2

One of the nice things about genetic algorithms is that they can easily be used for a diverse array of problem domains, whereas PSO for example seems best-suited for candidates of real-valued vectors (although I am aware of the use of the latter in combinatorial problems). It has, however, occurred to me that, by using genetic algorithms because they are familiar and easy to apply, I could be forgoing a better solution with a different metaheuristic. To avoid subjective discussion, I am of course looking for carefully gathered empirical results about this subject matter. I can't seem to find too many on my own.

Raphael
  • 72,336
  • 29
  • 179
  • 389
readyready15728
  • 121
  • 1
  • Your title is very broad, yet your actual post seems to be more focused. Can you try to update your title accordingly? – Yuval Filmus Sep 03 '17 at 15:58
  • I don't know what edit would be appropriate to achieve that outcome – readyready15728 Sep 03 '17 at 15:59
  • 1
    OK, here's what would achieve that outcome: 1. Edit the question to ask a question in the body of the question. 2. Make sure it is a specific question. If you just copy-paste the title into the body, we're probably going to tell you that your question is too broad. So, you'll probably need to ask a more specific question. 3. Then, choose a title that is a summary rather than an attempt at a complete statement of the question; see https://cs.meta.stackexchange.com/a/815/755. – D.W. Sep 03 '17 at 16:58
  • 1
  • Open-ended questions that call for a list of papers to read often aren't a good fit here. Can you articulate a question that has a single correct answer, and can be answered in a paragraph or two? 5. What research have you done? Have you searched on this site? I've noticed a few related questions: https://cs.stackexchange.com/q/56342/755, https://cs.stackexchange.com/q/561/755, https://cs.stackexchange.com/q/48245/755, and probably more.
    • – D.W. Sep 03 '17 at 16:59
  • I agree. A reference request like yours is too broad for Stack Exchange -- you ask for a survey of a whole research area! You need to narrow your focus considerably before a question of reasonable scope appears. Try talking to your advisor(s), search with Google Scholar and check out this guide to better (re)searches on [academia.SE]. – Raphael Sep 03 '17 at 17:16
  • D.W.: those results aren't what I'm looking for. They talk about (inter alia) when one ought to use something like genetic algorithms rather than some special-purpose technique. I already know that. None of these answers really fit my question. – readyready15728 Sep 04 '17 at 13:04
  • Raphael: I'm not asking for a survey of a whole research area. I'm asking for a survey of which metaheuristics specifically are suited for which problem domains. Google Scholar finds a little to that effect (e.g. https://pdfs.semanticscholar.org/e648/6f8870153660c0aeeea979db8357742ce6ef.pdf) but overall what I can find is pretty scant. – readyready15728 Sep 04 '17 at 13:08
  • Indeed one of the results that seems like it might be most directly relevant only talks about how, in the future, it may be possible to get better data on comparison:

    https://s3.amazonaws.com/academia.edu.documents/4623844/handbook_of_metaheuristics_2nd_edition.pdf?AWSAccessKeyId=AKIAIWOWYYGZ2Y53UL3A&Expires=1504534709&Signature=VMyG%2FHS4mfdd4SQOq73FMESQb2I%3D&response-content-disposition=inline%3B%20filename%3DArtificial_Immune_Systems.pdf#page=646

    I find it a bit funny because loss of source code is mentioned as a problem which would not be an issue if the researchers used GitHub.

     – readyready15728 Sep 04 '17 at 13:24
  • Nonetheless, the suggestions in that chapter might be useful to me personally. We'll see. – readyready15728 Sep 04 '17 at 13:26
  • An aside that may seem cheeky but one that I should probably add nonetheless, because it has to be said, is that giving easy, generic answers or asking for unnecessary clarifications just for the sake of having something to say is not helpful and something I see on Stack Exchange far too often. – readyready15728 Sep 04 '17 at 20:04