7
$\begingroup$

This is a question about "Program for computing tree decomposition of a graph".

Should we admit questions that are looking for code for certain algorithms ?

On the YES side, there was a question a while ago about #SAT solvers that I defended. There was also a recent question about transformations for SAT solvers. There's clearly interest in this topic (at least for SAT).

On the NO side, we could open ourselves up to a flood of 'can I find code for...' questions.

Thoughts ?

$\endgroup$
  • 6
    $\begingroup$ IMO your community needs to rethink this policy of being so very inordinately strict about "research level only" questions. But, that's just my humble opinion. $\endgroup$ – Jeff Atwood Apr 13 '11 at 5:39
  • 1
    $\begingroup$ I think this question is OK. As long as a question is interesting/helpful to other researchers (e.g. may come up in a coffee break discussion between them) and is not clearly off-topic (i.e. is not answerable by simple Googling/Wikipedia article/undergrad textbook/a typical undergrad taking an undergrad course on the topic) then it is OK leave it open. IIRC, the objective of restricting the scope to research level question was (and is) keeping the site interesting to researchers in TCS, so I agree with Jeff, we don't need to be too strict on it. $\endgroup$ – Kaveh Apr 13 '11 at 11:35
  • $\begingroup$ there is now a se site for Software Recommendations but it might be too applied for theoretical software areas. really like to see implementations of abstract algorithms/ papers, think they have many benefits, are useful to the community, sometimes rare & should be encouraged (both on the site & elsewhere). $\endgroup$ – vzn Apr 21 '15 at 22:02
21
$\begingroup$

There are some types of questions on implementation that I think are in scope:

"Chazelle's linear-time polygon triangulation algorithm is notoriously complex. Is there any peer-reviewed research that discusses any implementation of it?"

"Fibonacci heaps are much more complex than pairing heaps, but have better theoretical bounds. Are there any comparisons of the constant factors, perhaps with experimental evidence based on real implementations?"

"Are the state-of-the-art QSAT solvers fast enough to be used in flow-sensitive program analysis?"

I think even more might be in scope if the rule is "questions are considered to be 'research-level' roughly when they can be discussed between two professors or between two graduate students working on Ph.D.'s, but not usually between a professor and a typical undergraduate student." What about questions like:

"I need to have an open-source GPU-based SAT solver as a building block for testing the usefulness of my new algorithm. Do any exist?"

$\endgroup$
  • 7
    $\begingroup$ This is a good answer. In short, if the implementations are (highly) non-trivial of research-level algorithms, then they should be in scope. $\endgroup$ – Dave Clarke Apr 13 '11 at 7:51
  • 1
    $\begingroup$ @DaveClarke - Thanks! In response to your "non-trivial" qualifier, Fibonacci and pairing heaps aren't too complicated, so I don't think "I need a fibonacci heap for my research, are any around" is really research-level, even if "how fast are they/ is there an article in the jea.acm.org about them?" is. $\endgroup$ – jbapple Apr 13 '11 at 7:59

You must log in to answer this question.

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