Lately there have been some questions posed that seem to be homework:

Linear programming feasibility problem with strict positivity constraints

Approximately "solving" a linear system of equations without a feasible solution

I suspect that this one is also homework, but it seems the questioner (same as those above) has not bothered to transcribe the question in a sensible way:

Do we want to answer homework questions?

  • 2
    The user in question did not bother to respond to your last comment querying if they were asking a homework question. I deleted her most recent question because the quality is low and I don't feel that low quality homework questions are appropriate for the site. I guess this is the place for our users to disagree with the action. – Aron Ahmadia Jan 25 '12 at 21:36

I would say generally yes, for a couple of reasons:

  1. Scientific computation is a relatively advanced subject. You need to have a fair amount of expertise in both programming and your chosen field of science before you can even ask a question that falls under the domain of computational science. So the homework questions I'd expect to see on this site would tend to be more advanced than those on other science sites, and the people asking them correspondingly more educated. We shouldn't need to coddle them.
  2. The kinds of homework assignments one gets in a scientific computation class are usually pretty significant programming projects. It's not the sort of thing a student could ask a plz-send-teh-codez question about. At least not without it being painfully obvious - sure, someone could say "I need to simulate this N-body system and I don't know what to do," but that's just a crappy question, homework or not.

My point would be that I don't think we need to treat homework questions much differently from ordinary, non-homework questions. Obviously, if they're bad questions, we don't answer them; if they're off topic, we don't answer them; if they show zero effort on the part of the questioner, we don't answer them. But as long as a question is acceptable in all other respects, I don't think the fact that it arises in a homework assignment is going to make much difference.

  • 3
    I agree with David. Homework-esque questions are on topic as long as they fall within the scope of computational science, provided that the asker shows sufficient effort in understanding the problem and/or an attempted answer. – Paul Jan 26 '12 at 18:18

I agree with David Zaslavsky, but with a few caveats.

In the case of sara's questions, they're both pretty textbook questions in a linear programming class, and to answer those questions, you're generally only required to state something like, "I would solve the following LP," and give a formulation. So Dirk's answer for the positivity constraints question would essentially be the entire answer for the homework. (I am still kicking myself that I didn't think of what he did; again, well done, Dirk!) Similarly, the approximate solution question is pretty straightforward; I even have a deleted answer that solves the problem, but I'm not going to post it for a little while.

Some scientific computing problems don't require code; they only require a short description of how you would go about solving the problem, and for those types of homework assignments, I'm reticent about just posting a solution. I like the policies of Math.SE and Physics.SE; perhaps these FAQs could also be used to inspire guidelines.

  • 2
    +1 for doing something similar to Math.SE and Physics.SE. – David Ketcheson Jan 26 '12 at 6:38

Homework problems should not be solved. If it is suspected that a question relates to a homework problem, I think it should be encouraged to identify it as such (i.e., add a "homework" tag and prompt the poster to put up the entire problem and how far they've gotten instead of penalizing them) and then to provide hints rather than solutions. There is no shame in getting stuck and soliciting help, but we don't want "freeloaders" swamping the site.

  • I think there are legitimate questions that relate to homework problems, provided that we don't give them a copy-paste solution (much like the Math.SE and Physics.SE policies). I also agree with DavidZaslavsky that the programming tasks are non-trivial, so it's unlikely that a homework problem involving programming can be solved just from the answer provided on SciComp. However, questions involving derivations and pencil-and-paper math should not be as explicit or complete in their help. – Geoff Oxberry Feb 19 '12 at 2:23
  • Due to recent events, I'd like to clarify my point. Homework and research have different goals. Research requires the solution to a problem. Homework is supposed to install/reinforce skills/lessons and understanding. This requires a different type of answer. Due to the diversity of research in SC and hence expertise, the problem itself will not always be easily discerned as homework or research by all. – Deathbreath Feb 24 '12 at 18:37

I think the answer is yes - there's been several well-reason arguments as to why, but I'll add another:


It's entirely possible that someone (coughEpiGradcough) asking what feels like a basic question, or something with a textbook setup, is asking for help not because they're hoping for an easy mode assignment, but because they're trying to learn something on their own, and have gotten well and truly tangled up. Or are hip-deep in some particularly opaque problem that isn't being passed through the filter of someone who knows what they're doing.

Discouraging those kinds of questions is, in my entirely self-interested mind, bad.

  • As a highly interdisciplinary field has many even advanced researchers whose understanding is basic in some aspects of their research. So I think the self-learner aspect is applicable to much of our community. Hence, the complexity of a question is not relevant IMHO. – Deathbreath Feb 24 '12 at 18:32

You must log in to answer this question.

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