I am a mathematician who works primarily in the spectral theory of self-adjoint and unitary operators. Some of my research is of relevance in quantum walks, and I have one paper in particular that I would like to present in the Quantum Information Processing conference this January in Seattle.

The culture of the theoretical CS community seems very different from the math community. I even had to look up what it meant to "present a paper" at a conference, since we never use that language in math conferences!

I was wondering if I could get some advice on presenting my paper for consideration in this conference. Along with an arXiv link to my paper I am supposed to give them a 1-3 page extended abstract, which I suppose is a way to convince them my paper is relevant for the conference. What is normally expected in such a document?

For instance, to what extent should I talk about the mathematical machinery? Would the organizers care about the (very difficult and technical) details of the proof, or should I just state an algorithm, assert that it works and explain why it is relevant to computer science?

  • $\begingroup$ While QIP does not have published proceedings, it is still rather prestigious. The acceptance rate is, as far as I recall, around 20-30%. Thus, if you are unfamilar with QIP, it might be better to focus on the submission itself. $\endgroup$
    – user60974
    Commented Sep 9, 2016 at 17:02
  • 1
    $\begingroup$ Is this question about how to present the paper on the conference, or how to make a submission (extended abstract)? $\endgroup$
    – domotorp
    Commented Sep 9, 2016 at 18:44
  • 5
    $\begingroup$ The QIP 2015 program quantum-lab.org/qip2015/Program.php has some three-page abstracts linked — suggest you look through those as a guide to how to format yours. Advice for a talk itself is premature until you get your abstract accepted — unlike many mathematics conferences, you should not expect that anything you send will be accepted for presentation. $\endgroup$ Commented Sep 9, 2016 at 19:00
  • $\begingroup$ Chi provided a great answer for what talks should be like in a CS conference. As for QIP submissions, QIP is very competitive (with an acceptance rate of roughly 20%) and has a varied audience background (mostly TCS or physics, but also a few others). Do look at 3-page abstracts from previous years to get an idea of what you should include. It's an advertisement for why your paper is interesting and why people should read it or want to listen to a talk about it. $\endgroup$ Commented Sep 9, 2016 at 21:46
  • 1
    $\begingroup$ TCS values motivation and context in papers and talks. It seems math does not emphasize this. A three-page "extended abstract" should focus on why your results are interesting by motivating the question and relating it to previous work and only giving a statement of your contribution. $\endgroup$
    – Thomas
    Commented Sep 10, 2016 at 18:51

1 Answer 1


First and foremost, stay on time.

Usually, at conferences you are supposed to talk for a very limited amount of time, e.g. 25 minutes including 5 for questions. In such slot you will not be able to go deep into the technical details.

You should have a more "marketing" approach to presenting a paper. You have to convince the attendants about:

  1. what your main results are
  2. why your results are important
  3. why later on one should spend some time to actually read your paper

Try to be simple and direct. Most attendees will also have spent part of their mental energy on previous presentations. Chances are they will forget about your paper very soon (there are too many to remember for a human mind!). Your goal is to make them to remember something about your paper.

Leave the proofs out. You can briefly mention the proof technique if relevant, though ("by reduction to problem XYZ"). Provide the definitions and statements -- if they are very technical, you can be a bit vague ("under some regularity conditions..."). Clarity trumps precision here. Provide some background for people not in the field, but do not waste time in reviewing basic definitions. Here it is important to know what the audience is familiar with, which depends on the venue. E.g. in a TCS conference, everyone knows what a non-deterministic Turing machine is.

Write and use a few (computer) slides -- you can't waste time writing on a blackboard, and most venues will expect you to have a computer presentation ready. Bring your laptop to the presentation. For a 25 minutes talk, use around 20 slides. Beforehand, make a pretend presentation in your office, and use a timer to check you can do it on time. Optionally, you can take some notes about what to say on each slide, and use these notes during the talk.

Sum up the main points in the last slide to help people remember your talk.

Expect at least a question at the end. Sometimes a question can be not very relevant since the audience may not have grasped everything. Politely answer that anyway.

The goal of the 3-pages short abstract is similar: to convince people that your results are interesting, and that they should read the full version. Try to summarize the main points, trying to convey the general idea even if the space does not allow a proper discussion. Avoid the technical parts which are not needed to assess your results.

  • 1
    $\begingroup$ That point about the use of slides is interesting! In mathematics, slides are considered a bit of a "necessary evil". We use them a lot because they are convenient, but I think mathematicians prefer blackboard talks because they permit more of a give-and-take with the audience. $\endgroup$
    – Darren Ong
    Commented Sep 10, 2016 at 0:52
  • $\begingroup$ In (T)CS, I also consider them as a "necessary evil", and prefer the pace of the blackboard, as well as its space. A good blackboard can show a lot of details at the same time compared to a slide. Still, when time is short, slides can be effective. $\endgroup$
    – chi
    Commented Sep 10, 2016 at 7:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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