# Citations: overkill or not?

I've been using Sage to do a bunch of calculations for a math paper I'm now finishing. I'm definitely citing Sage, as well as PARI/GP (which I used within Sage, as well as on its own before I started using Sage). But I'm wondering about how much else to cite. Using the "get_systems" command, I've found that different functions I've used make use of GAP, GiNaC, MPFR, MPFI, Maxima, GMP, FLINT, and Mpmath, and I'm not even done looking yet. (Some details below.)

Does etiquette dictate that I cite (with both a mention in the text and a bibliography entry) each of those? On a gut level, that feels like a lot. And I don't know if any of those components are considered so basic that they're not worth mentioning. But I want to give credit where it's appropriate.

Some details on what is used for what: My work in Sage involves numerically calculating polylogarithms (which seems to involve GiNaC, MPFR, and MPFI), Stirling numbers of the first kind (GAP), estimating derivatives with finite difference methods (mpmath), finding complex roots of polynomials (PARI, GiNaC, MPFR), resultants (PARI, GiNaC), and lots of basic arithmetic of 1- and 2-variable polynomials over C (FLINT, Maxima, etc. etc. etc.). The work is more numerical than symbolic. There's a bit with large integers, but a lot more with high-precision (1000 digit, say) real/complex numbers.

Thanks, John

edit retag close merge delete

This might be a good question to ask on MathOverflow as well.

( 2011-10-21 00:38:12 +0200 )edit

Thanks, Benjamin -- I might post it there.

( 2011-10-21 13:24:26 +0200 )edit

Sort by ยป oldest newest most voted

The get_systems command just records what you used. So if you add 1+1 then GMP/MPIR is involved, though thats not really cite-worthy. If you are pushing the envelope in large integer computations, then you should definitely cite GMP/MPIR. Use your own judgement to decide what to cite.

Your question does hit on a difficult question: How to automatically decide what needs to be cited and what doesn't. Its probably fair to say that contributions to scientific software are under-cited. There is a citation package in the works that'll make it easier to cite things, and where individual methods can be marked as "cite-worthy" or not.

more

Thanks, Volker. I suspected that certain components were only contributing basic things. But I'm one of those users that only barely understands what's happening under the hood. I'll try to use my own judgment, but I don't trust it much here. I saw something about the new citation package that Neils Ranosch and others are working on. If it is functional right now, I'm too dense to know how to obtain it or make it work.

( 2011-10-21 12:38:14 +0200 )edit