it, so it went through a review process, then got bumped down to a lower journal (PRL to PR-A for the physics folks) started the acceptance process again...

Even in CS where things are mostly conference oriented, you're still probably looking at close to a year between finishing the research and seeing it published.

[yummyfajitas]

What algorithm and what type of improvement did you make? Also, can you prove the correctness/accuracy of your improvements?

The answer to these questions significantly affects the journal you should submit to. I can offer you suggestions if you give a little more detail.

You might want to work with someone at a local university on it. Doing research is hard, and writing papers is also difficult (especially your first paper).

[jfarmer]

Make sure you hook up with someone who has written papers before. Aside from the procedural stuff there's all sorts of stylistic issues when writing math papers (I call it math-ese).

[keefe]

Back in grad school when I was working on papers, it seemed pretty essential to have an experience PhD involved in the process. No matter how "great" the idea I had, there were always a lot of complications. For example, making sure I had the appropriate literature review done, did controlled comparisons against best of breed algorithms etc. I would write an abstract of your improvements with hard numbers and the conclusions that you draw from them and also a 2 page summary explaining all this, and email it to a friendly applied mathematician.

[auferstehung]

http://arxiv.org/help/submit

http://arxiv.org/list/cs.DS/recent