I asked about integer programming before on CV and it was locked as off topic. I'm unable to find a site dedicated to math programming and optimisation on stack exchange. Any one has idea?


Optimization per se isn't part of our site's interests. Like Calculus, Linear Algebra, Computer Science, and other disciplines, it is a tool we use, respect, and enjoy, but is not otherwise of interest in its own right. So, just as we tend to send pure Calculus, pure Linear Algebra, etc. questions (which do not have any explicit connection to statistics or machine learning) over to Mathematics, we will also close pure optimization questions. And just as we sometimes do answer questions about Calculus, etc., when they are clearly addressing a topic in statistics or machine learning, we sometimes answer optimization questions. But that's rare.

There are two places to look for communities that would answer pure optimization questions:

  1. Existing SE sites. Use the menu at the top to filter by category. I don't see any good candidates for general questions about optimization. However, some sites will field suitably framed questions of that nature, such as

    • Math Overflow (for research-level questions in optimization techniques, perhaps),

    • Mathematics (primarily for elementary problems, such as Lagrange multiplier questions, etc.),

    • Mathematica (which is interested in computer solutions using Mathematica--which despite the apparent narrowness can be an effective way to get answers, because Mathematica programs frequently are like executable mathematics), and especially

    • Computational Science.

    A more extensive look (which requires browsing through a hundred Beta sites, unfortunately) also turns up Data Science, which might like to address certain kinds of techniques that are used in data mining.

  2. Proposed SE sites ("Area 51"). I searched for "Optimization" and found nothing. If you're energetic and dedicated you might consider starting a proposal.

  • 2
    $\begingroup$ Computational science seems a good place to go. Thanks to @chl as well. $\endgroup$ – jf328 Jan 15 '15 at 16:27
  • 1
    $\begingroup$ I'd say Mathematics could handle way more sophisticated questions than lagrange multipliers. Any optimization topic that could conceivably be taught at the first year PhD student level is probably more on topic in Mathematics than Math Overflow. $\endgroup$ – WetlabStudent Jan 17 '15 at 3:37
  • 1
    $\begingroup$ @MHH Lagrange multipliers were mentioned only as an example, because I would expect far fewer readers of this meta site to be acquainted with quasiconvexity, MILPs, etc. "Elementary" is in the eye of the beholder, but I'm not qualified to give clearer guidance: as to where Mathematics ends and MathOverflow begins, I will have to leave that to those two communities. My sense is there's some overlap. $\endgroup$ – whuber Jan 17 '15 at 15:21
  • $\begingroup$ See also cs.stackexchange.com/questions/tagged/optimization $\endgroup$ – Firebug Aug 18 '16 at 15:41

To expand on the on-topic side of @whubers answer:

Firstly, there are applications of optimization that are on topic here: both optimization applied to the "usual" fitting of models and hyperparameter optimization.

Secondly, I think one fundamental difference between CV and more "mathematical" sites is that stats is about noisy data. In contrast most optimization heuristics I'm familiar with assume well behaved smooth target functions. So I'd also conster questions about optimization of noisy target functions on topic.

  • 3
    $\begingroup$ +1 Good points. I would add, though, that a question posed as a pure optimization problem--with no reference to its application in stats, machine learning, or anything else considered on-topic here--should usually be considered off topic. This is as much out of compassion for the OPs as it is for our community, because such questions rarely get any answer or comments at all here: they need to find a better home. $\endgroup$ – whuber Jan 27 '15 at 22:46

You must log in to answer this question.

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