The Astronomy Department at the University of Maryland (College Park) offers a one-credit astronomy scientific computing class, ASTR 288P: Introduction to Astronomical Programming, to provide undergraduates with a foundation in computing. This course is a prerequisite to an advanced-level three-credit course on Computational Astrophysics (ASTR 415).
In ASTR 288P, students learn to work with the UNIX terminal, get the basics of coding with Python and some C, and learn what makefiles are and how to install software, among other topics. The course also introduces students to the ASCL, as for the final class project, students (either alone or in pairs) pick a code from the ASCL, give a short presentation on how they installed and used it, and discuss how that code fits in the large scheme of computing in astrophysics. This allows the students to get a feel for the computational work the astro community is doing and is a good match to test the skills they should have learned in the class.
Until this week, a search in ADS for doctype:”software” keyword:”NASA” returned zero results. NASA has funded the ASCL to make its astronomy research software discoverable in ASCL and ADS. This required changes to the ASCL structure and the ADS data feed, and edits to some current records; it also entails mining various NASA software sites for codes that meet the ASCL’s criteria and creating appropriately tagged entries for them. In the first phase of the project, started in July, our wonderful developer Judy Schmidt (@SpaceGeck) worked her magic on our infrastructure, keywords have been added to some existing records, and ADS has ingested the first entries we’ve tagged with the NASA keyword. We can now see first results from this two-year project:
Additional changes will be coming to the ASCL in the coming months as we continue this funded work. We love this project; at its core, it’s a simple concept, and leverages existing resources (ADS, various NASA code sites, and ASCL) to make research software more discoverable and provides information about NASA software that was not readily available before. It furthers the excellent work NASA has been doing to release software, demonstrates yet another value of ADS (which has many superpowers!), and makes the ASCL more useful, too.
I gave a seminar at the Centre de Recherche Astrophysique de Lyon on Friday, September 14, at the invitation of Mohammad Akhlaghi, a post-doc there. Mohammad is very interested and has done a lot of work on reproducibility, ensuring that his work is reproducible and developing a reproducibility framework that can be adopted by others. The seminar took place on CRAL’s lovely historic campus at the Observatoire de Lyon in Saint-Genis-Laval. The title, abstract, and link to the slides are below.
Title: Make your code famous! (or at least discoverable).
Abstract: Source codes are increasingly important for the advancement of science in general and astrophysics in particular. Journal articles detail the general logic behind new results and ideas, but often the source codes that enable these results remain hidden from public view. In this presentation, I will discuss our recent study on the availability of source codes used for published research and how this affects the transparency and reproducibility of astro research. I will cover what the Astrophysics Source Code Library (ASCL, ascl.net) is, how to submit software to the resource, and the benefits of doing so. I will share what happens after software is submitted, how ASCL entries are indexed by ADS, the links between literature and software entries, and how an ASCL ID can be used for citing your code. I will cover good and bad ways to cite software, avenues for publishing software, and how journals are changing to include and recognize the contribution software makes to our discipline.
Abstract: Astrophysics research relies on software and all robust science requires transparency and reproducibility, yet the computational methods used in our discipline are often not shared or are difficult to find. In recent preliminary research, 40% of the software used in the 2015 papers we examined did not offer source code and restricting the reproducibility of this research. The Astrophysics Source Code Library (ASCL. ascl.net) registers astrophysics research source codes that have been used in refereed research, benefiting the field in numerous ways, including increasing the discoverability of software and making the published research record more robust. With over 1,700 codes, the ASCL is the largest indexed resource for astronomy research codes in existence. This free online registry was established in 1999 and is indexed by ADS and Web of Science. ASCL registration allows your software to be cited on its own merits and provides a citation method that is trackable and accepted by all astronomy journals and journals such as Science and Nature. This presentation covers the benefits of registering astronomy research software with the ASCL, upcoming changes that will enable greater software discovery initially for NASA software and potentially for software funded by other organizations, changes to the ASCL and ADS that benefit researchers, and our research into software use in astronomy.
Alice Allen, Astrophysics Source Code Library/University of Maryland
Robert J. Nemiroff, Michigan Technological University
Peter J. Teuben, University of Maryland
Most ASCL code entries have one or more links to articles that either describe or use the software in that entry. ADS ingests this information to associate the code with relevant literature. For example, the entry for 2LPTIC: 2nd-order Lagrangian Perturbation Theory Initial Conditions includes a link for an MNRAS paper in the “Appears in: field:
Going to the ADS entry for this software shows that the code is associated with a paper under Associated Articles:This, however, doesn’t tell you anything about the relationship between the article and the software. ADS and the ASCL have been working to improve this. The ASCL has been disambiguating these article links into Described in and Used in. At the ADS Hack Day event last month, ASCL provided ADS with disambiguated links for over 900 entries, and Carolyn Grant had these uploaded into ADS in a matter of minutes. (ADS folks are wizards, I tell you! Wizards! They work magic!!)
Currently, ASCL records appear the same, but for those records we have provided disambiguated article links, ADS displayed them as Described in and Used in, as you can see in the 2-DUST: Dust radiative transfer code entry, for which the ASCL lists two papers:
For ASCL records in which the Appears in link(s) have not been disambiguated, there is no change in how they are displayed in ADS. We have over 700 entries with article links still to be disambiguated and we continue this work; ADS will be ingesting the changes with their regular weekly ingest of ASCL data.