[Matplotlib-users] matplotlib at LIGO/GW observation

Benjamin Root ben.v.root at gmail.com
Tue Feb 16 15:30:32 EST 2016


You do bring up a very good point. With large projects like yours, it does
become difficult to explicitly cite every single piece of software that was
used, and how far does that go? For example, do you cite BLAS/ATLAS if you
were using numpy? A citation for Linux? For gcc?

Probably the best way to deal with that is to treat it like a dependency
problem. When a paper is published, cite the software that was immediately
used for the paper and/or any papers that have already been put out for a
particular portion of the software stack. So, ideally, several LIGO papers
would have already been put out about particular portions of the technical
stack that goes into great detail, and would have cited software more
exhaustively there. Then, this particular paper would only need to cite
those kinds of papers rather than redoing the entire citation stack again
and again. The software packages get their impact factor, and the authors
and reader maintain their sanity.

Yes, this is a bit of a chicken-n-egg problem because many editors frown
upon "software papers". I personally had a paper rejected once partly
because the editor thought I was shirking my authorship duties of
describing the procedures by footnoting a github link and citing
NumPy/SciPy and such. I am hopeful that these attitudes are changing, even
in the past few years since I submitted that paper. Your team's work is in
a fantastic position to broadly change these attitudes due to your
visibility across the sciences. Already, your efforts to make the source
code and data easily accessible has stunned many scientists who are not
used to this level of openness. I work in an atmospheric research shop, and
long-time researchers are just amazed by how easy it was to "run the
simulation" themselves via the jupyter notebooks! Being atmospheric
scientists, they never thought that they could possibly study anything
astrophysics related with just a click of a link.

Rock on!

Ben Root

On Tue, Feb 16, 2016 at 3:02 PM, Duncan Macleod <duncan.macleod at ligo.org>

> Hi Ben, all,
> You are correct, we haven't been very diligent in citing the software used
> in our results, mainly from the problem of having a very large software
> stack; everything from real-time interferometer operations and low-latency
> data analysis through detection characterisation and myriad offline
> analysis pipelines use hundreds of packages (C, C++, python, matlab, ROOT,
> ... on multiple OSs) which becomes hard to cite.
> I will, however, bring the subject up in the collaboration to try and
> collect citations from software so that we can get in the habit of
> providing proper references. Thanks for bringing this to our attention.
> Duncan
> On 16 February 2016 at 13:00, Benjamin Root <ben.v.root at gmail.com> wrote:
>> Speaking of citations, while we have your ear...
>> Some of us have noticed that the paper did not include any citations to
>> the scientific software that were utilized. This is somewhat of a new thing
>> to cite software, but matplotlib, numpy and other projects all have
>> suggested citations that we encourage researchers to use in their papers.
>> Many of us are also researchers, and contributions to projects like numpy
>> and matplotlib are often not treated as being on the same level as any
>> other publication because researchers rarely cite the software projects
>> they use.
>> http://matplotlib.org/citing.html
>> No hard feelings, we love what you guys have done. Just flagging it so
>> that you guys might do so in future papers.
>> Cheers!
>> Ben Root
>> On Tue, Feb 16, 2016 at 1:54 PM, Duncan Macleod <duncan.macleod at ligo.org>
>> wrote:
>>> Hi all,
>>> I'm a member of the LIGO Scientific Collaboration, and an author on the
>>> PRL paper linked by Nils. Figure 1 did indeed use matplotlib, as did all of
>>> the graphs in this paper (not the detector layout).
>>> The paper is 'open-access' with a Creative Commons Attribution 3.0
>>> License attached, so I believe the figures can be freely included in other
>>> documents as long as credits are given to the authors and the journal with
>>> an appropriate citation. It would be very nice to see this included as an
>>> example of using matplotlib for scientific analysis.
>>> Thanks
>>> D
>>> On 14 February 2016 at 07:52, Nils Becker <nilsc.becker at gmail.com>
>>> wrote:
>>>> Hello everyone,
>>>> as the direct observation of gravitational waves made its way round the
>>>> world a few days ago, I was pleased to see that they (very probably) used
>>>> matplotlib for their plots. They even used the new viridis colormap [1].
>>>> I could not confirm this directly for the plots in the paper but at
>>>> least the data analysis stack at LIGO seems to be built partly on python.
>>>> They provide scripts to reproduce the data analysis in python and use
>>>> matplotlib to plot it [2].
>>>> In any case, maybe it's an idea to contact LIGO to confirm this and ask
>>>> them if we could put the figure on the website gallery as a kind of "plot
>>>> of honor" or something? I mean there is a chance that it's going to be the
>>>> most famous plot done in matplotlib to this date.
>>>> Cheers
>>>> Nils
>>>> [1] http://journals.aps.org/prl/abstract/10.1103/PhysRevLett.116.061102
>>>> (page 2)
>>>> [2] https://losc.ligo.org/software/
>>>> _______________________________________________
>>>> Matplotlib-users mailing list
>>>> Matplotlib-users at python.org
>>>> https://mail.python.org/mailman/listinfo/matplotlib-users
>>> --
>>> Duncan Macleod
>>> duncan.macleod at ligo.org
>>> LIGO Data Grid systems development
>>> Louisiana State University
>>> _______________________________________________
>>> Matplotlib-users mailing list
>>> Matplotlib-users at python.org
>>> https://mail.python.org/mailman/listinfo/matplotlib-users
> --
> Duncan Macleod
> duncan.macleod at ligo.org
> LIGO Data Grid systems development
> Louisiana State University
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/matplotlib-users/attachments/20160216/737c98fc/attachment.html>

More information about the Matplotlib-users mailing list