On Thu, Oct 24, 2013 at 9:11 AM, Jed Brown <[email protected]> wrote:
> This is responding to Matt's comment, which is a topic that we could use > more feedback on. > > > https://bitbucket.org/petsc/petsc/commits/0a8e8854d85628d5034c7879b76533cb3504de85#general-comments > > Should the formal citations list (produced when running with -citations) > include underlying mathematical work or only tangible software that is > being executed? I'm hesitant to include the mathematical background, at > least by default, because it is inherently non-scalable. > I think we can differentiate between mathematical background and analysis, e.g. Proving that GMRES converges with these matrices, etc. and showing exactly how to structure an algorithm: Saad and Schultz, 96 I am for including publications that show how to structure things even if they do not produce code. I think in the future these kinds of publications will also feature code since the culture is changing. > We could perhaps have multiple levels of verbosity, one of which is > basically "further reading". > I think this could also be useful. > Even with only software, applying automated citations recursively is a > scalability risk similar to the advertising clause in 4-clause BSD, > though the standard for publication about software tends to > self-regulate a bit, along with the relative dearth of people interested > in maintaining scientific software packages. > If this does come to pass, we can add a "superceded" tag or something. Matt > What do you think PetscCitations should be? > -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener
