Impact on Scientific Research

[jarrodmillman]

See the parent issue #9 for background.

For the preliminary proposal, we wrote:

Scientific Research. The Scientific Python OSE underpins cutting-edge research across a vast spectrum
of scientific disciplines, from systems biology to materials science, and from astrophysics to
climate modeling [25]. This ecosystem supports endeavors ranging from large international
collaborations like CERN’s particle physics experiments to critical national resources such
as the Rubin Observatory [1, 10, 9], as well as countless individual research projects in
universities, national laboratories, and research centers worldwide [21].

[1] Francesco Andreuzzi. Improving throughput of CERN’s multi-language distributed
online data processing platform. Tech. rep. 2021.
[9] Fr´ed´eric Dux. “lightcurver: A Python Pipeline for Precise Photometry of Multiple-
Epoch Wide-Field Images”. In: arXiv preprint arXiv:2410.13625 (2024).
[10] V Daniel Elvira, Steven Gottlieb, Oliver Gutsche, and Benjamin Nachman. The
Future of High Energy Physics Software and Computing (V2. 1). Tech. rep. Fermi
National Accelerator Laboratory (FNAL), Batavia, IL (United States), 2022.
[21] Alok Kamatar, Mansi Sakarvadia, Valerie Hayot-Sasson, Kyle Chard, and Ian Foster.
“Lazy python dependency management in large-scale systems”. In: 2023 IEEE 19th
International Conference on e-Science (e-Science). IEEE. 2023, pp. 1–10.
[25] Johnny Wei-Bing Lin, Hannah Aizenman, Erin Manette Cartas Espinel, Kim Gun-
nerson, and Joanne Liu. An introduction to Python programming for scientists and
engineers. Cambridge University Press, 2022.

In the full proposal, we have space to significantly expand and improve this. Please let me know if you have better references for the above paragraph. It would also be helpful to have more concrete examples (preferably with references to github repos or published papers). While examples funded by other agencies are fine, it would be great to have more examples that are funded by NSF.

[matthewfeickert]

It would also be helpful to have more concrete examples (preferably with references to github repos or published papers).

This is a SciPy 2024 proceeding, which isn't the same as a journal publication, but us HEP people published How the Scientific Python ecosystem helps answer fundamental questions of the Universe, DOI: 10.25080/KMXN4784*. As myself and some other IRIS-HEP colleagues are co-authors on the proceedings, there's the IRIS-HEP NSF funding statement in there.

IRIS-HEP also organizes (now) annual analysis software training events with the US CMS and US ATLAS collaborations branches (2024 CMS, 2024 ATLAS events) where the focus is on getting early career researchers to become more comfortable with the Scientific Python ecosystem and the "PyHEP" ecosystem that we can build because of the robust and efficient tools. While I don't have any slides from those events that include it, in 2022 Jim Pivarski made a figure of the how the broader PyHEP ecosystem builds up from the Scientific Python ecosystem in a similar manner to the image that Jake VanderPlas gave in his 2017 PyCon keynote showing how the Scientific Python ecosystem builds from CPython up through more domain specific layers/shells.

While those figures have gotten used in multiple talks with the NSF, that's not something that is easy to point to. So I'll continue to dig through the IRIS-HEP publications to find some more concrete published examples.

Publications

• How the Scientific Python ecosystem helps answer fundamental questions of the Universe, Matthew Feickert, Nikolai Hartmann, Lukas Heinrich, Alexander Held, Vangelis Kourlitis, Nils Krumnack, Giordon Stark, Matthias Vigl, Gordon Watts, SciPy 2024 (10 Jul 2024). [NSF]
• IRIS-HEP Strategic Plan for the Next Phase of Software Upgrades for HL-LHC Physics, B. Bockelman et. al., arXiv 2302.01317 (02 Feb 2023). [NSF]

---

* At the time of posting the DOI is broken, but this has been reported to the SciPy proceedings committee.
[NSF] NSF funded

[matthewfeickert]

It would also be helpful to have more concrete examples (preferably with references to github repos or published papers)

I don't have exact references to particular software repositories from published papers, but the software related to the alignment of the LHCb experiment detector (which affects every publication from the LHCb collaboration) is public and uses the Scientific Python stack.